Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
The Receiver Analysis and Design from a System Point of View
José Miguel Caeiro Nogueira
Thesis to obtain the Master of Science Degree in
Electronics Engineering
Examination Committee
Chairperson: Prof. Doutor João José Lopes da Costa Freire
Supervisor: Prof. Doutor João Manuel Torres Caldinhas Simões Vaz Members of the Committee: Prof. Doutor Moisés Simões Piedade
Prof. Doutor Jorge Manuel dos Santos Ribeiro Fernandes
June 2013
Acknowledgements I would like to thank my family for all the support throughout my life, without them I wouldn’t be
where I am now. I would also like to thank my supervisor, Prof. João Vaz, for the excellent person that
he is and for all the guidance and support provided. And last but not least, I would like to thank my
friends and especially my college colleagues, because no one reaches this stage in life all by himself.
i
ii
Abstract One of the biggest challenges in wireless communication systems consists in achieving
flexible dual-band receivers with maximum hardware share and minimum power consumption. The
key for surpassing this challenge is the feasibility and performance evaluation for the different receiver
topologies at system level stage. This task has its degree of complexity, by the fact that the receiver
blocks can have different implementations leading to different performances.
Due to this fact, Agilent Genesys software will be used to perform the system level design
compliant with the 802.11g and 802.16e standards. The proposed architecture will be then exported to
Ptolemy software, where it will be validated with more realistic signal sources, insuring a clean
transition to the circuit level design.
Keywords – WLAN, Mobile WiMAX, System Level Architecture, Receiver, 802.11g, 802.16e.
iii
Resumo Um dos maiores desafios em sistemas de comunicação sem fios é obter receptores versáteis
de banda dupla, com a maior partilha de hardware e o mínimo de consumo. A solução para
ultrapassar este desafio começa pela avaliação da viabilidade e desempenho de diferentes topologias
para o receptor a nível sistémico. Esta tarefa tem o seu grau de complexidade, pelo facto de que os
blocos do receptor podem ter diferentes implementações que levam a diferentes desempenhos.
Devido a este facto, o programa Genesys da Agilent vai ser o software usado para a análise
sistémica de acordo com as normas 802.11g e 802.16e. A arquitetura proposta será depois exportada
para o programa Ptolemy, também da Agilent, onde será validada com fontes de sinal mais realistas,
assegurando uma transição sem problemas para o nível de circuito.
Palavras-Chave – WLAN, WiMAX Móvel, Arquitectura a Nível Sistémico, Receptor, 802.11g,
802.16e.
iv
Contents 1 Introduction .......................................................................................................................... 1
1.1 Motivation .................................................................................................................................. 1
1.1.1 System Level Design ......................................................................................................... 1
1.1.2 Dual Band Advantages ...................................................................................................... 1
1.2 State of the Art .......................................................................................................................... 2
1.2.1 Standards Evolution ........................................................................................................... 2
1.2.2 Mobile WiMAX vs LTE ....................................................................................................... 4
1.2.3 Wimax Growth (in the past 5 years) .................................................................................. 5
1.2.4 Plausible Architecture ........................................................................................................ 5
1.3 Aim of this Thesis ...................................................................................................................... 6
1.4 Thesis Organization .................................................................................................................. 6
2 System Level Concepts ...................................................................................................... 9
2.1 Receiver Design Basics ............................................................................................................ 9
2.1.1 Noise Factor ....................................................................................................................... 9
2.1.2 Sensitivity ......................................................................................................................... 10
2.1.3 Selectivity ......................................................................................................................... 11
2.1.4 BER .................................................................................................................................. 11
2.1.5 Non-linear Behavior ......................................................................................................... 11
2.1.6 Dynamic Range ............................................................................................................... 12
2.1.7 Blocking and Desensitization ........................................................................................... 13
2.2 Receiver Architecture Overview .............................................................................................. 14
2.2.1 Heterodyne Architecture .................................................................................................. 14
2.2.2 Image-Reject Architectures ............................................................................................. 20
2.2.3 Low-IF Architecture .......................................................................................................... 22
2.2.4 Homodyne Architecture ................................................................................................... 23
2.2.5 Comparison of Receiver Architectures ............................................................................ 28
2.3 Transmitter Design Basics ...................................................................................................... 28
2.3.1 EVM ................................................................................................................................. 28
2.3.2 Spectrum Mask ................................................................................................................ 30
2.4 Modulation Basics ................................................................................................................... 31
v
2.4.1 OFDM .............................................................................................................................. 31
2.4.2 OFDMA ............................................................................................................................ 33
3 Standards Analysis and Specifications .......................................................................... 35
3.1 Standards Analysis ................................................................................................................. 35
3.1.1 Maximum Input Signal ..................................................................................................... 35
3.1.2 Sensitivity ......................................................................................................................... 35
3.1.3 Noise Figure ..................................................................................................................... 36
3.1.4 Adjacent Channel Rejection ............................................................................................ 36
3.1.5 SNR and BER .................................................................................................................. 36
3.1.6 Spectrum Mask ................................................................................................................ 36
3.1.7 EVM ................................................................................................................................. 37
3.2 Performance Calculations ....................................................................................................... 38
3.2.1 Noise Figure ..................................................................................................................... 38
3.2.2 Propagation Loss ............................................................................................................. 38
3.2.3 ADC ................................................................................................................................. 39
3.3 Transmitter .............................................................................................................................. 40
3.4 Receiver .................................................................................................................................. 44
4 System Design ................................................................................................................... 49
4.1 Zero-IF Architecture ................................................................................................................ 49
4.1.1 Low Gain Mode ................................................................................................................ 50
4.1.2 High Gain Mode ............................................................................................................... 57
4.1.3 Adjacent Channel Rejection ............................................................................................ 62
4.1.4 EVM ................................................................................................................................. 64
4.1.5 Imbalance and Linearity ................................................................................................... 67
4.1.6 Results Discussion .......................................................................................................... 68
4.2 Low-IF Architecture ................................................................................................................. 69
4.2.1 Low Gain Mode ................................................................................................................ 70
4.2.2 High Gain Mode ............................................................................................................... 76
4.2.3 Adjacent Channel Rejection ............................................................................................ 81
4.2.4 EVM ................................................................................................................................. 84
4.2.5 Imbalance and Linearity ................................................................................................... 86
vi
4.2.6 Results Discussion .......................................................................................................... 87
4.3 Dual-Band Receiver Project .................................................................................................... 89
5 Conclusions ....................................................................................................................... 91
6 Appendix ............................................................................................................................ 93
7 References ......................................................................................................................... 97
vii
List of Tables Table 1 – IEEE 802.16 Standards. .......................................................................................................... 2 Table 2 – Mobile WiMAX Product Certification. ...................................................................................... 3 Table 3 – IEEE 802.11 Standards. .......................................................................................................... 4 Table 4 – Comparison of Receiver Architectures. ................................................................................. 28 Table 5 – Transmitter parameters. ........................................................................................................ 42 Table 6 – Front-end specifications. ....................................................................................................... 44 Table 7 – IF and BB specifications. ....................................................................................................... 44 Table 8 – Receiver global specifications. .............................................................................................. 45 Table 9 – LNA and PGA specifications. ................................................................................................ 45 Table 10 – Mixer specifications. ............................................................................................................ 45 Table 11 – LO specifications. ................................................................................................................ 45 Table 12 – IF and BB filters specifications. ........................................................................................... 46 Table 13 – ADC specifications. ............................................................................................................. 46 Table 14 – LNA, Mixer and PGA requirements. .................................................................................... 46 Table 15 – LNA, Mixer and PGA simulated requirements. ................................................................... 46 Table 16 – Low gain mode requirements. ............................................................................................. 50 Table 17 – 802.11g zero-IF low gain mode BER. ................................................................................. 52 Table 18 – 802.16e zero-IF low gain mode BER. ................................................................................. 55 Table 19 – High gain mode requirements. ............................................................................................ 57 Table 20 – 802.11g zero-IF high gain mode EVM. ............................................................................... 65 Table 21 – 802.11g zero-IF low gain mode EVM. ................................................................................. 65 Table 22 – 802.16e zero-IF high gain mode EVM. ............................................................................... 66 Table 23 – 802.16e zero-IF low gain mode EVM. ................................................................................. 66 Table 24 – 802.11g zero-IF imbalances. ............................................................................................... 67 Table 25 – 802.11g zero-IF linearity. ..................................................................................................... 67 Table 26 – 802.16e zero-IF imbalances. ............................................................................................... 67 Table 27 – 802.16e zero-IF linearity. ..................................................................................................... 67 Table 28 – 802.11g Zero-IF results. ...................................................................................................... 68 Table 29 – 802.16e Zero-IF results. ...................................................................................................... 69 Table 30 – 802.11g low-IF low gain mode BER. ................................................................................... 71 Table 31 – 802.16e low-IF low gain mode BER. ................................................................................... 74 Table 32 – Adjacent channel rejection ratios ........................................................................................ 84 Table 33 – 802.11g low-IF high gain mode EVM. ................................................................................. 84 Table 34 – 802.11g low-IF low gain mode EVM.................................................................................... 84 Table 35 – 802.16e low-IF high gain mode EVM. ................................................................................. 85 Table 36 – 802.16e low-IF low gain mode EVM.................................................................................... 85 Table 37 – 802.11g low-IF imbalances. ................................................................................................ 86 Table 38 – 802.11g low-IF linearity. ...................................................................................................... 86
viii
Table 39 – 802.16e low-IF imbalances. ................................................................................................ 87 Table 40 – 802.16e low-IF linearity. ...................................................................................................... 87 Table 41 – 802.11g low-IF results. ........................................................................................................ 87 Table 42 – 802.16e low-IF results. ........................................................................................................ 88
ix
List of Figures Figure 1 – Number of WiMAX Users by Region. ..................................................................................... 5 Figure 2 – Distortion caused by two adjacent channels. ....................................................................... 12 Figure 3 – Distortion caused by two interferes. ..................................................................................... 12 Figure 4 – Illustration of DR. .................................................................................................................. 12 Figure 5 – Illustration of SFDR. ............................................................................................................. 12 Figure 6 – (a) Interferer accompanying the received signal, (b) effect in time domain [24]. ................. 13 Figure 7 – Spectrum with potential blockers below 6 GHz [28]............................................................. 14 Figure 8 – Downconversion performed by the mixer [29]. .................................................................... 14 Figure 9 – Downconversion to IF [29]. .................................................................................................. 15 Figure 10 – Basic heterodyne architecture [29]. ................................................................................... 15 Figure 11 – Band selection and channel filtering [29]. .......................................................................... 15 Figure 12 – Image frequency problem [29]. .......................................................................................... 16 Figure 13 – Heterodyne architecture with an image-reject filter [29]. ................................................... 16 Figure 14 – Low IF trade-off [29]. .......................................................................................................... 16 Figure 15 – High IF trade-off [29]. ......................................................................................................... 17 Figure 16 – Dual-IF architecture [29]. .................................................................................................... 17 Figure 17 – Half IF problem [29]. ........................................................................................................... 18 Figure 18 – Quadrature demodulator [29]. ............................................................................................ 19 Figure 19 – Heterodyne architecture with I/Q demodulation [29]. ......................................................... 19 Figure 20 – Hartley architecture [29]. .................................................................................................... 20 Figure 21 – Weaver architecture [29]. ................................................................................................... 21 Figure 22 – Downconversion in Weaver architecture [29]. ................................................................... 22 Figure 23 – Secondary image problem [29]. ......................................................................................... 22 Figure 24 – Low-IF architecture. ........................................................................................................... 23 Figure 25 – Low-IF quadrature demodulation [29]. ............................................................................... 23 Figure 26 – Homodyne architecture [29]. .............................................................................................. 24 Figure 27 – Downconversion to BB [29]. ............................................................................................... 24 Figure 28 – DC Offset caused by LO self-mixing [29]. .......................................................................... 25 Figure 29 – DC Offset caused by a strong interferer [29]. .................................................................... 25 Figure 30 – Time varying offset [29]. ..................................................................................................... 25 Figure 31 – Even-order distortion. ......................................................................................................... 26 Figure 32 – Illustration of EVM. ............................................................................................................. 29 Figure 33 – Illustration of constellation points with different powers. .................................................... 30 Figure 34 – 802.11g transmitted spectrum mask. ................................................................................. 31 Figure 35 – Illustration of the OFDM concept........................................................................................ 32 Figure 36 – OFDM signal generation. ................................................................................................... 32 Figure 37 – IFFT operation. ................................................................................................................... 32 Figure 38 – OFDM and OFDMA time slots arrangement. ..................................................................... 33 Figure 39 – 802.11g transmitted spectrum mask. ................................................................................. 37
x
Figure 40 – 802.16e transmitted spectrum mask. ................................................................................. 37 Figure 41 – BER vs Eb/N0 for 4-QAM, 16 QAM, 64-QAM, and 1024-QAM modulations [42]. .............. 38 Figure 42 – 802.11g/802.16e Transmitter. ............................................................................................ 40 Figure 43 – 64-QAM constellation using Gray code. ............................................................................ 40 Figure 44 – OFDM subcarriers [45]. ...................................................................................................... 41 Figure 45 – Zero Padding used to shift aliases [46]. ............................................................................. 41 Figure 46 – OFDM subcarriers mapping [45]. ....................................................................................... 41 Figure 47 – OFDM symbol time structure [48]. ..................................................................................... 42 Figure 48 – 802.11g OFDM transmitted spectrum. ............................................................................... 43 Figure 49 – 802.16e OFDM transmitted spectrum. ............................................................................... 43 Figure 50 – 802.11g/802.16e receiver chain. ........................................................................................ 47 Figure 51 – 802.11g/802.16e Zero-IF Architecture. .............................................................................. 49 Figure 52 – 802.11g zero-IF low gain mode specifications. .................................................................. 51 Figure 53 – 802.11g zero-IF low gain mode constellation. ................................................................... 51 Figure 54 – 802.11g zero-IF low gain mode bitstreams. ....................................................................... 51 Figure 55 – 802.11g zero-IF low gain mode IP1dB. .............................................................................. 52 Figure 56 – 802.11g zero-IF low gain mode input power sweep. ......................................................... 52 Figure 57 – 802.11g zero-IF two-tone test. ........................................................................................... 53 Figure 58 – 802.11g zero-IF low gain mode IIP3. ................................................................................. 53 Figure 59 – 802.11g zero-IF low gain mode gain and NF. .................................................................... 53 Figure 60 – 802.16e zero-IF low gain mode specifications. .................................................................. 54 Figure 61 – 802.16e zero-IF low gain mode constellation. ................................................................... 54 Figure 62 – 802.16e zero-IF low gain mode bitstreams. ....................................................................... 55 Figure 63 – 802.16e zero-IF low gain mode IP1dB. .............................................................................. 55 Figure 64 – 802.16e zero-IF low gain mode input power sweep. ......................................................... 56 Figure 65 – 802.16e zero-IF two-tone test. ........................................................................................... 56 Figure 66 – 802.16e zero-IF low gain mode IIP3. ................................................................................. 56 Figure 67 – 802.16e zero-IF high gain mode Gain and NF. .................................................................. 57 Figure 68 – 802.11g zero-IF high gain mode specifications. ................................................................ 57 Figure 69 – 802.11g zero-IF high gain mode constellation. .................................................................. 58 Figure 70 – 802.11g zero-IF high gain mode IP1dB. ............................................................................ 58 Figure 71 – 802.11g zero-IF high gain mode input power sweep. ........................................................ 58 Figure 72 – 802.11g zero-IF high gain mode IIP3. ................................................................................ 59 Figure 73 – 802.11g zero-IF high gain mode Gain and NF. .................................................................. 59 Figure 74 – 802.16e zero-IF high gain mode specifications. ................................................................ 59 Figure 75 – 802.16e zero-IF high gain mode constellation. .................................................................. 60 Figure 76 – 802.16e zero-IF high gain mode IP1dB. ............................................................................ 60 Figure 77 – 802.16e zero-IF high gain mode input power sweep. ........................................................ 61 Figure 78 – 802.16e zero-IF high gain mode IIP3. ................................................................................ 61 Figure 79 – 802.16e zero-IF high gain mode Gain and NF. .................................................................. 61
xi
Figure 80 – 802.11g zero-IF adjacent channel power. .......................................................................... 62 Figure 81 – 802.11g zero-IF adjacent channel and desired signal. ...................................................... 62 Figure 82 – 802.11g zero-IF adjacent channel rejection test constellation. .......................................... 63 Figure 83 - 802.16e zero-IF adjacent channel power. .......................................................................... 63 Figure 84 – 802.16e zero-IF adjacent channel and desired signal. ...................................................... 63 Figure 85 – 802.16e zero-IF adjacent channel rejection test constellation. .......................................... 64 Figure 86 – Model EVM_WithRef [2]. .................................................................................................... 64 Figure 87 – 802.11g zero-IF high gain mode EVM constellation. ......................................................... 65 Figure 88 – 802.11g zero-IF low gain mode EVM constellation............................................................ 65 Figure 89 – 802.16e zero-IF high gain mode EVM constellation. ......................................................... 66 Figure 90 – Zero-IF 802.16e low gain mode EVM constellation. .......................................................... 66 Figure 91 – 802.11g/802.16e low-IF Architecture. ................................................................................ 69 Figure 92 – 802.11g low-IF low gain mode specifications. ................................................................... 70 Figure 93 – 802.11g low-IF low gain mode constellation. ..................................................................... 70 Figure 94 – 802.11g low-IF low gain mode bitstreams. ........................................................................ 71 Figure 95 – 802.11g low-IF low gain mode IP1dB. ............................................................................... 71 Figure 96 – 802.11g low-IF low gain mode input power sweep. ........................................................... 72 Figure 97 – 802.11g low-if two-tone test. .............................................................................................. 72 Figure 98 – 802.11g low-IF low gain mode IIP3. ................................................................................... 72 Figure 99 – 802.11g low-IF low gain mode Gain and NF. ..................................................................... 73 Figure 100 – 802.16e low-IF low gain mode specifications. ................................................................. 73 Figure 101 – 802.16e low-IF low gain mode constellation. ................................................................... 74 Figure 102 – 802.16e low-IF low gain mode bitstreams. ...................................................................... 74 Figure 103 – 802.16e low-IF low gain mode IP1dB. ............................................................................. 75 Figure 104 – 802.16e low-IF low gain mode input power sweep. ......................................................... 75 Figure 105 – 802.16e low-IF two-tone test. ........................................................................................... 75 Figure 106 – 802.16e low-IF low gain mode IIP3. ................................................................................. 76 Figure 107 – 802.16e low-IF high gain mode Gain and NF. ................................................................. 76 Figure 108 – 802.11g low-IF high gain mode specifications. ................................................................ 77 Figure 109 – 802.11g low-IF high gain mode constellation. .................................................................. 77 Figure 110 – 802.11g low-IF high gain mode IP1dB. ............................................................................ 77 Figure 111 – 802.11g low-IF high gain mode IP1dB. ............................................................................ 78 Figure 112 – Low-IF 802.11g high gain mode IIP3. .............................................................................. 78 Figure 113 – 802.11g low-IF high gain mode Gain and NF. ................................................................. 78 Figure 114 – 802.16e low-IF high gain mode specifications. ................................................................ 79 Figure 115 – 802.16e low-IF high gain mode constellation. .................................................................. 79 Figure 116 – 802.16e low-IF high gain mode IP1dB. ............................................................................ 80 Figure 117 – 802.16e low-IF high gain mode input power sweep. ....................................................... 80 Figure 118 – 802.16e low-IF high gain mode IIP3. ............................................................................... 80 Figure 119 – Low-IF 802.16e high gain mode Gain and NF. ................................................................ 81
xii
Figure 120 – 802.11g low-IF adjacent channel power. ......................................................................... 81 Figure 121 – 802.11g low-IF adjacent channel and desired input signal. ............................................. 82 Figure 122 – 802.11g low-IF adjacent channel rejection test constellation. ......................................... 82 Figure 123 – 802.16e low-IF adjacent channel power. ......................................................................... 82 Figure 124 – 802.16e low-IF adjacent channel spectrum and desired input signal. ............................. 83 Figure 125 – 802.16e low-IF adjacent channel rejection test constellation. ......................................... 83 Figure 126 – Low-IF 802.11g high gain mode EVM constellation......................................................... 84 Figure 127 – 802.11g low-IF low gain mode EVM constellation. .......................................................... 85 Figure 128 – 802.16e low-IF high gain mode EVM constellation. ......................................................... 85 Figure 129 – Low-IF 802.16e low gain mode EVM constellation. ......................................................... 86 Figure 130 – Dual-band zero-IF architecture. ....................................................................................... 89 Figure 131 – Dual-band low-IF architecture. ......................................................................................... 90 Figure 132 – 802.11g zero-IF receiver. ................................................................................................. 93 Figure 133 – 802.11g low-IF receiver. ................................................................................................... 93 Figure 134 – 802.16e zero-IF receiver. ................................................................................................. 93 Figure 135 – 802.16e low-IF receiver. ................................................................................................... 93 Figure 136 – 802.11g transmitter. ......................................................................................................... 94 Figure 137 – 802.16e transmitter .......................................................................................................... 94 Figure 138 – 802.11g/802.16e demodulator. ........................................................................................ 94 Figure 139 – 802.11g zero-IF receiver. ................................................................................................. 94 Figure 140 – 802.16e zero-IF receiver. ................................................................................................. 95 Figure 141 – 802.11g low-IF receiver. ................................................................................................... 95 Figure 142 – 802.16e low-IF receiver. ................................................................................................... 95
xiii
List of Abbreviations ADC Analog to Digital Converter AP Access Point AWGN Additive White Gaussian Noise BB Baseband BER Bit Error Rate BPF Band Pass Filter BW Bandwidth BWA Broadband Wireless Access CCK Complementary Code Keying CMOS Complementary Metal Oxide Semiconductor CP Cyclic Prefix DAC Digital to Analogue Converter DCR Direct Conversion Receiver DR Dynamic Range DSP Digital Signal Processor EDA Electronic Design Automation ERP Extended Rate PHY FDD Frequency Division Duplex FFT Fast Fourier Transform ICI Inter Channel Interference IEEE Institute of Electrical and Electronics Engineers IF Intermediate Frequency IFFT Inverse Fast Fourier Transform IIP3 Input Referred Third Order Intercept Point IMT International Mobile Telecommunications IP1dB Input 1 dB Compression Point IP2 2nd Order Intercept Point ISI Inter Symbol Interference ISP Internet Service Provider LNA Low Noise Amplifier LO Local Oscillator LOS Line of Sight LPF Low Pass filter LTE Long Term Evolution MOS Metal Oxide Semiconductor NF Noise figure NLOS Non Line of Sight OFDM Orthogonal Frequency Division Multiplexing OFDMA Orthogonal Frequency Division Multiple Access OIP3 Output Referred Third Order Intercept Point PAPR Peak to Average Power Ratio PGA Programmable Gain Amplifier PHY Physical Layer QAM Quadrature Amplitude Modulation RF Radio Frequency SC Single Carrier SFDR Spurious Free Dynamic Range SNR Signal to Noise Ratio TDD Time Division Duplex VREF Reference Voltage WiMAX Worldwide Interoperability for Microwave Access WLAN Wireless Local Area Network WMAN Wireless Metropolitan Area Network
xiv
1 Introduction
1.1 Motivation
1.1.1 System Level Design
Nowadays, the system level design of a receiver is mainly accomplished using spreadsheets.
This method has many limitations, including the number of different topologies that can be explored
within a deadline is reduced and when block requirements are changed, it takes too long to transfer
this changes to circuit level. Leaving most of the work dependent on the designer’s experience, this
produces weak solutions in functionality and quality. More effective approaches are necessary, since
radio frequency (RF) components have a longer design cycle when compared with digital ones.
However as years pass by, the wireless market demands a higher level of integrability and a
shorter time to market. As it is known these two requirements don’t always combine. In this sense
Agilent developed the two Electronic Design Automation (EDA) tools that will be used in this thesis,
Genesys [1] and Ptolemy [2], simplifying the designer’s work on the design flow and system level
optimization and its validation [3].
1.1.2 Dual Band Advantages
Institute of Electrical and Electronics Engineers (IEEE) 802.11 Wireless Local Area Network
(WLAN), also known as WiFi, is most certainly the wireless technology with the highest acceptance
worldwide, becoming nowadays the default interface in almost every electronic device. Meanwhile,
Mobile Worldwide Interoperability for Microwave Access (WiMAX) continues to attract a lot of attention
in the telecommunication world, especially by manufacturers and Internet Service Providers (ISP).
Mobile WiMAX (IEEE 802.16e), the first standard to introduce mobility to WiMAX, can be viewed as
the bridge between high speed WLANs and high mobility of cellular networks.
These two technologies are quite complementary when compared:
• Mobile WiMAX was developed for Wireless Metropolitan Area Network (WMAN) with a
coverage range of a few kilometers, being able to cover entire cities. While WiFi is for WLAN with a
transmission range up to 100m, proving mostly indoor coverage within hotspot, campus, enterprise
and home environments;
• Mobile WiMAX is mostly for commercial networks deployed by ISPs. However, WiFi is
mainly for non-commercial usage, such as home and enterprise networks;
• Mobile WiMAX supports high mobility so that users can have Internet access inside a
moving car or a train up to 120 km/h, but WiFi is mainly for nomadic users, who use this technology
1
while barely moving from the same place. Altough WiFi devices are not optimized to support mobility,
they can accept nothing more than the “walking speed” velocity [4].
Merging Mobile WiMAX and WiFi gives a true meaning to the “Internet on the Go” concept. By
combining the high speeds of both WLAN and WMAN, ISPs are able to provide fixed, portable and
mobile broadband internet access. And obviously, combining these two technologies produces a
major saving in device costs [5].
1.2 State of the Art
1.2.1 Standards Evolution
1.2.1.1 Mobile WiMAX
The development of the IEEE 802.16 standard on Broadband Wireless Access (BWA) started
in 1999, by IEEE 802.16 Working Group. The initial goal was only to provide fixed wireless services,
but it was expanded to offer mobility in IEEE 802.16e. WiMAX Forum is responsible for the
commercial profile of IEEE 802.16 standard since 2001.
The fixed BWA service in Line of Sight (LOS) environment of 10-66 GHz band was approved
in 2001 (IEEE 802.16-2001). In 2003, IEEE 802.16a standard was developed in the Non Line of Sight
(NLOS) environment of 2-11 GHz band including three types of Physical Layer (PHY) layers, Single
Carrier (SC), Orthogonal Frequency Division Multiplexing (OFDM), and Orthogonal Frequency
Division Multiple Access (OFDMA). Later, these standards were revised by IEEE 802.16d and its final
version, IEEE 802.16-2004, was approved.
In 2002, IEEE 802.11e was introduced to enhance the standards by offering mobility and it
was approved in December of 2005. Its system profile was released in February 2006. Table 1 shows
the main characteristics of the IEEE 802.16 standards [4].
Table 1 – IEEE 802.16 Standards.
802.16-2001 802.16a 802.16d 802.16e
Frequency band
10-66 GHz (LOS)
2-11 GHz (NLOS) 10-66 GHz (LOS)
2-11 GHz (NLOS) 10-66 GHz (LOS) 2-11 GHz (NLOS)
PHY layer SC SC, OFDM, OFDMA
SC,OFDM, OFDMA
SC, OFDM, OFDMA
Duplex TDD, FDD TDD, FDD TDD, FDD TDD, FDD
Mobility Fixed Fixed Fixed Mobile Release
date Apr. 2002 Apr. 2003 Oct. 2004 Feb. 2006
2
The product support and certification began in 2007. Table 2 describes the band class groups
in release 1.0 [6].
Table 2 – Mobile WiMAX Product Certification.
Band Class
Frequency band (GHz)
Bandwidth (MHZ) Region
1.A 2.3-2.4
8.75 Korea, South Asia
1.B 5 & 10
2.A 2.305-2.320, 2.345-2.360
3.5
United States/Canada 2.B 5
2.C 10
3.A 2.496-2.690 5 & 10 United States/Europe
4.A
3.3-3.4
5
China/India 4.B 7
4.C 10
5.A
3.4-3.8
5
Europe/Asia 5.B 7
5.C 10
The IEEE 802.16-2009 standard, which contains some enhancements relative to IEEE
802.16e, is the second revision of the IEEE 802.16 standard. It acts as base standard for the IEEE
802.16m. IEEE 802.16m is an amendment whose development started in 2007 by the IEEE 802.16
Working Group. It is an advanced air interface oriented to 4G networks and devices. The biggest
enhancement that it brought, is probably the fact that IEEE 802.16m systems will be able to support
transfer rates up to 1 Gbit/sec while maintaining backward compatibility with the existing Mobile
WiMAX systems [7]. 802.16m was approved as an International Mobile Telecommunications-
Advanced (IMT-Advanced) technology in 2010 [8]. Products, supporting 802.16m, were expected to
become available in 2012 [9].
1.2.1.2 WiFi
The IEEE standard 802.11 development on WLAN began in 1991 by the IEEE 802.11 Working
Group.
IEEE 802.11a was presented in 1999 and is based on OFDM at the 5 GHz band, while IEEE
802.11b, also presented in 1999, is based on the Complementary Code Keying (CCK) at the 2.4 GHz
band. The 802.11b products arrived at the market around the same time with great success due its
backward compatibility and fast transmission rates. IEEE 802.11a was later introduced to the market
3
in 2002 with great difficulties. It was not backward compatible and the devices were more expensive,
its higher transmission rates weren’t enough to overcome those two downsides.
IEEE 802.11g was introduced in 2003, in the same year compatible products arrived at the
market. IEEE 802.11g was defined as an extension of 802.11b and is backwards compatible with it
supporting the same rates. It also supports the same transmission rates as 802.11a since both use the
same modulation. This advantages ruined the small chances that 802.11a had on market acceptance.
Nowadays, IEEE 802.11g remains the most popular 802.11 PHY in the market at a global
scale. Mainly due to its fast transmission rates, lower cost devices and backward compatibility. Table 3
compares the various PHYs of the 802.11 [4].
Table 3 – IEEE 802.11 Standards.
1.2.2 Mobile WiMAX vs LTE
With Long Term Evolution (LTE) starting to make his way in the market of 4G networks, it is
inevitable the comparison that people are starting to make between LTE and Mobile WiMAX. Many
think that Mobile WiMAX is nothing more than a dead technology, but this is not even close to reality.
Others are confused and don’t know even what to think about that. Well, Mobile WiMAX and LTE are
quite similar with the only differences residing in speed and openness of the networks.
Network Speed
Mobile WiMAX is capable of transfer rates up to 30 Mb/s. LTE is faster offering speeds up to
100 Mb/s. However, 802.16m products will soon reach the market enabling Mobile WiMAX to provide
speeds up to 1 Gb/s.
Network Openness
LTE requires a SIM card in order to function, making it exclusive of some ISPs and devices
with SIM interfaces. Mobile WiMAX does not require a SIM card, making it a lot more compatible and
open technology. A Mobile WiMAX device with one Client ID can be used in various networks at
different locations. The reverse is also possible, the same device can be configured with multiple
Client IDs and used within the same network.
Standard Modulation Frequency band Transmission Rates (Mbps)
802.11a OFDM 5GHz 6, 9, 12, 18, 24, 36, 48, 54
802.11b CCK 2.4 GHz 1, 2, 5.5, 11
802.11g OFDM, CCK 2.4 GHz 6,9, 12, 18, 24, 36, 48, 54 + 802.11b rates
4
WiMAX has more supporters backing it such as Intel, Samsung, Alvarion, Beceem, Cisco,
Clearwire, Huawei, UQC, Yota, Airspan and ZTE [10]. And because of the two differences stated
above, Mobile WiMAX is expected to become faster, maintaining a wider global coverage than LTE in
a near future [11].
1.2.3 Wimax Growth (in the past 5 years)
In 2007, WiMAX Forum made a forecast [12] to predict WiMAX worldwide growth until 2012.
Figure 1 illustrates the user numbers evolution by major world region.
Figure 1 – Number of WiMAX Users by Region.
In February 15 of 2011, WiMAX Forum announced a WiMAX worldwide coverage of 823
million people in approximately 149 countries. This represented a growth of 215 million people since
December 2009 [13]. In August 16 of 2011, WiMAX Forum announced that the total number of 20
million global subscribers was reached. This was considered a dramatic growth, reported by operators
worldwide. Also in 2011, WiMAX equipment, by itself, was estimated to be a USD $2 billion industry.
It’s obvious that investments in WiMAX are still flowing. And considering that one subscriber can
represent more than one user, the 2007 forecast has all the chances to be reached or even surpassed
before the end of 2012 [14].
1.2.4 Plausible Architecture
Concerning the design of IEEE 802.11g standard WiFi receivers, the Direct Conversion
Receiver (DCR) has been chosen for highly integrated systems and low power consumption [15] [16].
For mobile receivers, the level of integration, flexibility, and power dissipation are crucial factors [17].
DCR is less complex than a heterodyne receiver, since it does not require an external image-reject
filter [18] and the receiver baseband (BB) stage includes only low-pass filters (LPF) and
programmable gain amplifiers (PGA) [15]. However, it has disadvantages such as DC offset that can
only be canceled with digital signal processor (DSP) help, quadrature demodulator impairments must
5
be minimized with proper layout design combined with DSP aid [19], and flicker noise that became
important due to signal with relatively low level at baseband circuits.
In the design of WiMAX receivers for the IEEE 802.16e standard, DCR architecture is mainly
chosen for the same reasons as WiFi receivers [20]. It requires less and simpler filters than the
heterodyne receiver. Band-pass filters (BPF) are not used because the intermediate frequency (IF) is
at DC, so the image rejection does not apply [21].
In the design of Mobile WiMAX/Wifi dual-band receivers, DCR architecture is used for the
same reasons referred above for the individual standard receivers [22]. Also, 802.11g the 802.16e
signals do not have a subcarrier at DC, which relaxes the requirements of carrier suppression and 2nd
order intermodulation, making the DCR the ideal choice [23].
If the designer is willing to invest in the DSP level, then the zero-IF architecture should be
chosen since its problems can be corrected in the digital domain lowering the analog to digital
converter (ADC) requirements.
Otherwise, there is the low-IF option, in which the DC offset is not a huge problem. But, since
the signal is centered at 10 MHz, this leads to a higher ADC sampling frequency and consequently to
higher power consumption in the IF chain.
1.3 Aim of this Thesis Receivers’ architectures need to be chosen and designed at a system level prior to circuit
implementation. That choice depends on the cost, power consumption, complexity and performance.
Performance is, the only factor, conditioned by the standards. Therefore, the specifications extraction
from the standards is very important.
The main goal of this thesis is to study which is the most suitable receiver architecture to fulfill
the specifications imposed by both the 802.11g and 802.16e standards, considering the aspects
referred above, and assuming a complementary metal oxide semiconductor (CMOS) technology
implementation.
1.4 Thesis Organization This thesis report is organized in five chapters. Chapter 1 is the introduction. Next chapters
are organized as follows:
• Chapter 2 provides an overview of various receiver design basics and architectures. It also
describes some transmitter design basics and introduces the OFDM modulation and its variations;
• Chapter 3 contains the standards analysis and specifications along with the transmitter and
demodulator components;
• Chapter 4 describes the steps taken in the single standard architectures design;
6
• Chapter 5 addresses the feasibility and design of the dual-band receiver;
• Chapter 6 concludes with a summary and presents future work.
7
8
2 System Level Concepts
2.1 Receiver Design Basics A receiver performance is usually limited by four factors: noise, nonlinearities, circuit
imbalances and components quality factor. The following parameters that are used to characterize a
receiver are related to these factors. A receiver must be capable to demodulate one channel from a
group of very close frequency channels that are associated with a certain communication system. This
system obeys to a certain standard. But in free space several other communication systems coexist,
and the receiver must reject them. This is why the receiver design is so demanding. In terms of
interferes they will be associated with other channels from the same system (in-band interferes) and
signals from other systems (out-of-band interferes).
2.1.1 Noise Factor
Noise factor can be seen as the measurement of the signal to noise ratio (SNR) degradation
between the input and output of a system. Since any real system is noisy, its output noise power will
be more amplified than the output signal power. This means that the SNR will always suffer a
reduction. The noise factor, F , is defined as,
/ 1/
in in
out out
S NFS N
= ≥ (1)
where inS and inN are the input signal and noise available powers, and outS and outN are the output
signal and noise available powers, respectively. The input noise is defined as thermal noise produced
from a matched load at 0 290T K= , that is, 0Ni kT B= , where k is the Boltzmann constant and B the
equivalent noise bandwidth (BW). Noise figure (NF) is obtained as ( ) 10logNF dB F= .
For a system with m stages in cascade the total F is given by the Friis equation,
321
1 1 2 1 ( 1)
1 11 ......
mtotal
m
F FFF FG G G G G −
− −−= + + + + (2)
where mF is the noise factor and mG the available power gain of m stage calculated with the
respective input impedance. It is possible to see in (2) that the noise characteristics of a cascaded
system are dominated by the first stage, therefore it is desirable that the first stage has a considerable
gain and low F [24].
9
2.1.2 Sensitivity
The sensitivity can be defined as the minimum signal level that a receiver can discern while
maintaining the service quality. Acceptable quality is usually quantified at the receiver output by the
minimum SNR value for analog modulated input signals, or maximum Bit Error Rate (BER) value for
digital modulated input signals. This problem arises because at the receiver output (at IF or BB), in
addition to the desired channel, other signals can appear at the same frequency. If the receiver input
has only the desired RF signal, the disturbing signal is usually in-band noise. If the receiver input has
other in-band and out-of-band interferes, the receiver output can present a fraction of those signals
that can degrade sensitivity. How these interfere signals are downconverted to output frequency will
be explained latter. Following the well-known sensitivity due to in-band noise is calculated. If at the
receiver input only the desired channel is present, the input noise plus the receiver internal generated
noise are the responsible for the sensitivity value.
The sensitivity calculation begins by expanding NF,
in
out
SNRNFSNR
= (3)
/in in
out
P NSNR
= (4)
where inP is the input signal power and inN the input noise power. By equating inP ,
in in outP N NF SNR= ⋅ ⋅ (5)
with input noise power given by,
0inN k T B= ⋅ ⋅ (6)
and B the equivalent channel BW. Substituting (6) in (5), after applying dBm to both sides,
0[ ] 10log( ) [ ]in outP dBm kT B NF SNR dB= + + (7)
Because 010 log( ) 174 /kT dBm Hz= − is a constant, a further simplification can be made
and the final equation for sensitivity is obtained,
[ ] [ ]min min174 / 10login outP dBm dBm Hz NF B SNR dB= − + + + (8)
The sum of the first three terms is the total noise of the system, also known as noise floor [24].
10
2.1.3 Selectivity
Selectivity is the receiver ability to demodulate the desired channel in the presence of other in-
band channels, being the adjacent channel rejection the most problematic.
The overall receiver selectivity is usually associated with IF and BB stages. Along the receiver
chain, active components have a non-linear behavior, which generates overload, modulation
distortion, spurious signals and spurious responses. To minimize these, frequency selectivity BPFs
and LPFs are included in the receiver. These filters improve the receiver selectivity especially in the
BB stage. Unfortunately, the filters quality factor isn’t high enough and trade-offs have to be done [25].
2.1.4 BER
When the RF signal information is digital, the receiver output BER parameter is frequently
used. It shows how many bits are incorrectly interpreted from a large number of bits. It can be viewed
as the rate at which the receiver misinterprets a ‘1’ by a ‘0’ and vice-versa. In RF design, the BER
requirement is usually converted to a SNR requirement. BER depends mainly on the type of
modulation used, coding techniques and demodulation method [26].
2.1.5 Non-linear Behavior
Non-linear behavior of active components produce intermodulation products and/or harmonics
on their outputs [27]. The intermodulation products appear at frequencies ,n mf ,
, 1 2n mf n f m f= ± ⋅ ± ⋅ (9)
where n and m are positive integer numbers and 1f and 2f the input signals frequencies. The order
of each intermodulation product is given by n m+ . Harmonic distortion is also described by (9) for
products where 0n = or 0m = .
Some intermodulation products may fall close or inside the desired channel with a power level
large enough to distort it. The 3rd order intermodulation products at frequencies 1 22 f f− and 2 12 f f−
are the most problematic when two input signals are close in frequency.
In Figure 2, the distortion of the system results from the interference between two adjacent
channels, which is caused by the fact that one of the 3rd order intermodulation products falls on top of
the desired channel. This can also occur by the combination of two out-of-band interferes (Figure 3).
11
Figure 2 – Distortion caused by two adjacent channels.
Figure 3 – Distortion caused by two interferes.
2.1.6 Dynamic Range
In RF design there are two definitions of dynamic range (DR) [24]. One, simply called dynamic
range, is defined as the division of the maximum tolerable input signal power by the sensitivity. DR is
limited by compression at the upper end and by the internal receiver noise at the lower end (Figure 4).
The other is called spurious free dynamic range (SFDR). The lower end is still the sensitivity.
The upper end is defined by the maximum input level in a typical two-tone test for which the 3rd order
intermodulation products equal the output receiver noise (Figure 5). The SFDR quantifies the
maximum level of interferers that a receiver can tolerate while still maintaining its service quality, even
in the presence of a small input signal.
Figure 4 – Illustration of DR.
Figure 5 – Illustration of SFDR.
12
2.1.7 Blocking and Desensitization
Desensitization lowers the SNR at the receiver output. This phenomenon derives from
compression that occurs when the received signal is accompanied by a large interferer, this is
illustrated in Figure 6(a). The large excursions of the interferer lead to a reduction of the receiver gain,
as illustrated in Figure 6(b).
Figure 6 – (a) Interferer accompanying the received signal, (b) effect in time domain [24].
Desensitization is mainly determined by the low noise amplifier (LNA) due to its compressing
characteristics. It is a non-linear device and its behavior can be described by,
2 31 2 3( ) ( ) ( ) ( )y t x t x t x tα α α≈ + + (10)
Therefore, desensitization can be quantified assuming that 1 1 2 2( ) cos cosx t A t A tω ω= + ,
where the first and second terms represent the desired signal and the interferer, respectively.
Considering the third order characteristic of (10), the output at 1ω is,
2 21 3 1 3 2 1 1
3 3( ) cos ...4 2
y t A A A tα α α ω = + + +
(11)
which, for 1 2A A , is reduced to,
21 3 2 1 1
3( ) cos ...2
y t A A tα α ω = + +
(12)
Thus, the gain of the desired signal is equal to 21 3 23 2Aα α+ , which is a decreasing function
of 2A if 1 3 0α α < , the usual case. If 2A is considerably large, the gain can even drop to zero. In this
case, it is said that the signal is blocked. The term blocker, in RF design, represents interferes that
desensitize the receiver even if the gain is not reduced to zero [24].
In order for a proper frequency plan to be determined, it is advisable that the designer knows
which wireless applications could act as blockers to the desired band. Some potential blockers are
shown in Figure 7.
13
Figure 7 – Spectrum with potential blockers below 6 GHz [28].
2.2 Receiver Architecture Overview
2.2.1 Heterodyne Architecture
It is an architecture that uses one or more IFs. The superheterodyne receiver was invented by
Armstrong in 1918, in this thesis it will be referred as heterodyne. Channel filtering is proven to be very
difficult at high frequencies. So, it was created a method of translating the desired signal to a much
lower frequency allowing channel filtering with a reasonable Q. Illustrated in Figure 8, a mixer is
responsible for this translation.
Figure 8 – Downconversion performed by the mixer [29].
The frequency of the signal,
( ) cos( )RF RF RFv t A tω= (13)
is downconverted by multiplying it with a sinusoid generated by the local oscillator (LO).
( ) cos( )LO LO LOv t A tω= (14)
In this way the impulses at LOω shift the desired signal to RF LOω ω± .
[ ]( ) cos( ) cos( ) cos( ) cos( )2
RF LOX RF LO RF LO RF LO
A Av t t t t tω ω ω ω ω ω= ⋅ = − + + (15)
The component at RF LOω ω+ is removed by the LPF in Figure 8, leaving the signal at a
frequency, called IF, of RF LOω ω− . This operation is called downconversion.
14
( ) cos( )2
RF LOIF RF LO
A Av t tω ω= − (16)
Heterodyne receivers use an LO frequency unequal to RFω , which results in an IF different
from zero (Figure 9).
Figure 9 – Downconversion to IF [29].
In reality, the signal received by the antenna has not only the desired signal but also blockers
and other interferers, sometimes even in the same band. So a realistic receiver needs more filtering.
Receivers incorporate a band-select filter (BPF1) in the front end stage, which selects the
entire band and rejects out-of-band interferers (Figure 10).
Figure 10 – Basic heterodyne architecture [29].
The type of filtering, in Figure 10 (BPF2), is called channel selection filtering, it selects the
desired signal channel rejecting the interferers in other channels (Figure 11).
Figure 11 – Band selection and channel filtering [29].
2.2.1.1 Image Frequency
Heterodyne receivers have a problem called the image frequency. Supposing that one
interferer exists at 2IM LO RFω ω ω= − . If BPF1 attenuation is not high enough, this signal is
downconverted to IF and corrupts the desired signal (Figure 12).
15
Figure 12 – Image frequency problem [29].
One solution consists in attenuating the image signal at the RF mixer input. This is
accomplished using an image-reject filter, BPF2 (Figure 13). This filter can have small losses in the RF
band with a large attenuation in the image band, only if 2 IFω is large enough.
Figure 13 – Heterodyne architecture with an image-reject filter [29].
2.2.1.2 Trade-off between Image Rejection and Selectivity
The desired channel and the image have a frequency difference equal to 2 IFω . Thus, for a
high image rejection a large value for IFω is necessary. But the principle of a heterodyne receiver
consists in translating the desired channel to a low enough frequency, so that the channel selection
filters become feasible. However, increasing IFω results in a higher Q for the IF filter. In Figure 14
and Figure 15, two cases corresponding to a high and low values of IF are shown to illustrate the
trade-off.
Low IF case
Figure 14 – Low IF trade-off [29].
16
Image rejection is worse because at high frequencies BPF2 is difficult to build with a high Q.
Channel filter BPF3 can have a high Q because IF is low enough. So channel selection is better, which
improves selectivity.
High IF case
Figure 15 – High IF trade-off [29].
Image rejection is better because IF is higher, so BPF2 is easier to build. Channel selection is
worse because channel filter BPF3 center frequency is higher.
If low IF solution is chosen, selectivity is better but residual image in the desired channel
reduces the receiver sensitivity. This is why in a receiver the trade-off between low and high IF is
designated trade-off between sensitivity and selectivity.
2.2.1.3 Dual IF
The image rejection selectivity trade-off can be improved if a dual IF topology (Figure 16) is
used. A higher IF1 can be used, simplifying or avoiding BPF2 and the selectivity is improved by BPF4 at
lower frequencies.
Figure 16 – Dual-IF architecture [29].
In the second downconversion the image problem can also appear, but because the frequency
is lower it is not so severe for BPF3.
2.2.1.4 Half IF Problem
If an interferer at ( ) / 2RF LOω ω+ appears at the antenna and is not filtered, two things can
happen (Figure 17):
17
• The interferer mixes with the LO and produces an intermodulation product at / 2IFω . If the
following stages produce 2nd order distortion it will fall on top of downconverted channel;
• The interferer suffers 2nd order distortion before the mixer and, if the LO has too much 2nd
order harmonic, it will fall on top of down-converted channel.
To minimize this problem, RF and IF paths must have low 2nd order distortion, the LO must
have low 2nd harmonic and the interferer must be filtered before the LNA.
Figure 17 – Half IF problem [29].
Important remarks about the possible heterodyne topologies:
• BPF1 is usually off-chip. Sometimes is the duplexer used in full-duplex transceivers;
• BPF2 is also off-chip because of its high Q value. But these filters are 50 Ω matched, so the
LNA output must be 50 Ω matched. Meaning that buffers must be used increasing power
consumption. The cost is also higher;
• Off-chip filters are less expensive if standard frequencies are chosen;
• Channel tuning is usually made by changing LO1 frequency. In this case LO2 frequency has
a fixed value.
2.2.1.5 Signal Demodulation
If the lower IF stage produces a slow enough signal for the ADC, then the signal demodulation
to BB can be done in the digital domain (DSP). The BB demodulation can also be done in the analog
domain, if the last IF is zero.
Modern communication systems use digital modulation techniques. This means that instead of
a single BB signal, there are two BB signals, I(t) and Q(t). Demodulation has to be done by mixing with
two signals in quadrature using a quadrature demodulator (Figure 18).
18
Figure 18 – Quadrature demodulator [29].
Assuming that the downconversion is to BB ( LO IFω ω= ), as illustrated in Figure 18,
quadrature downconversion is performed by mixing ( )x t with a LO with quadrature outputs.
( ) ( ) cos( ) ( )sin( )x IF x IFx t I t t Q t tω ω= + (17)
The resulting outputs are called quadrature BB signals,
( ) ( ) ( )( ) cos( ) cos(2 ) (2 )2 2 2
x x xI LO LO LO
I t I t Q ta x t t t sin tω ω ω= = − + (18)
( ) ( ) ( )( )sin( ) cos(2 ) (2 )
2 2 2x x x
Q LO LO LOQ t Q t I ta x t t t sin tω ω ω= = − + (19)
Although having the same frequency Ia e Qa are separated in phase, and when combined can
reconstruct the original information.
After low-pass filtering,
( )( )2
xI tI t = (20)
( )( )
2xQ tQ t = (21)
For digital I/Q demodulation a typical receiver is shown in Figure 19.
Figure 19 – Heterodyne architecture with I/Q demodulation [29].
19
Channel selection can be made at IF1 which means that LO2 has a fixed value. But another
recent solution is to vary both oscillators’ frequencies. That is called the sliding IF topology.
Because of the high IF1 value there is no need for the image-reject filter. Oscillators LO1 and
LO2 frequencies are proportional, meaning that only one synthesizer is needed. Also since LO and RF
signals have a large separation, crosstalk between them is reduced.
2.2.2 Image-Reject Architectures
Image-reject architectures suppress the image without filtering, avoiding the trade-off between
image rejection and selectivity. This architectures’ principle is to cancel the image at the output by
adding two signals with opposite signs.
2.2.2.1 Hartley Architecture
Hartley’s circuit mixes the input signal with the quadrature phases of the LO, low-pass filters
the resulting signals and shifts the in-phase one by -90º before adding them (Figure 20).
Figure 20 – Hartley architecture [29].
The input signal is ( ) cos( ) cos( )RF RF IM IMx t A t A tω ω= + , where the first term is the desired
signal and the second is the image.
Assuming low side injection ( IM LO RFω ω ω< < ), ( )x t is multiplied by the LO outputs and the
high-frequency terms removed by the LPFs. The following signals are obtained at points a and c,
( ) sin( ) sin( )2 2RF IM
RF LO LO IMA Aa t t tω ω ω ω= − − + − (22)
( ) cos( ) cos( )2 2RF IM
RF LO LO IMA Ac t t tω ω ω ω= − + − (23)
After the -90º phase-shift, the signal at point b is obtained,
20
( ) cos( ) cos( )2 2RF IM
RF LO LO IMA Ab t t tω ω ω ω= − − − (24)
At the output the image is canceled by the addition of ( )b t and ( )c t , the resulting signal is
( ) cos( )RF RF LOy t A tω ω= − .
The main drawback of Hartley architecture is its sensitivity to amplitude and phase
mismatches between LO signals. In a realistic design other mismatches exist, for example, in the
mixers and in the phase-shifter.
2.2.2.2 Weaver Architecture
The Weaver architecture avoids the issues of Hartley architecture. As shown in Figure 21, the
Weaver architecture replaces the 90º phase-shifter with another quadrature mixing.
Figure 21 – Weaver architecture [29].
Beginning with the signals 1( )a t and 2 ( )a t after the first quadrature downconversion, to
formulate the architecture behavior (Figure 22),
1 1 1( ) sin( ) sin( )2 2RF IM
RF LO LO IMA Aa t t tω ω ω ω= − − + − (25)
2 1 1( ) cos( ) cos( )2 2RF IM
RF LO LO IMA Aa t t tω ω ω ω= − + − (26)
Simplifying,
1 1( ) sin( )2
IM RFIF
A Aa t tω−= (27)
2 1( ) sin( )2
IM RFIF
A Aa t tω+= (28)
21
The second quadrature mixing operation is performed resulting in,
[ ]1 1 2 1 2( ) sin( ) sin( )4
IM RFIF LO IF LO
A Ab t t tω ω ω ω−= − + + (29)
[ ]2 1 2 1 2( ) sin( ) sin( )4
IM RFIF LO IF LO
A Ab t t tω ω ω ω+= − − + + (30)
Neglecting the high frequency terms in 1( )b t and 2 ( )b t ,
2( ) sinRF IFy t A tω= (31)
Figure 22 – Downconversion in Weaver architecture [29].
The image in the first stage is suppressed, but if an image interferer appears in second stage
input it will lead to a distortion of the desired channel (Figure 23). This can happen if an interferer
appears at the input with a frequency equal to 1 22 2LO LO RFω ω ω+ − .
Figure 23 – Secondary image problem [29].
If the second stage makes a BB direct conversion the problem does not exist. If not, the LPF
must be replaced by a BPF for 1IFω .
2.2.3 Low-IF Architecture
In Figure 24 is presented the low-IF architecture. Its approach is quite similar to the Weaver
architecture, with the exception that the signals 1( )a t and 2 ( )a t are sent to the ADCs. The second
down-conversion to BB is performed in the digital domain. Low-IF has all the Homodyne advantages
22
with no DC offset problem, but it may suffer from I/Q impairments in the RF analogue quadrature
downconversion.
IF can be as low as half of the channel BW. But such a low IF implies that the analog
quadrature demodulator is working close to the RF frequency. This leads to an imperfect image
rejection due to the demodulator impairments. A proper layout design and techniques based on RF
complex filters placed before the quadrature demodulator are used to reduce this limitation [29].
Figure 24 – Low-IF architecture.
Assuming a perfect quadrature demodulator, the complex spectrum at its output, Y, is a left
shifted version by LO frequency of the RF spectrum, X, as shown in Figure 25.
Figure 25 – Low-IF quadrature demodulation [29].
But in practice, IF I/Q chain is not perfectly symmetric, so quadrature relation between IF
image and signal will be corrupted. A solution for this, is to place a complex BPF at the beginning of
the IF chain, centered at +IF, attenuating the –IF image and high order products. Thus, if the following
IF stages (PGA and ADC) are not symmetrical, the signal is lesser corrupted since the image was
already attenuated the complex BPF. Finally the channel selection is performed at BB by a LPF [29].
2.2.4 Homodyne Architecture
Also known as DCR or zero-IF receiver, this architecture (Figure 26) has some issues, but it is
still very popular in wireless telecommunications.
23
Figure 26 – Homodyne architecture [29].
It is an architecture that downconverts the RF signal directly to BB. This means that
LO RFω ω= and 0IFω = (Figure 27).
Figure 27 – Downconversion to BB [29].
The major advantage of the homodyne receiver is that it doesn't suffer from the image
frequency problem, so no image-reject filter is needed. Also, IF filters are not necessary because there
is no IF. Channel select LPF is usually active. A good trade-off noise-linearity-power must exist
between this filter, the ADC and a possible amplifier between them.
2.2.4.1 Issues [29]
LO leakage emission
Due to the fact that the LO and the RF frequencies are the same and the LO-RF mixer
isolation is not so high, part of the LO signal can be radiated by the antenna.
DC offsets
If at the RF and LO mixer inputs the same interferer signal appears, on the mixer IF port a DC
component is generated. This component can saturate the following stages and ruin the receiver
performance. This is called LO self-mixing and can happen in several ways that are described below:
• Due to low mixer LO-RF isolation, the LO signal is successively reflected in the RF chain
ports mismatches and mixes with itself (Figure 28).
24
Figure 28 – DC Offset caused by LO self-mixing [29].
• A strong interferer can cross the RF chain, leaking to the LO port and mixing with itself
(Figure 29).
Figure 29 – DC Offset caused by a strong interferer [29].
• The previous effects produce time-invariant DC offsets. The following one produces a time
varying offset (Figure 30) which is very difficult to distinguish from the desired signal.
Figure 30 – Time varying offset [29].
Any leak from the LO to the antenna can reflect in a moving object and produces a time
varying self-mixing effect. This is worst if the receiver is moving.
Some solutions to solve the problems of these DC offsets:
• The DC component is filtered with an HPF. This increases the BER because the signal has
at DC its higher energy. There are modulation techniques that reduce the spectrum importance near
DC (DC-free coding) but this is more appropriate to wideband channels;
25
• Time invariant offset is corrected by measuring the output offset, passing through an ADC
and storing it in a memory. This is done in the receiver idle state. In the burst mode the memory value
is added (or subtracted) at the mixer output to eliminate the offset;
• Time varying offset is corrected by averaging the output with interferes and offset through
an ADC and storing it in a fast memory. The interferers are random, but the time invariant offset aren’t.
Then the value is added (or subtracted) at the mixer output to eliminate the offset when the receiver is
on.
Even-order distortion
Assuming that two strong interferences appear at the LNA input close to the desired frequency
(Figure 31).
Figure 31 – Even-order distortion.
1 1 2 2cos cosa RF RFV A t A tω ω= + (32)
If the LNA experiences some even-order distortion, 21 2 ...b a aV g V g V= + + , where 1g is the
LNA gain and 2g the 2nd order distortion coefficient. The low frequency beat at 1 2RF RFω ω− and the
DC component caused by the 2nd order distortion can be expressed as
2 2
1 22 2 1 2 1 2cos( ) ...
2b RF RFA AV g g A A tω ω
+= + − +
(33)
In reality, all mixers have some feedthrough from RF to IF, and because of that a DC
component will corrupt the BB signal. Also, since 1 2RF RFω ω≈ the second term will fall inside BB
signal. To minimize this problem a LNA with a low value of 2g must be designed, that is, with a low
2nd order intercept point (IP2) value.
Now supposing that a strong AM interferer appears at the LNA input close to the desired
frequency
[ ]1 11 ( ) cos cosa m RFV A m t t tω ω= + ⋅ (34)
26
The 2nd order distortion produces the following low frequency products and the DC offset,
2 2 2
12
( ) ( )1 2 ( )cos( ) cos(2 ) ...2 2 2a m m
A m t m tV g m t t tω ω
= + + + +
(35)
The low frequency modulating signal produces DC and close to DC interferers. The mixer is
also a source of 2nd order distortion, sometimes even more important than the LNA. If the Va
interferers appear at Vb, the same effect in the mixer stage will produce DC and close to DC interferers
at BB. Once again, a good IP2 mixer must be designed.
I/Q mismatch
Errors in the amplitude and quadrature relations of the quadrature demodulator will affect the
BB constellation. If only amplitude mismatch occurs, a gain factor affects the BB signals, which is not
too severe. On the other hand, if quadrature mismatch occurs it will degrade the BB signals in a way
that each one has a small portion of the other.
Flicker noise
The gain between the antenna and the mixer output is not very high (around 30dB), resulting
in a mixer output BB signal with a low magnitude. The BB amplifiers and filters can introduce 1/f noise
that corrupts the SNR close to DC. This effect can be minimized by:
• Using an active mixer instead of a passive one to increase front-end gain.
• Using large metal oxide semiconductor (MOS) devices to minimize 1/f noise.
• Usually DC offset cancellation schemes can reduce 1/f noise components bellow 1/Tc,
where Tc is the offset cancellation period.
• Using DC-free coding on the signal. If high-pass filtering is done, 1/f noise is reduced and
does not affect the signal.
27
2.2.5 Comparison of Receiver Architectures
The advantages and disadvantages between heterodyne, low-IF and zero-IF architectures are
summarized in Table 4.
Table 4 – Comparison of Receiver Architectures.
Advantages Disadvantages
Heterodyne Relaxes specs of RF, IF
and BB circuits Minor impairments in
I/Q demodulator
× Image-frequency filter × Complex circuit
Low-IF Less filters Easier image rejection
× Impairments in I/Q demodulator
Homodyne Simpler circuit No image rejection
problem Less filters
× 1/f noise and DC offset kill the performance
× Slow DC offset settling time × Impairments in I/Q
demodulator × Local oscillator pulling
2.3 Transmitter Design Basics Error vector magnitude (EVM) and spectrum mask are two major specifications of a
transmitter. EVM can be seen as a number that indicates the modulation quality of a signal being
related to the BER in a digital transmission. The spectrum mask specifies the attenuation of the
adjacent and non-adjacent channel interferers at frequencies beyond the desired channel BW.
2.3.1 EVM
The EVM takes into account all the effects that can degrade the modulation quality, such as
amplifier nonlinearities, amplitude and phase impairments from quadrature mixing, noise, image
rejection, and digital to analogue converter (DAC) inaccuracies.
The EVM measures the deviation of a sampled vector from the ideal vector, this is illustrated
in Figure 32.
28
Figure 32 – Illustration of EVM.
EVM is defined as the error vector magnitude normalized to the ideal vector magnitude,
error
reference
XEVM
X= (36)
Considering vector’s magnitude in terms of power, EVM can also be expressed as,
P
error
reference
PEVM = (37)
in decibels,
10 10( ) 20log 10logerror error
referencereference
X PEVM dBPX
= = (38)
or as a percentage,
(%) 100%error
reference
XEVM
X= × (39)
The constellation point with the highest power will be used as a reference, and the error at
each constellation point will be averaged if the obtained constellation points have different powers.
This is illustrated in Figure 33.
29
Figure 33 – Illustration of constellation points with different powers.
And can be expressed as,
2 2
12max
1 ( )N
i ii
I QNEVM
X=
∆ + ∆=
∑ (40)
where N is the number of constellation points, Xmax is the magnitude of the reference vector to
constellation point with the highest power, and ∆Ii and ∆Qi are offsets of the measured constellation
point and the ideal constellation point [30]. Before the calculation of EVM, the demodulated
constellation must be normalized in terms of average power signal to equalize the reference
constellation.
2.3.2 Spectrum Mask
When a modulated signal goes through a nonlinear system, its BW is enlarged by odd-order
nonlinearities. This is caused by the created mixing products between the individual components of
the modulated signal.
These intermodulation products cause another problem, they create interference with adjacent
channels. This happens because channels are very close to each other in order to maximize spectrum
efficiency. This is the main reason why the standards require accordance to a spectrum mask.
As an example, the 802.11g signal is formed by 52 subcarriers with modulated data around
each one. The data contained in each subcarrier can nonlinearly interact with itself, causing
intermodulation distortion. An ideal transmitted OFDM signal should appear like the one shown in
Figure 34.
30
The 802.11g spectrum mask requires that the transmitted signal spectrum to be more than 20,
28, and 40 dBc below the peak of the modulated signal with offset frequencies of 11, 20, and 30 MHz,
respectively, away from the center of the desired channel.
Figure 34 – 802.11g transmitted spectrum mask.
The accomplishment of the spectrum mask specifications by a properly designed system is
conditioned by the intermodulation products, which are a consequence of the non-linearity of the
power amplifier [31].
2.4 Modulation Basics The modulation scheme used in 802.11g is Extended Rate PHY-OFDM (ERP-OFDM). The
Mobile WiMAX uses OFDMA as the radio access method for improved multipath performance in
NLOS environments.
2.4.1 OFDM
An OFDM channel uses multiple low rate modulated subcarriers closely spaced to each other.
The subcarriers sidebands overlap, but because the subcarriers are mathematically orthogonal there
is now interference. So, OFDM modulation has a high spectral efficiency.
In OFDM, the high rate input stream is divided into multiple parallel low rate substreams.
Then, each low rate substream modulates one individual subcarrier. The modulation type is usually a
spectral efficient one. The modulation used in this thesis will be 64-quadrature amplitude modulation
(64-QAM) because it is the most demanding in the standards. At last, all the subcarriers are
transmitted at the same time [32]. Figure 35 illustrates the OFDM concept.
31
Figure 35 – Illustration of the OFDM concept.
Figure 36 shows in more detail how an OFDM signal is built before being transmitted.
Figure 36 – OFDM signal generation.
A serial bitstream is converted into N parallel streams, each one being mapped to a symbol
using QAM modulation. An inverse Fast Fourier Transform (IFFT) is computed on each set of
symbols, giving as result a set of BB complex time-domain signals (Figure 37 [33]).
Figure 37 – IFFT operation.
The IFFT of a signal ak can be expressed as,
21
0( )
j ktNT
kk
v t a eπ−
=
=∑ (41)
32
where ak is the complex I/Q symbol that modulates the kth complex carrier, N is the number of
subcarriers and T is the OFDM symbol time, being 1/T the carrier spacing.
The complex output (with rectangular coordinates I and Q) is upconverted to the carrier
frequency. Finally the radiofrequency I and Q components are summed to form the transmitted signal
[34] [35],
2( ) Re[ ( ) ]cj f tks t v t a e π= (42)
[ ]1
0cos(2 / arg[ ])
N
k c kk
a f k T t aπ−
=
= + +∑ (43)
2.4.2 OFDMA
OFDMA is a multi-user access technique based on OFDM modulation that allows sharing the
same channel by several users. A subset of carriers is assigned to one user, so a subchannel is
created for that user. For the same user more than one subchannel can be assigned, these
subchannel are not necessarily adjacent.
Subchannelization defines which subchannels can be assigned to the mobile units depending
on their channel conditions and data requirements. In this way, a mobile WiMAX base station can
allocate within the same time slot more transmit power for lower SNR units and less power for higher
SNR units.
In OFDM only one mobile unit can transmit in one time slot and use the entire channel, this is
illustrated in Figure 38(a). In OFDMA, several mobile units are able to transmit at the same time slot in
more than one subchannel, as illustrated in Figure 38(b) [32].
Figure 38 – OFDM and OFDMA time slots arrangement.
33
34
3 Standards Analysis and Specifications
3.1 Standards Analysis Every integrated circuit project starts with a preliminary task known as system level design.
The purpose of this systemic task consists in determining the individual specifications of each receiver
component, in a way that the overall receiver fulfills the standard specifications.
The IEEE 802.11g [36] and 802.16e [37] were analyzed and the following requirements were
obtained.
3.1.1 Maximum Input Signal
IEEE 802.11g Requirement: The maximum signal level at the receiver input is –20 dBm.
IEEE 802.16e Requirement: The maximum signal level at the receiver input is –30 dBm.
3.1.2 Sensitivity
IEEE 802.11g Requirement: The receiver input sensitivity is -65 dBm.
IEEE 802.16e Requirement: The receiver input sensitivity is derived according to [37]:
6
1010114 10log 10log S Used
ss RXFFT
F NR SNR R IMPLoss NFN
− × ×= − + − + + +
(44)
Where:
• SNR is 20 dB as tabled in [37], for 64-QAM modulation at a 54 Mbps data rate;
• R is the number of data repetitions, one repetition was considered;
• Fs is 22.4 MHz for a 20 MHz BW signal;
• Nused is the number of data subcarriers used, 1680 [32], [38];
• NFFT should be 2048 [38], but since the transmitted signal only contains data, the used
number was 4096 in order to pass the spectral mask requirements.
• Implementation loss is 5 dB;
• NF is 8 dB.
With all these values replaced in (44) sensitivity can now be calculated,
1022.4 1680114 20 10log 1 10log 5 8 71
4096ssR dBm× = − + − + + + = −
(45)
35
3.1.3 Noise Figure
IEEE 802.11g Requirement: NF should be less than 10 dB.
IEEE 802.16e Requirement: NF should be less than 8 dB.
3.1.4 Adjacent Channel Rejection
IEEE 802.11g Requirement: Adjacent channels at 2.4 GHz are at ± 25 MHz spacing. The
adjacent channel rejection shall be measured by setting the desired signal’s strength 3 dB above
sensitivity. The power difference between the interfering and the desired channel is the adjacent
channel rejection. The corresponding rejection, measured at the receiver input, shall be no less than
-1 dB while keeping the BER requirements.
IEEE 802.16e Requirement: The adjacent channel rejection shall be measured by setting the
desired signal’s strength 3 dB above the sensitivity. The corresponding rejection, measured at the
receiver input, shall be no less than 4 dB while maintaining the BER requirements.
3.1.5 SNR and BER
IEEE 802.11g Requirement: Minimum SNR requirement, at the receiver output, for BER=10-3
is 20 dB (64-QAM at a 54 Mbps data rate). This can be confirmed in [28] [22].
IEEE 802.16e Requirement: Minimum SNR requirement, at the receiver output, for BER=10-6
is 20 dB (64-QAM at a 54 Mbps data rate) as seen in 3.1.2.
3.1.6 Spectrum Mask
IEEE 802.11g Requirement: The spectral density of the transmitted signal shall fall within the
spectral mask as shown in Figure 39.
36
Figure 39 – 802.11g transmitted spectrum mask.
IEEE 802.16e Requirement: The spectral density of the transmitted signal shall fall within the
spectral mask as shown in Figure 40.
Figure 40 – 802.16e transmitted spectrum mask.
Both spectral masks take into account that the transmitted signals have a peak to average
power ratio (PAPR) of 12 dB [39].
3.1.7 EVM
IEEE 802.11g Requirement: EVM has a maximum value of -25 dB for a 64-QAM modulation
at a 54 Mbps data rate [40] [41].
IEEE 802.16e Requirement: EVM has a maximum value of -31 dB for a 64-QAM at a 54
Mbps data rate [41].
37
3.2 Performance Calculations
3.2.1 Noise Figure
As seen in 3.1.5, the 802.11g requires a BER=10-3, which in Figure 41 translates to a Eb/N0 of
approximately 13.46 dB [28]. The 802.16e requires a BER=10-6, translating into an Eb/N0 of
approximately 19 dB. Since the SNR is given by [28],
( ) 0 210 log( / ) 10log logbSNR E N M R= + × (46)
where R is the coding rate ¾ and M the constellation density, 64. Substituting these values in (46), the
minimum theoretical values of SNR are obtained, 18 dB for the 802.11g and 19 dB for the 802.16e. In
practical cases, to take into account additional imperfections, a safe margin up to 3 dB is added to
SNR. It is possible to see in 3.1.5 that both standards take this into consideration.
Figure 41 – BER vs Eb/N0 for 4-QAM, 16 QAM, 64-QAM, and 1024-QAM modulations [42].
With the value of the SNR, the required NF of the system can be calculated. The considered
BW of the signal of both standards is 20 MHz. The sensitivity values were defined in 3.1.2. With these
values replaced in (8) the NF required for the 802.11g is 16 dB, and for the 802.16e is 10 dB. These
calculated values for NF are above the ones defined by the standards, and because of that the
standard’s values will be considered. A receiver should have the lowest NF as possible, otherwise it
won’t be a competitive product [31].
3.2.2 Propagation Loss
The free space losses, in situations in which the distance between the access point (AP) and
the receiver is far greater than the wavelength, can be calculated by [31],
38
[ ]2 24 410log 10logd dfL dB
cπ πλ
= =
(47)
where L is the propagation loss, d is the distance between the AP and the receiver, λ is the
wavelength of the RF signal, f is the signal frequency and c the speed of light in vacuum,
83 10 /m s× . In (47) are not taken into account the antenna gain, absorption and reflection losses, and
other factors.
For the 802.11g at 2.4 GHz, if the receiver is placed 1.5 meter away from the AP transmitting
at its maximum output power, 200 mW (23 dB) [36], the signal will experience 40 dB attenuation, and
at the receiver input power level will be -20 dBm (maximum allowed). At a distance of 250 meters, the
signal will suffer 88 dB attenuation, arriving at the receiver with a -65 dBm power level (sensitivity).
Now let’s consider the 802.16e case at 3.5 GHz, transmitting at its maximum output power of
200 mW (23 dB). This power level was considered equal to the previous case since it is also allowed
by the 802.16e standard [37]. Assuming the receiver is placed 3 meters away from the AP, the signal
will experience 53 dB attenuation, arriving at the receiver with a -30 dBm power level (maximum
allowed). At a distance of 350 meters, the signal would suffer a 94 dB attenuation arriving at the
receiver with a power of -71 dBm (sensitivity).
This is only valid under the conditions referred above.
3.2.3 ADC
The DR of an ADC can be calculated by its maximum SNR [43],
[ ] 6.02 1.76ADC bDR dB n= + (48)
In mobile systems, at least 10 bit ADCs are frequently used [43] to prevent a higher BER due
to the ADC’s quantization noise [44].
The ADC average input power must be -21 dB and -30 dBm for the 802.11g and 802.16e,
respectively, for the signal to be properly decoded. Since OFDM has a PAPR of 12 dB, the ADC
needs to be able to handle a maximum power of -9 dBm. The ADC studied in this thesis has a 10 bit
resolution and a peak to peak voltage of 1 V. Considering that the ADC input power, in watts, is given
by,
21
2in
inin
VPZ
= (49)
the input impedance can be calculated and is equal to 1 kΩ.
The ADC NF can be obtained by,
39
20log 6.02 10log 10log 196.2ADC pp in sNF V n Z f= − − − + (50)
Considering a zero-IF architecture, the ADC has a sampling frequency of 20 MHz, the
minimum possible while still avoiding aliasing. Thus the NF is,
3 620 log1.0 6.02 10 10log10 10log(20 10 ) 196.2 33ADCNF dB= − × − − ⋅ + = (51)
Now considering a low-IF architecture, the same ADC has a sampling frequency of 40 MHz.
For the same reason, a NF of approximately 30 dB is obtained.
3.3 Transmitter The transmitter blocks for the 802.11g and 802.16e are shown in Figure 42.
Figure 42 – 802.11g/802.16e Transmitter.
In this thesis it was only considered the transmission of the data (payload) at the maximum
speed, 54 Mbps. Therefore, the data bits are divided in 6-bit groups and mapped to a 64-QAM
constellation using Gray codification (Figure 43). The transmitters built in Ptolemy are shown in the
Appendix in Figure 136 and Figure 137.
Figure 43 – 64-QAM constellation using Gray code.
Before the IFFT, the data to be transmitted is loaded into the IFFT buffer. In Figure 44 it is
shown that an OFDM symbol is made of two types of subcarriers: The “Used subcarriers” transport the
transmitted data; the “Null subcarriers”, that consist of DC subcarrier and guard subcarriers, aren’t
used for data transmission [45].
40
Figure 44 – OFDM subcarriers [45].
The purpose of using guard subcarriers is to decay the signal spectrum edges creating the
Fast Fourier Transform (FFT) rectangular shape [45]. This helps to reduce aliasing [46] and it
becomes easier to recover the signal at the receiver side [47].
In the transmitter the aliasing is produced by the DAC sampling process at a rate of 1/Tsymbol
and is right next to the OFDM signal. This happens because the values at the output of the IFFT need
to be sampled by the DAC before the OFDM signal arrives to its final shape to be transmitted. The
aliasing is shifted by padding certain positions of the IFFT input sequence with zeros. This is illustrated
in Figure 45.
Figure 45 – Zero Padding used to shift aliases [46].
The mapping of the subcarriers at the input of the IFFT buffer is show in Figure 46. The IFFT
parameters are specified in Table 5.
Figure 46 – OFDM subcarriers mapping [45].
41
Then, the IFFT transforms the signal from the frequency domain to the time domain. After that,
a guard interval is inserted at the beginning of each symbol. Its purpose is to avoid ISI and inter
channel interference (ICI). To eliminate ISI the guard interval has a duration longer than the multipath
channel maximum delay. Furthermore, to remove ICI the guard time is cyclically extended within this
guard time, as shown in Figure 47. This cyclical extension is also known as cyclic prefix (CP) [46].
Figure 47 – OFDM symbol time structure [48].
Now, that the time domain OFDM symbol is built, it will then modulate a RF carrier with a
symbol rate specified in Table 5. The symbol rate was calculated based on the equations given by [2],
since they provided the symbol rate compliant with the standards. For the 802.11g the symbol rate
was obtained by the equation,
( 6)802.11 2 order
gSymbolrate bandwidth −= × (52)
where the BW of the signal is 20 MHz and the order of the IFFT is 8 satisfying 2order Carriers= . For
the 802.16e the symbol rate was calculated by,
802.16 2eSymbolrate bandwidth oversamplingrate= × × (53)
where the BW of the signal is 20 MHz, the corresponding oversampling rate is 28/25 and the IFFT
order is 12.
The simulator modulator component reference voltage parameter (VREF) was calibrated
accordingly to the desired output power.
Table 5 – Transmitter parameters.
802.11g 802.16e
IFFT size 256 4096
Used subcarriers 52 1680
Null subcarriers DC ; 27-229 DC ; 841-3255
Guard interval 13
IFFTSize× 18
IFFTSize×
Symbol rate [MHz] 80 44.8
Carrier frequency [MHz] 2450 3500
In Figure 48 and Figure 49, are show the transmitted spectrums for both standards with an
output power of 23 dBm and carrying data only.
42
Figure 48 – 802.11g OFDM transmitted spectrum.
Figure 49 – 802.16e OFDM transmitted spectrum.
It is possible to observe that both spectrums fulfill the spectral masks requirements (marked in
red).
43
3.4 Receiver Initially, a state of the art research was made to find the 802.11g and 802.16 receivers’ typical
requirements in terms of block level specifications:
Table 6 – Front-end specifications.
Ref. Receiver/ Standard
BPF LNA Mixer LO IL
[dB] Gain [dB]
NF [dB]
IIP3 [dBm]
Gain [dB]
NF [dB]
IIP3 [dBm] Phase Noise
[18] Zero-IF 802.11a 2 25/10 −95dBc/100 kHz
[49] Zero-IF 802.11a/g 16 3 -1 10 10 15
[50] Zero-IF 802.11a/b/g 2 18 3 3 12 7 -102 dBc/1 MHz
[19] Zero-IF 802.11a/b/g 20/10 2.5 4 12 11 -115 dBc/1 MHz
[22] Zero-IF 802.11g/802.16e 1.5 18/0 -7 10 8.6 -110dBc/1 MHz.
[23] Zero-IF 802.11g/802.16e -100dBc/100kHz
-95 dBc/100kHz
[21] Zero-IF 802.16e
[39] Zero-IF 802.16e 20/10 2.5 -6 10 10 5
[51] Zero-IF 802.16e 25
[52] Zero-IF 802.16e -90 dBc/100 kHz
-120 dBc/1MHz
[53] Low-IF 802.11a/g 20 1.6 16
[54] Zero-IF/ Low-IF 802.16e
17/5 2.5 2 6 15
[55] Low-IF 802.16e 25/10 -15/-4 10 11 2.7 -110 dBc/1MHz
Table 7 – IF and BB specifications.
Ref. LPF PGA ADC
Order Type Freq. [MHz]
IL [dB]
Gain [dB]
NF [dB]
IIP3 [dBm]
N bits
Freq. [MHz]
DR [dB]
[49] 35 28 5 [50] 4th Butterworth 10 [19] 4th Chebyschev 3-20 57 10 40 [22] 3rd Butterworth 3 50/20 12 55 [23] 5th Butterworth 1.9-11 40 [21] 3rd Butterworth [39] 40/5 16 16 [51] 5th Butterworth 66 [53] Polyphase 40 [54] 2nd Butterworth 20/0 [55] 10 [56] 5th Butterworth 2.5-20 65
44
Table 8 – Receiver global specifications.
Ref. Receiver/ Standard
Receiver Gain [dB]
NF [dB]
IP1dB [dB]
IIP3 [dBm]
[18] Zero-IF 802.11a 66 <10 -/-27 >-19
[50] Zero-IF 802.11a/b/g 7.5 -26/-10
[19] Zero-IF 802.11g 3.5 -/8
[23] Zero-IF 802.11g/802.16e 3.4/4
[21] Zero-IF 802.16e 40 5.8 -26 -14
[39] Zero-IF 802.16e 68/23 6.5 -16/-7.8
[51] Zero-IF 802.16e 3.3
[52] Zero-IF 802.16e 80 4.5 -6
[53] Low-IF 802.11a/g 33.4 4.1 -26 -16.6
[54] Zero-IF/Low-IF 802.16e 48 7 -16/-8
[55] Low-IF 802.16e 37 3.7
[56] Low-IF 802.16e 93/1 4.6
The block specifications were summarized in the tables below.
Table 9 – LNA and PGA specifications.
Gain [dB] NF [dB] IIP3 [dBm]
LNA High gain mode [16 , 25]
[1.6 ; 3] [-15 ; 4] Low gain mode [0 , 10]
PGA High gain mode [20 , 66]
[16 ; 28] [5 ; 16] Low gain mode [0 , 20]
Table 10 – Mixer specifications.
Gain [dB] NF [dB] IIP3 [dBm]
[2 ; 16] [6 , 11] [3 ; 15]
Table 11 – LO specifications.
Power [dBc] Offset
[-104 , -90] 100 kHz
[-125 , -102] 1 MHz
45
Table 12 – IF and BB filters specifications.
Type Order BW [MHz] Insertion Loss [dB]
Chebyshev 4th [3 , 20] [3 , 10]
Butterworth 2nd to 5th [2.5 , 20]
Table 13 – ADC specifications.
Resolution Frequency DR
10 bit 40 MHz 55 dB
Since the heterodyne architecture is no longer an option for mobile devices, due to the
reasons specified in 1.2.4, the following simulations were only focused on the Zero-IF and Low-IF
architectures.
Two of the biggest concerns when designing a receiver are the sensitivity and the gain. The
main goal for it, is to achieve a low NF and a high input referred third order intercept point (IIP3). Since
these parameters are dependent, it is crucial to ensure a proper distribution of the gain over the
receiver chain. The performance of a receiver is mainly affected by its sensitivity, intermodulation
products and adjacent and alternate channel rejection [43]. This will be discussed in the following
sections.
Based on the research above, the LNA, Mixer and PGA linearity was defined by the following
values in Table 14.
Table 14 – LNA, Mixer and PGA requirements.
LNA IIP3
Mixer IIP3
PGA IIP3
-6 dBm 15 dBm 14 dBm
But, both simulation software demand output referred third order intercept point (OIP3) LNA
and PGA values, and the translations was made taking into account their maximum gain, as show in
Table 15.
Table 15 – LNA, Mixer and PGA simulated requirements.
LNA OIP3
Mixer IIP3
PGA OIP3
12 dBm 15 dBm 32 dBm
The receiver chain is shown in Figure 50. The demodulator does, exactly, the inverse
operation of the transmitter. The guard interval is removed from the digital signal and a FFT is
performed, converting the signal to the frequency domain. Next, the null subcarriers are removed so
that the bitstream can be properly extracted, from the complex QAM signal, by the decoder.
46
Figure 50 – 802.11g/802.16e receiver chain.
The demodulator built in Ptolemy is shown in the Appendix in Figure 138.
47
48
4 System Design
In this chapter two individual receivers will be designed for the 802.11g and 802.16e,
respectively. For each receiver, the zero-IF and low-IF architectures will be considered, in a total of
four receivers. The Genesys [1] and Ptolemy [2] softwares from Agilent Technologies were chosen to
perform the system level simulations.
Genesys provides graphical diagnosis of impairments such as mismatches and frequency
spurious mixing products that are ignored by spreadsheets. It also allows system designs to be
exported to Agilent ADS for further development at the circuit and DSP levels.
Ptolemy is oriented for electronic systems with special emphasis on communication
applications. One of its main advantages is the ability to make mixed domains simulations. This
means, for instance, that a system can contain several numerical or behavioral domain blocks and
also blocks that are described by electrical circuits. This capability is important to help the design of
transmitters or receivers architectures to cope with modern specifications. Ptolemy has the capability
of communicating with the equipment of Agilent, programming signal generators and acquiring signals
from oscilloscopes and spectrum analyzers.
4.1 Zero-IF Architecture The RF signal is band selected by the BPF and amplified by the LNA. Then, it is down
converted to BB by two quadrature mixers. In BB the signal is filtered by two LPFs, removing the high
frequency components from the mixer. The PGAs provide the ADC a signal with constant power for
proper digitalization. It is assumed that both the LNA and the PGAs have built-in AGC inputs. The
Zero-IF architecture is the same for both standards, as shown in Figure 51, only the block
specifications differ.
Figure 51 – 802.11g/802.16e Zero-IF Architecture.
The 802.11g and 802.16e don’t specify an intermodulation requirement. Although, it is
possible to express the receiver linearity in terms of the input 1 dB compression point (IP1dB) [50].
The LNA is one of most important blocks, since it needs to have a large DR to cover the input
signal. To prevent itself and following blocks saturation, the gain of LNA and PGA must be reduced for
49
strong input signals. Therefore, the receiver operates in two distinct modes [39]: low gain and high
gain modes.
4.1.1 Low Gain Mode
In the low gain mode, the receiver input signal is already strong and doesn’t require a
significant amount of amplification. The 802.11g and the 802.16e use OFDM modulation, hence a
back off of 10~12 dB, resulting from the PAPR of the OFDM signal varying envelope, must be taken
into account between the maximum average input signal and the IP1dB [39].
So, for the 802.11g, the maximum average received signal is -20 dBm, translating into IP1dB
of -10 dBm due to a 10 dB backoff. Usually the IIP3 is larger than the IP1dB by 9~10 dB, meaning an
IIP3 of 0 dBm is required [50].
In the 802.16e, the maximum received signal is -30 dBm. Then, an IP1dB of -18 dBm is
required, considering a backoff of 12 dB. The required IIP3 is -8 dBm [39]. The low gain mode
requirements are shown in Table 16.
Table 16 – Low gain mode requirements.
802.11g 802.16e
IP1dB [dBm] -10 -18
IIP3 [dBm] 0 -8
4.1.1.1 802.11g
The BPF has an insertion loss of 2 dB with a passband of 100 MHz centered at 2450 MHz
[57]. The 4th order LPF has an insertion loss of 5 dB [43] and a cutoff frequency of 12 MHz. The 10 bit
ADC has an input impedance of 1 kOhm, a VREF of 0.95 V and a sampling frequency of 21 MHz. For
an efficient sampling and proper data recovery, the ADC input power was set to aprox. -21 dBm, as
explained in 3.2.3. The LO phase noise was defined to be 95 dBc/Hz@100 kHz [18] and 102
dBc/Hz@1 MHz [50].
Based on the specification ranges summarized in 3.4, the block level specifications were
obtained and adjusted by simulation for the low gain mode as show in Figure 52.
50
Figure 52 – 802.11g zero-IF low gain mode specifications.
It was considered that the signal is received, in additive white Gaussian noise (AWGN)
conditions, at a 54 Mbps data rate with a frequency of 2450 MHz at -20 dBm. The 802.11g zero-IF
receiver built in Ptolemy is shown in the Appendix in Figure 139. The demodulated signal constellation
is shown in Figure 53.
Figure 53 – 802.11g zero-IF low gain mode constellation.
The received bitstream is perfectly synchronized with the transmitted one. In Figure 54, the
first 300 bits show that both bitstreams are synchronized from the beginning.
Figure 54 – 802.11g zero-IF low gain mode bitstreams.
51
The BER was simulated over the 1000 received bits, and it passed the standard requirement.
This is show in Table 17, imported from Ptolemy.
Table 17 – 802.11g zero-IF low gain mode BER.
The following simulations were made with Genesys. The Genesys 802.11g zero-IF receiver is
depicted in the Appendix in Figure 132. A linear amplifier was used to simulate the ADC NF. The
obtained IP1dB was -7.1 dBm. In Figure 55 is shown its evolution among the receiver blocks.
Figure 55 – 802.11g zero-IF low gain mode IP1dB.
An input power sweep was performed in order to determine the output power saturation. In
Figure 56, it can be observed that none of the receiver blocks compresses.
Figure 56 – 802.11g zero-IF low gain mode input power sweep.
A two-tone test was performed to obtain IIP3 (Figure 57). Two strong tones with -10 dBm were
applied at the receiver input with offsets of 20 MHz and 40 MHz from the carrier frequency. The lower
52
intermod product is downconverted to BB falling inside the IF BW, in this way it is possible to simulate
its interference.
Figure 57 – 802.11g zero-IF two-tone test.
The obtained IIP3 was 3.4 dBm. In Figure 58, is shown its evolution among the receiver
blocks.
Figure 58 – 802.11g zero-IF low gain mode IIP3.
The obtained total gain and NF were -2.8 dB and 32.8 dB, respectively. Their evolution among
the receiver blocks is shown in Figure 59.
Figure 59 – 802.11g zero-IF low gain mode gain and NF.
53
4.1.1.2 802.16e
The BPF has an insertion loss of 2 dB with a passband of 200 MHz centered at 3500 MHz
[58]. The LPF has an insertion loss of 5 dB and a corner frequency of 10 MHz. The 10 bit ADC has an
input impedance of 1 kOhm, a VREF of 0.95 V and a sampling frequency of 20 MHz. For an efficient
sampling and proper data recovery, the ADC input power was set to aprox. -30 dBm, as explained in
3.2.3. The LO phase noise was defined to be 95 dBc/Hz@100 kHz [23] and 110 dBc/Hz@1 MHz [55].
Based on the specification ranges summarized in 3.4, is possible to verify that the 802.11g
and 802.16e specifications are quite similar. Thus, the block level specifications were obtained and
adjusted by simulation for the low gain mode as show in Figure 60.
Figure 60 – 802.16e zero-IF low gain mode specifications.
It was considered that the signal is received, in AWGN conditions, at a 54 Mbps data rate with
a frequency of 3500 MHz at -30 dBm. The 802.16e zero-IF receiver built in Ptolemy is shown in the
Appendix in Figure 140. The demodulated signal constellation is shown in Figure 61.
Figure 61 – 802.16e zero-IF low gain mode constellation.
The received bitstream is perfectly synchronized with the transmitted one. In Figure 62, the
first 300 bits show that both bitstreams are synchronized from the beginning.
54
Figure 62 – 802.16e zero-IF low gain mode bitstreams.
The BER was simulated over the 1000 received bits, and it passed the standard requirement.
This is show in Table 18, imported from Ptolemy.
Table 18 – 802.16e zero-IF low gain mode BER.
The following simulations were made with Genesys. The Genesys 802.11g zero-IF receiver is
depicted in the Appendix in Figure 134. A linear amplifier was used to simulate the ADC NF. The
obtained IP1dB was -7.1 dBm. In Figure 63 is shown its evolution among the receiver blocks.
Figure 63 – 802.16e zero-IF low gain mode IP1dB.
An input power sweep was performed in order to determine the output power saturation. In
Figure 64, it can be observed that none of the receiver blocks compresses.
55
Figure 64 – 802.16e zero-IF low gain mode input power sweep.
A two-tone test was performed to obtain IIP3 (Figure 65). Two strong tones with -20 dBm were
applied at the receiver input with offsets of 20 MHz and 40 MHz from the carrier frequency. The lower
intermod product is downconverted to BB falling inside the IF BW, in this way it is possible to simulate
its interference.
Figure 65 – 802.16e zero-IF two-tone test.
The obtained IIP3 was 2.8 dBm. In Figure 66, it is shown its evolution among the receiver
blocks.
Figure 66 – 802.16e zero-IF low gain mode IIP3.
The obtained total gain and NF were -2.1 dB and 32.1 dB, respectively. Their evolution among
the receiver blocks is shown in Figure 67.
56
Figure 67 – 802.16e zero-IF high gain mode Gain and NF.
4.1.2 High Gain Mode
In the high gain mode, the received signal is weak and needs a significant amount of
amplification. In the 802.11g and in the 802.16e, the IP1dB in the high gain mode is set to -26 dBm.
Thus, an IIP3 of -16 dBm is required for both standards [50] [39], as show in Table 19.
Table 19 – High gain mode requirements.
802.11g 802.16e
IP1dB [dBm] -26 -26
IIP3 [dBm] -16 -16
4.1.2.1 802.11g
The block level specifications are the same as in low gain mode (4.1.1.1), with a few
exceptions. The LNA and PGA gains were changed to 18 dB, as shown in Figure 68.
Figure 68 – 802.11g zero-IF high gain mode specifications.
It was considered that the signal is received, in AWGN conditions, at a 54 Mbps data rate with
a frequency of 2450 MHz at -65 dBm.
57
The demodulated signal constellation is shown in Figure 69.
Figure 69 – 802.11g zero-IF high gain mode constellation.
The signal was recovered with a BER lower than the required by the standard as in low gain
mode (4.1.1.1). The obtained IP1dB was -20.6 dBm. In Figure 70 it is shown its evolution among the
receiver blocks.
Figure 70 – 802.11g zero-IF high gain mode IP1dB.
An input power sweep was performed in order to determine the output power saturation. In
Figure 71, it can be observed that none of the receiver blocks compresses.
Figure 71 – 802.11g zero-IF high gain mode input power sweep.
58
A two-tone test was performed to obtain IIP3 with the same configuration as in low gain mode
(4.1.1.1), with the two tones at -20 dBm. The obtained IIP3 was -3.6 dBm. In Figure 72, it is shown its
evolution among the receiver blocks.
Figure 72 – 802.11g zero-IF high gain mode IIP3.
The total gain and NF were 42.2 dB and 5.4 dB, respectively. Their evolution among the
receiver blocks is shown in Figure 73.
Figure 73 – 802.11g zero-IF high gain mode Gain and NF.
4.1.2.2 802.16e
The block level specifications are the same as in low gain mode (4.1.1.2), with a few
exceptions. The LNA and PGA gains were changed to 18 dB, as shown in Figure 74.
Figure 74 – 802.16e zero-IF high gain mode specifications.
59
It was considered that the signal is received, in AWGN conditions, at a 54 Mbps data rate with
a frequency of 3500 MHz at -71 dBm. The demodulated signal constellation is shown in Figure 75.
Figure 75 – 802.16e zero-IF high gain mode constellation.
The signal was recovered with a BER lower than the required by the standard as in low gain
mode (4.1.1.2). The obtained IP1dB was -18.5 dBm. In Figure 76, it is shown its evolution among the
receiver blocks.
Figure 76 – 802.16e zero-IF high gain mode IP1dB.
An input power sweep was performed in order to determine the output power saturation. In
Figure 77, it can be observed that none of the receiver blocks compresses.
60
Figure 77 – 802.16e zero-IF high gain mode input power sweep.
A two-tone test was performed to obtain IIP3 with the same configuration as in low gain mode
(4.1.1.2), with the two tones at -30 dBm. The obtained IIP3 was -4 dBm. In Figure 78, it is shown its
evolution among the receiver blocks.
Figure 78 – 802.16e zero-IF high gain mode IIP3.
The obtained total gain and NF were 38.9 dB and 5.05 dB, respectively. Their evolution
among the receiver blocks is shown in Figure 79.
Figure 79 – 802.16e zero-IF high gain mode Gain and NF.
61
4.1.3 Adjacent Channel Rejection
4.1.3.1 802.11g
The adjacent channel rejection was simulated as explained in 3.1.4 and show in Figure 80.
Two completely different modulating bitstreams were used for the desired signal and interferer. The
average power of the interfering signal was initially set to -63 dBm to calibrate the receiver gain for
proper demodulation. Its average power was then raised until the BER requirement was no longer
fulfilled.
Figure 80 – 802.11g zero-IF adjacent channel power.
The receiver was working in high gain mode and the PGA was operating with a 15 dB gain.
The average power of the interfering could be raised up to -39 dB. This results in a 23 dB adjacent
channel rejection, as shown in Figure 81.
Figure 81 – 802.11g zero-IF adjacent channel and desired signal.
By looking at the constellation in Figure 82, it can be seen that the signal is at the limit of being
properly decoded.
62
Figure 82 – 802.11g zero-IF adjacent channel rejection test constellation.
4.1.3.2 802.16e
The adjacent channel rejection was simulated as explained in 3.1.4 and show in Figure 83.
Two completely different modulating bitstreams were used for the desired signal and interferer. The
average power of the interfering signal was initially set to -64 dBm to calibrate the receiver gain for
proper demodulation. Its average power was then raised until the BER requirement was no longer
fulfilled.
Figure 83 - 802.16e zero-IF adjacent channel power.
The receiver was working in high gain mode and the PGA was operating with 12 dB gain. The
maximum average power of the interfering could be raised up to -62 dB. This results in a 6 dB
adjacent channel rejection, as shown in Figure 84.
Figure 84 – 802.16e zero-IF adjacent channel and desired signal.
63
By looking at the constellation in Figure 85, it can be seen that the signal is at the limit of being
properly decoded.
Figure 85 – 802.16e zero-IF adjacent channel rejection test constellation.
4.1.4 EVM
4.1.4.1 802.11g
For 802.11g EVM simulation the Ptolemy component presented in Figure 86. This model was
used since it accepts custom OFDM signals, like the ones in this work.
Figure 86 – Model EVM_WithRef [2].
The parameters symbol rate and the number of samples per frame are the same as those defined
in the transmitter in 3.3. The symbol at which the data collection for the EVM measurement starts is
the first, since the two bitstreams are synchronized from the beginning, as show in Figure 54. The
considered burst size is 1000, the number of bits transmitted. The SymDelayBound parameter is set to
-1 since no synchronization is needed, the two bitstreams are already synchronized. Since the
standards specify that the EVM measurement should be performed over multiple bursts and the result
should be averaged, 10 bursts were considered.
The signal was transmitted with an output power of 0 dBm at a 54 Mbps data rate. The two
receiver modes were studied. In high gain mode, the obtained EVM was approximately -30 dB as
show in Table 20.
64
Table 20 – 802.11g zero-IF high gain mode EVM.
And the corresponding constellation compared to the ideal one is shown in Figure 87.
Figure 87 – 802.11g zero-IF high gain mode EVM constellation.
In low gain mode, the obtained EVM was approximately -33.9 dB as shown in Table 21.
Table 21 – 802.11g zero-IF low gain mode EVM.
The corresponding constellation, compared to the ideal one is shown in Figure 88.
Figure 88 – 802.11g zero-IF low gain mode EVM constellation.
65
4.1.4.2 802.16e
For the 802.11g EVM simulation it was used the same element shown in Figure 86, for the
same reasons. The signal was transmitted with an output power of 0 dBm at a 54 Mbps data rate. The
two receiver modes were studied. In high gain mode, the obtained EVM was approximately -23.4 dB
as show in Table 22 .
Table 22 – 802.16e zero-IF high gain mode EVM.
And the corresponding constellation compared to the ideal one is shown in Figure 89.
Figure 89 – 802.16e zero-IF high gain mode EVM constellation.
In low gain mode, the obtained EVM was approximately -26.2 dB as shown in Table 23.
Table 23 – 802.16e zero-IF low gain mode EVM.
The corresponding constellation, compared to the ideal one is shown in Figure 90.
Figure 90 – Zero-IF 802.16e low gain mode EVM constellation.
66
4.1.5 Imbalance and Linearity
Phase and gain imbalances were added to see their impact on the BER requirement. The
OIP3 values of the LNA and PGA were lowered to see the minimum value that still manages to pass
the BER requirements.
4.1.5.1 802.11g
Phase and gain imbalances were modified individually. The imbalance ranges, for which the
BER is fulfilled are shown in Table 24.
Table 24 – 802.11g zero-IF imbalances.
Gain [dB] Phase [º]
Low gain mode [-0.7 , 1.5] [-4.6 , 5.2]
High gain mode [-0.6 , 1.1] [-3.6 , 5.4]
The LNA and PGA OIP3 values were lowered individually in low gain mode. Their minimum
values are shown in Table 25.
Table 25 – 802.11g zero-IF linearity.
LNA PGA
OIP3 [dBm] 3 -12
4.1.5.2 802.16e
Phase and gain imbalances were modified individually. The imbalance ranges, for which the
BER is fulfilled are shown in Table 26.
Table 26 – 802.16e zero-IF imbalances.
Gain [dB] Phase [º]
Low gain mode [-0.7 , 0.7] [-4.8 , 4.8]
High gain mode [-0.5 , 0.3] [-1.2 , 0.7]
The LNA and PGA OIP3 values were lowered individually in low gain mode. Their minimum
values are shown in Table 27.
Table 27 – 802.16e zero-IF linearity.
LNA PGA
OIP3 [dBm] -7 -22
67
4.1.6 Results Discussion
The 802.11g required a LPF cutoff frequency of 12 MHz and an ADC sampling frequency of
21 MHz, due to its wider spectrum. For the 802.16e, which has a thinner spectrum, a LPF cutoff
frequency of 10 MHz and an ADC sampling frequency of 20 MHz were enough. The simulated ADC
VREF value was 0.95 V instead of 1 V. This was dictated by the 802.16e requirements. Anticipating a
dual band project, this value was used for both standards.
It was shown that both IP1dB and IIP3 values are mainly determined by the LNA. In Table 28,
it is shown that the linearity and sensitivity requirements of the 802.11g were fulfilled. Even the
obtained EVM at high gain mode fulfills its requirement.
The NF used to validate the standard requirement is always the one obtained in high gain
mode, because in this mode signal quality can be affected and easily degraded by noise. The NF
obtained in high gain mode is lower than the required.
It has a quite large adjacent channel rejection, as desirable. And in every simulation that was
made, the BER requirement is achieved.
Table 28 – 802.11g Zero-IF results.
Obtained Required
IP1dB low gain mode -7.1 dBm ≥ -10 dBm
high gain mode -20.6 dBm ≥ -26 dBm
IIP3 low gain mode 3.4 dBm ≥ 0 dBm
high gain mode -4 dBm ≥ -16 dBm
EVM low gain mode -33.9 dB
≤ -25 dB high gain mode -30 dB
NF 5.4 dB ≤ 10 dB
Adjacent Channel Rejection 23 dB ≥ -1 dB
BER 0.000 ≤ 10-3
In EVM simulations it is possible to observe that the spreading of the constellation points is, in
part, due to LO phase noise [31]. Most noticeable in high gain mode, since the input signal is weak
with low power making it more vulnerable to noise.
In Table 29, it is shown that the linearity and sensitivity requirements of the 802.16e were also
fulfilled, with a considerable margin. The NF in high gain mode is lower than the required. The
adjacent channel rejection has a 2 dB margin to the minimum required. Only the obtained EVM values
didn’t fulfill the standard requirement.
68
Table 29 – 802.16e Zero-IF results.
Obtained Required
IP1dB low gain mode -7.1 dBm ≥ -18 dBm
high gain mode -18.5 dBm ≥ -26 dBm
IIP3 low gain mode 2.8 dBm ≥ -8 dBm
high gain mode -6 dBm ≥ -16 dBm
EVM low gain mode -26.2 dB
≤ -31 dB high gain mode -23.4 dB
NF 5 dB ≤ 8 dB
Adjacent Channel Rejection 6 dB ≥ 4 dB
BER 0.000 ≤ 10-6
4.2 Low-IF Architecture The RF signal is band selected by the BPF and amplified by the LNA. Then, it is down
converted to an IF of 10 MHz by two quadrature mixers. IF was chosen to be equal to half of the RF
signal BW, in order to make the signal left edge closer to zero. Thus, the ADC sampling frequency can
be as low as possible. At IF the signal is filtered by two LPFs removing the high frequency
components from the mixer. The PGAs provide the ADC a signal with constant power for proper
digitalization. It is assumed that both the LNA and the PGAs have built-in AGC inputs. The low-IF
architecture is the same for both standards, as shown in Figure 91, only the block specifications differ.
Figure 91 – 802.11g/802.16e low-IF Architecture.
The low-IF follows exactly the same considerations made for the zero-IF, and therefore the
two gain modes are also studied for the low-IF architecture. By the research made in 3.4, it is possible
to conclude that both architectures, zero-IF and low-IF, have almost the same block level
specifications. Thus, the zero-IF requirements and block level specifications were imported and
adapted for the low-IF architecture.
69
4.2.1 Low Gain Mode
In this mode the receiver input signal is already strong and doesn’t require a significant
amount of amplification. The IP1dB and IIP3 requirements are the same as Zero-IF low gain mode
(4.1.1).
4.2.1.1 802.11g
The BPF has an insertion loss of 2 dB with a passband of 100 MHz centered at 2450 MHz
[57]. The 4th order LPF has an insertion loss of 5 dB [43] and a cutoff frequency of 20 MHz. The 10 bit
ADC has an input impedance of 1 kOhm, a VREF of 0.95 V and a sampling frequency of 40 MHz. For
an efficient sampling and proper data recovery, the ADC input power was set to aprox. -21 dBm, as
explained in 3.2.3. The LO phase noise was defined to be 95 dBc/Hz@100 kHz [18] and 102
dBc/Hz@1 MHz [50]. In Figure 92, are shown the block level specifications used for low gain mode.
Figure 92 – 802.11g low-IF low gain mode specifications.
It was considered that the signal is received, in AWGN conditions, at a 54 Mbps data rate with
a frequency of 2450 MHz at -20 dBm. The 802.11g low-IF receiver built in Ptolemy is shown in the
Appendix in Figure 141. The demodulated signal constellation is shown in Figure 93.
Figure 93 – 802.11g low-IF low gain mode constellation.
70
The received bitstream is perfectly synchronized with the transmitted one. In Figure 94, the
first 300 bits show that both bitstreams are synchronized from the beginning.
Figure 94 – 802.11g low-IF low gain mode bitstreams.
The BER was simulated over the 1000 received bits, and it passed the standard requirement.
This is show in Table 30, imported from Ptolemy.
Table 30 – 802.11g low-IF low gain mode BER.
The following simulations were made with Genesys. The Genesys 802.11g low-IF receiver is
depicted in the Appendix in Figure 133. A linear amplifier was used to simulate the ADC NF. The
obtained IP1dB was -7.1 dBm. In Figure 95 is shown its evolution among the receiver blocks.
Figure 95 – 802.11g low-IF low gain mode IP1dB.
An input power sweep was performed in order to determine the output power saturation. In
Figure 96, it can be observed that none of the receiver blocks compresses.
71
Figure 96 – 802.11g low-IF low gain mode input power sweep.
A two-tone test was performed to obtain IIP3 (Figure 97). Two strong tones with -10 dBm were
applied at the receiver input with offsets of 20 MHz and 40 MHz from the carrier frequency. The lower
intermod product is downconverted to BB falling inside the IF BW, in this way it is possible to simulate
its interference.
Figure 97 – 802.11g low-if two-tone test.
The obtained IIP3 was 1.5 dBm. In Figure 98, is shown its evolution among the receiver
blocks.
Figure 98 – 802.11g low-IF low gain mode IIP3.
72
The obtained total gain and NF were -3 dB and 33 dB, respectively. Their evolution among the
receiver blocks is shown in Figure 99.
Figure 99 – 802.11g low-IF low gain mode Gain and NF.
4.2.1.2 802.16e
The BPF has an insertion loss of 2 dB with a passband of 200 MHz centered at 3500 MHz
[58]. The LPF has an insertion loss of 5 dB and a corner frequency of 20 MHz. The 10 bit ADC has an
input impedance of 1 kOhm, a VREF of 0.95 V and a sampling frequency of 40 MHz. For an efficient
sampling and proper data recovery, the ADC input power was set to aprox. -30 dBm, as explained in
3.2.3. The LO phase noise was defined to be 95 dBc/Hz@100 kHz [23] and 110 dBc/Hz@1 MHz [55].
In Figure 100, are shown the block level specifications used for the low gain mode.
Figure 100 – 802.16e low-IF low gain mode specifications.
It was considered that the signal is received, in AWGN conditions, at a 54 Mbps data rate with
a frequency of 3500 MHz at -30 dBm. The 802.16e low-IF receiver built in Ptolemy is shown in the
Appendix in Figure 142. The demodulated signal constellation is shown in Figure 101.
73
Figure 101 – 802.16e low-IF low gain mode constellation.
The received bitstream is perfectly synchronized with the transmitted one. In Figure 102, the
first 300 bits show that both bitstreams are synchronized from the beginning.
Figure 102 – 802.16e low-IF low gain mode bitstreams.
The BER was simulated over the 1000 received bits, and it passed the standard requirement.
This is show in Figure 31, imported from Ptolemy.
Table 31 – 802.16e low-IF low gain mode BER.
The following simulations were made with Genesys. The Genesys 802.16e low-IF receiver is
depicted in the Appendix in Figure 135. A linear amplifier was used to simulate the ADC NF. The
obtained IP1dB was -7.1 dBm. In Figure 103 is shown its evolution among the receiver blocks.
74
Figure 103 – 802.16e low-IF low gain mode IP1dB.
An input power sweep was performed in order to determine the output power saturation. In
Figure 104, it can be observed that none of the receiver blocks compresses.
Figure 104 – 802.16e low-IF low gain mode input power sweep.
A two-tone test was performed to obtain IIP3 (Figure 105). Two strong tones with -20 dBm
were applied at the receiver input with offsets of 20 MHz and 40 MHz from the carrier frequency. The
lower intermod product is downconverted to BB falling inside the IF BW, in this way it is possible to
simulate its interference.
Figure 105 – 802.16e low-IF two-tone test.
The obtained IIP3 was 0.8 dBm. In Figure 106, it is shown its evolution among the receiver
blocks.
75
Figure 106 – 802.16e low-IF low gain mode IIP3.
The obtained total gain and NF were -2 dB and 32 dB, respectively. Their evolution among the
receiver blocks is shown in Figure 107.
Figure 107 – 802.16e low-IF high gain mode Gain and NF.
4.2.2 High Gain Mode
In the high gain mode, the received signal is weak and needs a significant amount of
amplification. The IP1dB and IIP3 requirements for this mode are the same as in zero-IF high gain
mode (4.1.2).
4.2.2.1 802.11g
The block level specifications are the same as in low gain mode (4.2.1.1), with a few
exceptions. The LNA and PGA gains were changed to 18 dB, as shown in Figure 108.
76
Figure 108 – 802.11g low-IF high gain mode specifications.
It was considered that the signal is received, in AWGN conditions, at a 54 Mbps data rate with
a frequency of 2450 MHz at -65 dBm. The demodulated signal constellation is shown in Figure 109.
Figure 109 – 802.11g low-IF high gain mode constellation.
The signal was recovered with a BER lower than the required by the standard as in low gain
mode (4.2.1.1). The obtained IP1dB was -20.5 dBm. In Figure 110, it is shown its evolution among the
receiver blocks.
Figure 110 – 802.11g low-IF high gain mode IP1dB.
An input power sweep was performed in order to determine the output power saturation. In
Figure 111, it can be observed that none of the receiver blocks compresses.
77
Figure 111 – 802.11g low-IF high gain mode IP1dB.
A two-tone test was performed to obtain IIP3 with the same configuration as in low gain mode
(4.2.1.1) with the two tones at -20 dBm. The obtained IIP3 was -5.3 dBm. In Figure 112, it is shown its
evolution among the receiver blocks.
Figure 112 – Low-IF 802.11g high gain mode IIP3.
The obtained total gain and NF were 42 dB and 8.3 dB, respectively. Their evolution among
the receiver blocks is shown in Figure 113.
Figure 113 – 802.11g low-IF high gain mode Gain and NF.
78
4.2.2.2 802.16e
The block level specifications are the same as in low gain mode 4.2.1.2, with a few exceptions.
The LNA and PGA gains were changed to 18 dB and 15 dB, respectively, as shown in Figure 114.
Figure 114 – 802.16e low-IF high gain mode specifications.
It was considered that the signal is received, in AWGN conditions, at a 54 Mbps data rate with
a frequency of 3500 MHz at -71 dBm. The demodulated signal constellation is shown in Figure 115.
Figure 115 – 802.16e low-IF high gain mode constellation.
The signal was recovered with a BER lower than the required by the standard as in low gain
mode (4.2.1.2). The obtained IP1dB was -18.6 dBm. In Figure 116, it is shown its evolution among the
receiver blocks.
79
Figure 116 – 802.16e low-IF high gain mode IP1dB.
An input power sweep was performed in order to determine the output power saturation. In
Figure 117, it can be observed that none of the receiver blocks compresses.
Figure 117 – 802.16e low-IF high gain mode input power sweep.
A two-tone test was performed to obtain IIP3 with the same configuration as in low gain mode,
(4.2.1.2) with the two tones at -30 dBm. The obtained IIP3 was -5.7 dBm. In Figure 118, it is shown its
evolution among the receiver blocks.
Figure 118 – 802.16e low-IF high gain mode IIP3.
80
The obtained total gain and NF were 39 dB and 7.9 dB, respectively. Their evolution among
the receiver blocks is shown in Figure 119.
Figure 119 – Low-IF 802.16e high gain mode Gain and NF.
4.2.3 Adjacent Channel Rejection
4.2.3.1 802.11g
The adjacent channel rejection was simulated as explained in 3.1.4 and show in Figure 120.
Two completely different modulating bitstreams were used for the desired signal and interferer. The
average power of the interfering signal was initially set to -63 dBm to calibrate the receiver gain for
proper demodulation. Its average power was then raised until the BER requirement was no longer
fulfilled.
Figure 120 – 802.11g low-IF adjacent channel power.
The receiver was working in high gain mode and the PGA was operating with a 15 dB gain.
The average power of the interfering could be raised up to -38 dB. This results in a 24 dB adjacent
channel rejection, as shown in Figure 121.
81
Figure 121 – 802.11g low-IF adjacent channel and desired input signal.
By looking at the constellation in Figure 122, it can be seen that the signal is at the limit of
being properly decoded.
Figure 122 – 802.11g low-IF adjacent channel rejection test constellation.
4.2.3.2 802.16e
The adjacent channel rejection was simulated as explained in 3.1.4 and show in Figure 123.
Two completely different modulating bitstreams were used for the desired signal and interferer. The
average power of the interfering signal was initially set to -64 dBm to calibrate the receiver gain for
proper demodulation. Its average power was then raised until the BER requirement was no longer
fulfilled.
Figure 123 – 802.16e low-IF adjacent channel power.
82
The receiver was working in high gain mode and the PGA was operating with a 12 dB gain. The
maximum average power of the interfering could be raised only up to -64 dB. This results in a 4 dB
adjacent channel rejection, as shown in Figure 124.
Figure 124 – 802.16e low-IF adjacent channel spectrum and desired input signal.
In Figure 125, it can be seen why the interferer power can’t be further raised, since the
constellation points are already quite far apart at the minimum rejection required.
Figure 125 – 802.16e low-IF adjacent channel rejection test constellation.
4.2.3.3 Adjacent channel rejection with impairments
In the low-IF architecture the adjacent channel rejection is also affected by I/Q impairments,
especially at the RF quadrature demodulator. So, for this rejection simulation, typical values for
demodulator phase and gain imbalance were considered, 2º and 1%, respectively. The gain
imbalance was defined for Q channel relative to I channel, therefore Q is multiplied by a factor,
2010gain imbalance
g = (54)
where g is 1.01, resulting in a gain imbalance of approximately 0.08 dB. For these imbalances, the
adjacent channel rejection values still fulfill the standards adjacent rejection requirements. Their
values, for both standards, are presented in Table 32.
83
Table 32 – Adjacent channel rejection ratios
802.11g 802.16e
12 dB 4 dB
4.2.4 EVM
The EVM simulation was made using the same methodology as in Zero-IF (4.1.4).
4.2.4.1 802.11g
The signal was transmitted with an output power of 0 dBm and at a 54 Mbps data rate. The
two receiver modes were studied. In high gain mode, the obtained EVM was approximately -30 dB, as
show in Table 33.
Table 33 – 802.11g low-IF high gain mode EVM.
And the corresponding constellation, compared to the ideal one is shown in Figure 126.
Figure 126 – Low-IF 802.11g high gain mode EVM constellation.
In low gain mode, the obtained EVM was approximately -34 dB, as shown in Table 34.
Table 34 – 802.11g low-IF low gain mode EVM.
The corresponding constellation, compared to the ideal one is shown in Figure 127.
84
Figure 127 – 802.11g low-IF low gain mode EVM constellation.
4.2.4.2 802.16e
The signal was transmitted with an output power of 0 dBm at a 54 Mbps data rate. The two
receiver modes were studied. In high gain mode, the obtained EVM was approximately -23 dB, as
shown in Table 35.
Table 35 – 802.16e low-IF high gain mode EVM.
And the corresponding constellation compared to the ideal one is shown in Figure 128.
Figure 128 – 802.16e low-IF high gain mode EVM constellation.
In low gain mode, the obtained EVM was approximately -26 dB, as shown in Table 36.
Table 36 – 802.16e low-IF low gain mode EVM.
85
The corresponding constellation, compared to the ideal one is shown in Figure 129.
Figure 129 – Low-IF 802.16e low gain mode EVM constellation.
4.2.5 Imbalance and Linearity
Phase and gain imbalances were added to see their impact on the BER requirement. The
OIP3 values of the LNA and PGA were lowered to see the minimum value that still manages to pass
the BER requirements.
4.2.5.1 802.11g
Phase and gain imbalances were modified individually. The imbalance ranges, for which the
BER is fulfilled are shown in Table 37.
Table 37 – 802.11g low-IF imbalances.
Gain [dB] Phase [º]
Low gain mode [-1.1 , 2.9] [-8.2 , 6.4]
High gain mode [-0.9 , 2.3] [-5.7 , 7.3]
The LNA and PGA OIP3 values were lowered individually in low gain mode. Their minimum
values are shown in Table 38.
Table 38 – 802.11g low-IF linearity.
LNA PGA
OIP3 [dBm] 3 -12
4.2.5.2 802.16e
Phase and gain imbalances were modified individually. The imbalance ranges, for which the
BER is fulfilled are shown in Table 39.
86
Table 39 – 802.16e low-IF imbalances.
Gain [dB] Phase [º]
Low gain mode [-1.6 , 1.6] [-8.6 , 9.1]
High gain mode [-1 , 0.1] [-3.7 , 0.7]
The LNA and PGA OIP3 values were lowered individually in low gain mode. Their minimum
values are shown in Table 40.
Table 40 – 802.16e low-IF linearity.
LNA PGA
OIP3 [dBm] -6 -19
4.2.6 Results Discussion
In the low-IF simulation the only exception was the ADC VREF value. It had to be 0.95 V
instead of 1 V. This was dictated by the 802.16e requirements. Anticipating a dual band project, this
value was used for both standards.
It was shown that both IP1dB and IIP3 values are mainly determined by the LNA. In Table 41,
it is shown that the linearity and sensitivity requirements of the 802.11g were fulfilled. Even the
obtained EVM at high gain mode fulfills its requirement.
The NF used to validate the standard requirement is always the one obtained in high gain
mode, because in this mode signal quality can be affected and easily degraded by noise. The NF
obtained in high gain mode is lower than the required. It has a quite large adjacent channel rejection,
as desirable. And in every simulation that was made, the BER requirement is achieved.
Table 41 – 802.11g low-IF results.
Obtained Required
IP1dB low gain mode -7.1 dBm ≥ -10 dBm
high gain mode -20.5 dBm ≥ -26 dBm
IIP3 low gain mode 1.5 dBm ≥ 0 dBm
high gain mode -5.3 dBm ≥ -16 dBm
EVM low gain mode -34 dB
≤ -25 dB high gain mode -30 dB
NF 8.3 dB ≤ 10 dB
Adjacent Channel Rejection 24 dB ≥ -1 dB
BER 0.000 ≤ 10-3
87
In EVM simulations it is possible to observe that the spreading of the constellation points is, in
part, due to LO phase noise [31]. Most noticeable in high gain mode, since the input signal is weak
with low power making it more vulnerable to noise.
In Table 42, it is shown that the linearity and sensitivity requirements of the 802.16e were also
fulfilled, with a considerable margin. The NF in high gain mode and the adjacent channel rejection
barely meet the requirements. Only the obtained EVM values didn’t fulfill the standard requirement.
Table 42 – 802.16e low-IF results.
Obtained Required
IP1dB low gain mode -7.1 dBm ≥ -18 dBm
high gain mode -18.6 dBm ≥ -26 dBm
IIP3 low gain mode 0.8 dBm ≥ -8 dBm
high gain mode -5.7 dBm ≥ -16 dBm
EVM low gain mode -26 dB
≤ -31 dB high gain mode -23 dB
NF 7.9 dB ≤ 8 dB
Adjacent Channel Rejection 4 dB ≥ 4 dB
BER 0.000 ≤ 10-6
88
4.3 Dual-Band Receiver Project In the previous chapter, it was seen that the 802.16 dictates the BB NF distribution mainly in
the low-IF architecture. In the zero-IF architecture the LPF cutoff and ADC sampling frequencies are
defined by the 802.11g. And the zero-IF architecture achieves a lower NF than the low-IF one.
In the dual-band architectures two LNAs were used, which is a much more efficient way of
receiving signals in one band rejecting signals from the other band. Although dual-band select filters
and antennas exist, it is quite difficult for a double-band (even switched) LNA to cope with strong
signals approximately 1 GHz aside. The increased circuit size isn’t relevant when compared to its
advantages.
Which architecture is more suitable for a dual band receiver is a decision that the designer will
have to make based on the project restrictions. In the zero-IF case, the problems that arise from the
DC offset and 1/f noise have to be corrected with the DSP interacting with the analog DC levels [50].
But because the output signal is at baseband, the ADC sampling frequency is lower and so is the
power consumption. The proposed dual-band zero-IF architecture is shown in Figure 130.
Figure 130 – Dual-band zero-IF architecture.
Low-IF doesn’t suffer from DC offset and 1/f noise, but the fact that the signal is centered
around 10 MHz leads to a higher power consumption of the IF chain, since it needs an ADC with a
higher sampling frequency. The proposed dual-band low-IF architecture is presented in Figure 131
89
Figure 131 – Dual-band low-IF architecture.
90
5 Conclusions Although, WiMAX current status and the number of subscribers in Q1 2013 are not yet
available, WiMAX is definitely not a dead technology as many people thought, and will continue to
evolve as an excellent WiFi complement.
It’s crucial to extract the standards requirements in first place, before proceeding to the system
level conception of the architecture. It’s also essential to have an idea of the specifications that each
block is able to achieve, so that the system level values are possible to implement.
The receiver output SNR should be calculated in order to know the receiver total NF. This is
important for knowing how to distribute the NF and gain among the receiver blocks.
The 802.11g and 802.16e transmitters were built according to the standards requirements to
transmit only payload at a 54 Mbps data rate with OFDM using 64-QAM modulation, which is the most
demanding case.
The demodulator was built to extract the received bitstream. So it could be compared to the
transmitter modulating bits to verify both standards BER requirements.
The zero-IF and low-IF were designed at a system level, and tested with the modulated OFDM
signal sent by the transmitter in order to verify the standards requirements.
The 802.16e dictates the receiver NF distribution among the receiver blocks of both
architectures.
Both architectures were proposed for a dual-band receiver, since the final choice depends on
the presence or absence of project restrictions.
Concerning future work, error correction could be added to the transmitted signal to see if this
makes the 802.16e EVM requirement achievable.
A low-IF architecture with a complex band-pass filter could be designed and compared with
the one designed in this thesis to verify the standards’ requirements.
The mobile WiMAX transmitter and receiver could be upgraded to support the new mobile
standard, 802.16m.
91
6 Appendix
Figure 132 – 802.11g zero-IF receiver.
Figure 133 – 802.11g low-IF receiver.
Figure 134 – 802.16e zero-IF receiver.
Figure 135 – 802.16e low-IF receiver.
Figure 136 – 802.11g transmitter.
Figure 137 – 802.16e transmitter
Figure 138 – 802.11g/802.16e demodulator.
Figure 139 – 802.11g zero-IF receiver.
Figure 140 – 802.16e zero-IF receiver.
Figure 141 – 802.11g low-IF receiver.
Figure 142 – 802.16e low-IF receiver.
7 References [1] Genesys RF and Microwave Design Software,
http://www.home.agilent.com/agilent/product.jspx?nid=-34275.0.00&cc=PT&lc=eng.
[2] Agilent Ptolemy Simulator,
http://www.home.agilent.com/agilent/product.jspx?nid=-34333.804577.00&cc=PT&lc=eng.
[3] Yanmei Li, "A Platform-Based Approach to Low-Power Receiver Design", PhD Thesis, 2008.
[4] B. G. Lee and S. Choi, Broadband Wireless Access and Local Networks: Mobile WiMax and WiFi:
Artech House, 2008.
[5] O. A. Nabil Alrousan, ljilja Trajkovic, "Handover mechanism of mobile WiMAX (802.16e) with Wi-Fi
(802.11g) Technology."
[6] K. Etemad, "Overview of Mobile WiMAX Technology and Evolution," IEEE Communications
Magazine, 2008.
[7] S. Ahmadi, Mobile WiMAX: A Systems Approach to Understanding IEEE 802.16m Radio Access
Technology: Elsevier Science, 2010.
[8] Wimax Forum, "IEEE 802.16m approved as IMT-Advanced Technology", Press Release, 2010.
[9] Wimax Forum, "WiMAX and the IEEE 802.16m Air Interface Standard", 2010.
[10] Wimax Forum, "Industry Leaders Back Mobile WiMAX™ Enhancements Printer-friendly version",
Press Relase, 2010.
[11] "What is the Difference Between WiMax and LTE ?", http://www.goingwimax.com, 2009.
[12] WiMAX Forum, "WiMAX Forum WiMAX Technology Forecast (2007-2012)", 2007.
[13] Wimax Forum, "WiMAX™ on Track to Cover One Billion by EOY 2011", Press Release, 2011.
[14] Wimax Forum, "WiMAX Subscriptions Surpass 20 Million Globally", Press Release, 2011.
[15] C. Minyeon, O. Dongjin, C. Induck, and K. Ickjin, "A 64-dB spurious free dynamic range CMOS
baseband analog chain for IEEE 802.11a/b/g WLAN," in Radio and Wireless Symposium (RWS),
2011 IEEE, 2011, pp. 315-318.
[16] Z. Pengfei, "CMOS RF receiver: from system architecture to circuit implementation," in ASIC,
2005. ASICON 2005. 6th International Conference On, 2005, pp. 1157-1162.
[17] D. Jakonis, K. Folkesson, J. Dbrowski, P. Eriksson, and C. Svensson, "A 2.4-GHz RF sampling
receiver front-end in 0.18µm CMOS," Solid-State Circuits, IEEE Journal of, vol. 40, pp. 1265-
1277, 2005.
[18] L. Liang-Hui, L. Fu-Lin, and C. Huey-Ru, "Complete RF-System Analysis of Direct Conversion
Receiver (DCR) for 802.11a WLAN OFDM System," Vehicular Technology, IEEE Transactions on,
vol. 56, pp. 1696-1703, 2007.
[19] K. Vavelidis, I. Vassiliou, T. Georgantas, A. Yamanaka, S. Kavadias, G. Kamoulakos, et al., "A
single-chip, 5.15GHz-5.35GHz, 2.4GHz-2.5GHz, 0.18/spl mu/m CMOS RF transceiver for
802.11a/b/g wireless LAN," in Solid-State Circuits Conference, 2003. ESSCIRC '03. Proceedings
of the 29th European, 2003, pp. 221-224.
[20] M. Locher, J. Kuenen, A. Daanen, H. Visser, B. H. Essink, P. P. Vervoort, et al., "A Versatile, Low
Power, High Performance BiCMOS MIMO/Diversity Direct Conversion Transceiver IC for
WiBro/WiMAX (802.16e)," Solid-State Circuits, IEEE Journal of, vol. 43, pp. 1731-1740, 2008.
[21] L. I. M. Alunan, M. T. G. de Leon, and C. R. K. Roque, "System-level simulation and analysis of a
WiMAX direct-conversion receiver in 90nm CMOS," in TENCON 2010 - 2010 IEEE Region 10
Conference, 2010, pp. 1129-1134.
[22] B. J. Farahani and M. Ismail, "WiMAX/WLAN radio receiver architecture for convergence in
WMANS," in Circuits and Systems, 2005. 48th Midwest Symposium on, 2005, pp. 1621-1624 Vol.
2.
[23] M. Simon, P. Laaser, E. Riccio, U. Basaran, D. Friedrich, Y. Raman, et al., "A high performance
dual band/dual mode CMOS RF transceiver for WiMAX and WLAN systems," in Wireless
Technology, 2008. EuWiT 2008. European Conference on, 2008, pp. 81-84.
[24] B. Razavi, RF Microelectronics: Pearson Education, 2011.
[25] X. Li and M. Ismail, Multi-Standard Cmos Wireless Receivers: Analysis and Design: Kluwer
Academic Publishers, 2002.
[26] A. Do, "Energy-Aware Design of 2.4-GHz CMOS Receiver Front-Ends for WPANs," PhD Thesis,
2010.
[27] D. R. D. L. González, "Automatic Design-Space Exploration of Integrated Multi-Standard Wireless
Radio Receivers," Licenciate Thesis, 2006.
[28] J. Laskar, B. Matinpour, and S. Chakraborty, Modern Receiver Front-Ends: Systems, Circuits, and
Integration: Wiley-Interscience, 2004.
[29] J. Vaz, "Radio Architectures," Wireless Integrated Circuits and Systems Group, 2012/2013.
[30] M. Steer, Microwave and RF Design: A Systems Approach: SciTech Pub., 2009.
[31] A. Behzad, Wireless LAn Radios: System Definition to Transistor Design: Wiley-IEEE, 2007.
[32] ATDI, "Mobile WiMax," 2007.
[33] KEITHLEY, "An Introduction to Orthogonal Frequency Division Multiplex Technology,"
Presentation Slides, 2004.
[34] Wikipedia, "Orthogonal frequency-division multiplexing",
http://en.wikipedia.org/wiki/Orthogonal_frequency-division_multiplexing, 2012.
[35] Z. Abate, WiMax RF Systems Engineering: Artech House, 2009.
[36] IEEE Std 802.11-2007, "Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer
(PHY) Specifications" revision of IEEE Std 802.11-1999, June 2007.
[37] IEEE Std 802.16-2009, "Part 16: Air Interface for Broadband Wireless Access Systems", revision
of IEEE Std 802.16-2004, May 2009.
[38] B. Zarlingo, "From Fixed WiMAX 802.16d to Mobile 802.16e: RF and Digital Baseband Testing,"
Agilent Technologies, 2005.
[39] D. Ayadi, S. Rodriguez, M. Loulou, and M. Ismail, "System level design of radio frequency
receiver for IEEE 802.16 standard," in Design and Test Workshop, 2008. IDT 2008. 3rd
International, 2008, pp. 82-86.
[40] LITEPOINT, "Practical Manufacturing Testing of 802.11 OFDM Wireless Devices," Handbook,
2010.
[41] K. L. Edward Gebara, Chang-Ho Lee and Joy Laskar "Integrated CMOS PA Technology for
Cellular Digital Radios," Presentation Slides.
[42] L. Smaini, RF Analog Impairments Modeling for Communication Systems Simulation: Application
to OFDM-based Transceivers: Wiley, 2012.
[43] Q. Gu, RF System Design of Transceivers for Wireless Communications: Springer, 2005.
[44] S.-L. H. Chien-Hung Kuo, "A pipelined ADC with Digital Correction for IEEE 802.11a WLAN,"
Department of Electrical Engineering Tamkang University.
[45] J. Xu, "Custom OFDM Signal Generation Using SystemVue," Application Note, Agilent EEsof
EDA, 2010.
[46] E. H. Miguel, "Fiber-based Orthogonal Frequency Division Multiplexing Transmission Systems,"
Master Thesis, 2010.
[47] M. Rodrigues, "Orthogonal Frequency Division Multiplexing (OFDM): A Primer," Laboratory for
Communication Engineering Department of Engineering, University of Cambridge.
[48] F. J. V. Ramjee Prasad, "OFDMA WiMAX Physical Layer," Aalborg University, Denmark, 2010.
[49] F. Agnelli, G. Albasini, I. Bietti, A. Gnudi, A. Lacaita, D. Manstretta, et al., "Wireless multi-standard
terminals: system analysis and design of a reconfigurable RF front-end," Circuits and Systems
Magazine, IEEE, vol. 6, pp. 38-59, 2006.
[50] M. Brandolini, P. Rossi, D. Manstretta, and F. Svelto, "Toward multistandard mobile terminals -
fully integrated receivers requirements and architectures," Microwave Theory and Techniques,
IEEE Transactions on, vol. 53, pp. 1026-1038, 2005.
[51] F. Beaudoin, T. Zortea, G. Deliyannides, M. Hiebert, M. McAdam, M. Venditti, et al., "A fully
integrated tri-band, MIMO transceiver RFIC for 802.16e," in Radio Frequency Integrated Circuits
Symposium, 2008. RFIC 2008. IEEE, 2008, pp. 113-116.
[52] J. Deguchi, D. Miyashita, Y. Ogasawara, G. Takemura, M. Iwanaga, K. Sami, et al., "A Fully
Integrated 2 x 1 Dual-Band Direct-Conversion Mobile WiMAX Transceiver With Dual-Mode
Fractional Divider and Noise-Shaping Transimpedance Amplifier in 65 nm CMOS," Solid-State
Circuits, IEEE Journal of, vol. 45, pp. 2774-2784, 2010.
[53] C. Carta, R. Vogt, and W. Bachtold, "Multiband monolithic BiCMOS low-power low-IF WLAN
receivers," Microwave and Wireless Components Letters, IEEE, vol. 15, pp. 543-545, 2005.
[54] S. Rodriguez, A. Rusu, and M. Ismail, "4G CMOS nanometer receivers for mobile systems:
Challenges and solutions," in Signals, Circuits and Systems, 2009. ISSCS 2009. International
Symposium on, 2009, pp. 1-4.
[55] A. Taparia, S. A. Nakvi, and B. Banerjee, "A low-IF WiMAX RF transceiver in 0.18µm CMOS
technology," in Circuits and Systems Workshop: System-on-Chip - Design, Applications,
Integration, and Software, 2008 IEEE Dallas, 2008, pp. 1-4.
[56] H. Da-Rong, K. Shiau-Wen, and P. Yi-Hsin, "A WiMAX Receiver with Variable Bandwidth of 2.5 -
20 MHz and 93 dB Dynamic Gain Range in 0.13µm CMOS Process," in Radio Frequency
Integrated Circuits (RFIC) Symposium, 2007 IEEE, 2007, pp. 369-372.
[57] EPCOS, "SAW Components " September 02, 2008.
[58] NTK, "NTK RF Components for WiMAX," ELECTRONIC COMPONENTS DIVISION, 2008.