Gigabit Access in Wirelessworkshop.nkn.in/2013/images/presentation/2nd... · A large MIMO technology demonstrator project Goal Demonstrate Gigabit transmission over-the-air Joint

GIGABIT ACCESS IN WIRELESS

A. Chockalingam

Department of ECE, IISc

Second Annual NKN Workshop

Bangalore

18 October 2013

A. Chockalingam ( Department of ECE, IISc ) Gigabit Access in Wireless 18 October 2013 1 / 53

Outline

1 GIGABIT WIRELESS - STATE-OF-THE-ART

2 MIMO - AN ACE PHY FEATURE IN GIGABIT WIRELESS

3 SPATIAL MODULATION - ANOTHER ACE PHY FEATURE


Gigabit Wireless - State-of-the-art

A recent (Dec’2012) wireless demonstration

Parameter Value

Data rate 10 Gbps

Bandwidth 400 MHz

Spectral efficiency 25 bps/Hz

Carrier frequency 11 GHz

Environment Urban

Mobility 9 km/hr

Technology 8 × 16 MIMO64-QAM

(a) (b)



Another recent (May’2013) wireless demonstration

Parameter Value

Data rate 1.056 Gbps

Bandwidth ?

Spectral efficiency ?

Carrier frequency 28 GHz

Distance 2 km

Technology * Adaptive antenna array* 64 antenna elements



Gartner’s hype cycle

Source: Internet



Moore’s law drives wireless data rates

Source: SPAWC’2010 plenary talk slides of Dr. Gerhard Fettweis















Increasing wireless data rates

New spectrum (bps)

increase BW (e.g., 60 GHz band, mm wavelength, 7 GHz BW)

+: unlicensed (free)

-: propagation characteristics, devices, short range, cost

Increase QAM size (bps/Hz)

MIMO (bps/Hz)

+: Theory has predicted unlimited capacity

-: Practicality, complexity, cost

Dense deployments (bps/Hz/km2)

Femtocells

+: 1000x speed up (claimed)

-: interference management, backhaul, cost



Evolution to Gigabit WiFi (and beyond)

IEEE Band BW Data rates PHY features Spectral Eff.Standard per channel (bps/Hz)

802.11b 2.4 GHz 5 MHz 11,5.5,2,1 Mbps DS-SS, CCK 0.5, 2

802.11g 2.4 GHz 20 MHz 1 - 54 Mbps OFDM 2.5

802.11a 5 GHz 20 MHz 54,48,36,24,18 OFDM12,9,6 Mbps 64 subcarriers 2.5

MIMO-OFDM802.11n 5/2.4 GHz 20/40 MHz 600 Mbps 4 × 4 MIMO 15

128 subcarriers

802.11ac 5 GHz 80/160 MHz 1 Gbps MU-MIMO 6.25

802.11ad 60 GHz 7 GHz up to 7 Gbps Beamforming < 2

HEW ? ? ? ? ↑

HEW: High Efficiency WiFi

Bands other than 2 GHz and 5 GHz

802.11af (White-Fi): TV white spaces, sub-1GHz (cognitive radio, geographic sensing)802.11ah: non-TV white spaces, sub-1GHz (Internet of Things (IoT), Machine to Machine (M2M))

802.11aj: 60 GHz (5 GHz BW) – China-centric

Emerging use cases (under discussion) – relevant for India



IEEE 802.11 MAC

Two protocols

PCF: Point coordination function (polling)

DCF: Distributed coordination function (random access)

DCFCSMA/CA

RTS/CTS handshake before transmission of data packet

Avoids hidden node problem

ACK for data packet

Backoff mechanism to resolve collisions

backoff parameters: CWmin, CWmax

Minimum silence periods between transmissions

DIFS: DCF Inter-Frame Spacing

SIFS: Short Inter-Frame Spacing

Shorter minimum waiting implies higher priority (ACK, CTS)



IEEE 802.11 MAC - CSMA/CA

Source: http://secowinet.epfl.ch/slides



IEEE 802.11 MAC (DCF) Throughput

IEEE 802.11b

Max. raw data rate: 11 Mbps

Useful throughput is much less due to CSMA/CA overhead

Application using TCP: 5.9 Mbps

Application using UDP: 7.1 Mbps

CSMA/CA overhead

Min. overhead for sending one data packet

= Tx time of (1 RTS + 1 CTS + 1 ACK + 3 SIFS + 1 DIFS + 4

preambles)

In addition, loss due to collision and retransmissions



DCF MAC (In)efficiency in Gigabit WiFi

Suppose

RTS = CTS = ACK = Preamble = 20 bytes

SIFS = 16 µsec, DIFS = 34 µsec, Data packet = 2500 bytes

Assume ideal conditions

No channel errors, no collision (i.e., point-to-point Tx)

Case a) say, Rate = 54 Mbps. Useful throughput?

Ans: 42.3 Mbps (about 78% of 54 Mbps)

Case b) say, Rate = 1 Gbps. Useful throughput?

Ans: 194 Mbps (only 19.4% of 1 Gbps)

MAC (in)efficiency is a concern in Gigabit WiFi



Back to Gartner’s hype cycle

Hype cycle for Communication and Networking, 2011

Source: Internet



Back to Gartner’s hype cycle

Hype cycle for Communication and Networking, 2013

Source: Internet


MIMO - An ace PHY feature in Gigabit wireless

MIMO – an ace PHY feature

Why MIMO?

nt : # of transmit antennas, nr : # receive antennas

# Antennas Error Probability (Pe) Capacity (C), bps/Hz

SISO

nt = nr = 1 Pe ∝ SNR−1 C = log(SNR)

SIMO

nt = 1, nr > 1 Pe ∝ SNR−nr C = log(SNR)

MIMO

nt > 1, nr > 1 Pe ∝ SNR−nt nr C = min(nt , nr ) log(SNR)

MIMO technology scores high on

Spectral efficiency

Power efficiency

Link reliability



Increasing spectral efficiency: QAM vs MIMO

(c) SISO/SIMO with 64-QAM (d) MIMO with nt = 6 and BPSK

Spectral efficiency in both systems: 6 bps/Hz



Increasing spectral efficiency: QAM vs MIMO

0 5 10 15 20 25 30 35 40 45 5010

-4

10-3

10-2

10-1

100

Average recieved SNR (dB)

Bit Error Rate

SISO, nt=1, nr=1, 64-QAMSIMO, nt=1, nr=6, 64-QAMMIMO, nt=6, nr=6, BPSK

6 bps/Hz



Large MIMO systems

Larger the number of antennas, better will be the

spectral efficiency

power efficiency

reliability

Large MIMO systems

MIMO systems where communication terminals use

tens to hundreds of antennas

Achieve very high spectral efficiencies in the range of

tens to hundreds of bps/Hz



Technological challenges

Placement of large no. of antenna elements

Feasible in moderately sized communication terminals

Use high carrier frequencies (small carrier wavelengths);

e.g., 5 GHz, 60 GHz

Compact antenna arrays

RF technologies

Multiple IF/RF transmit and receive chains

Spatial modulation

Allows use of less number of Tx RF chains than the

number of Tx antennas

Large MIMO signal processing

Signal detection, channel estimation, decoding, precoding

Channel hardening in large random matrices help



Channel hardening in large random matrices

Magnitude plots of HHH for different sizes of random matrix H

02

46

8

0

2

4

6

8-10

-5

0

5

10

15

8 x 8

(e) 8 × 8





02

46

8

0

2

4

6

8-10

-5

0

5

10

15

8 x 8

(i) 8 × 8

010

2030

40

0

10

20

30

40-40

-20

0

20

40

60

32 x 32

(j) 32 × 32





02

46

8

0

2

4

6

8-10

-5

0

5

10

15

8 x 8

(m) 8 × 8

010

2030

40

0

10

20

30

40-40

-20

0

20

40

60

32 x 32

(n) 32 × 32

0 20 40 60 80 100

0

50

100-50

0

50

100

150

96 x 96

(o) 96 × 96





02

46

8

0

2

4

6

8-10

-5

0

5

10

15

8 x 8

(q) 8 × 8

010

2030

40

0

10

20

30

40-40

-20

0

20

40

60

32 x 32

(r) 32 × 32

0 20 40 60 80 100

0

50

100-50

0

50

100

150

96 x 96

(s) 96 × 96 (t) 256 × 256



Simple algorithms – Good performance

Local search based signal detection

1 2 3 4 5 6 7 8 9 1010

-6

10-5

10-4

10-3

10-2

10-1

100

Average Received SNR (dB)

Bit

Err

or R

ate

ZF-LAS (1 x 1)ZF-LAS (10 x 10 MIMO)ZF-LAS (50 x 50 MIMO)ZF-LAS (100 x 100 MIMO)ZF-LAS (200 x 200 MIMO)ZF-LAS (400 x 400 MIMO)

Increasing # antennasimproves BER performance

BPSK

————————-* K. V. Vardhan, S. K. Mohammed, A. Chockalingam, and B. S. Rajan, A low-complexity detector for large MIMO systems and

multicarrier CDMA systems, IEEE J. Sel. Areas Commun., vol. 26, no. 3, pp. 473-485, Apr. 2008.



Project NAVA

A large MIMO technology demonstrator project

Goal

Demonstrate Gigabit transmission over-the-air

Joint project: IISc, DRDO, and private industry

IISc provides system design, core algorithms and IPs

Private industry : develop/manufacture main subsystems



NAVA

IP

Ethernet

Application

UDP

IP

Ethernet

Application

UDP NAVA

NAVA

B

Terminal

NAVA

A

Ethernet

10GbEthernet

10Gb

Ethernet

Video

server

NAVA

MACEthernet

ClientVideoserver

Terminal

Client

MAC

NAVA PHY NAVA PHY



NAVA

System



NAVA

High level specifications

Parameter Value

Data rate 1 Gbps

Bandwidth 40 MHz

Spectral efficiency 25 bps/Hz

Carrier frequency 2.5 GHz

No. transmit antennas 16

No. receive antennas 20

Frequency plan

40 MHz 40 MHz

Downlink Uplink

2.475 GHz 2.725 GHz



NAVA - Antenna unit

20-antenna MIMO cube at 2.5 GHz

technology: PIFA



NAVA - RF unit

16 Tx chains: IF: 220 ± 20 MHZ; RF: 2725 ± 20 MHz

20 Rx chains: RF: 2475 ± 20 MHz; IF: 140 ± 20 MHz



NAVA Baseband unit

.

.

.

.

.

.

35 MHz20MHz±

20MHz±

20MHz±

Interfa

ce

Ethernet

T

x

R

x

Rx

Tx

1 2 n

MAC

FPGA

DAC1

DAC2

DAC16

ADC1

ADC2

ADC20

Keyboard

Display

UserData

10Gb

Ethernet

SerialConfiguration

(Laptop/PC) Interface

NAVA

NAVA

NAVA

Rx FPGAs

connectors

BU

MAC PHY

220 MHz

140 MHz

±20MHz24.8 MHz

20

FPGA

SMA

(From

Tx

connectors

16 SMA

(to RFU)

RFU)(Front panel)

/



NAVA - Baseband unit



NAVA - Digital board






NAVA - IF converter board






Inside NAVA FPGAs



NAVA terminal



Large multiuser MIMO (proposed architecture for 5G)

BS with hundreds of antennas & tens of users with 1 antennaeach

Massive MIMO, Hiper-MIMO, Higher-order MIMO, Large-scale MIMO

S. K. Mohammed, A. Chockalingam, and B. S. Rajan, A low-complexity precoder for large multiuser MISO systems,

Proc. IEEE VTC’2008, pp. 797-801, May 2008.



Large multiuser MIMO


Spatial modulation - Another ace PHY feature

Spatial modulation

Space shift keying (SSK)

nt transmit antennas; 1 transmit RF chain

m = log2 nt bits choose an antenna

chosen antenna transmits a tone; other antennas remain silent

bits conveyed through antenna index (m bpcu)

nt = 64, 1 Tx RF chain, 6 bps/Hz



Spatial modulation

SSK performance

0 5 10 15 20 25 30 35 40 45 5010

-4

10-3

10-2

10-1

100

Average recieved SNR (dB)

Bit E

rro

r R

ate

SISO, Nt=1, Nr=1, 64-QAMSIMO, Nt=1, Nr=6, 64-QAMMIMO, Nt=6, Nr=6, BPSKSSK, Nt=64, Nr=6

6 bps/Hz



Spatial modulation

An M-ary modulation symbol (e.g., M-QAM) is sent on the chosen

antenna

m + log2 M bpcu

Data bits to SM signal mapping for m = 2, nt = 4

Antenna sel. SM Tx. signal Status of Tx antennas (nt = 2m = 4)

bits, m = 2 vector, x Antenna 1 Antenna 2 Antenna 3 Antenna 4

0 0 [x , 0, 0, 0]T x ∈ AM OFF OFF OFF

0 1 [0, x , 0, 0]T OFF x ∈ AM OFF OFF

1 0 [0, 0, x , 0]T OFF OFF x ∈ AM OFF

1 1 [0, 0, 0, x ]T OFF OFF OFF x ∈ AM



GSM

Two limitations in SM and SSK

nt limited to powers of 2

number of RF chains restricted to 1

GSM removes both the above restrictions

nt is not restricted to power of 2

nrf transmit RF chains, 1 ≤ nrf ≤ nt

In GSM

nrf out of nt antennas will be active simultaneously

an nrf × nt switch connects RF chains to Tx antennas

each active antenna will send a M-ary symbol on it

remaining nt − nrf antennas remain silent

Spectral efficiency of GSM

R =

⌊

log2

(nt

nrf

)⌋

︸︷︷︸

# ant. sel. bits

+ nrf log2 M︸︷︷︸

# M-ary modln. bits

bpcu



GSM

Total no. of antenna activation patterns: L =(

nt

nrf

)

Only 2K, K =

⌊

log2

(

nt

nrf

)

⌋

patterns are needed

Select any K patterns out of L patterns and form a set

Call this set as ‘antenna activation pattern set’, S

An example:

Let nt = 4, nrf = 2, M = 4 (i.e., 4-QAM)

=⇒ L = 6, K = 2, R = 6 bpcuPossible activation patterns (L = 6):

{

[1, 1, 0, 0], [1, 0, 1, 0], [0, 1, 0, 1], [0, 0, 1, 1], [0, 1, 1, 0], [1, 0, 0, 1]}

Chosen activation patterns (2K= 4):

S ={

[1, 1, 0, 0], [1, 0, 1, 0], [0, 1, 0, 1], [0, 0, 1, 1]}



GSM

Data bits to GSM signal mapping for nt = 4, nrf = 2

6 bpcu for 4-QAM

Data bits Ant. activity Antenna status

K = 2 bits pattern Antenna 1 Antenna 2 Antenna 3 Antenna 4

0 0 [1, 1, 0, 0]T x1 ∈ AM x2 ∈ AM OFF OFF

0 1 [1, 0, 1, 0]T x1 ∈ AM OFF x2 ∈ AM OFF

1 0 [0, 1, 0, 1]T OFF x1 ∈ AM OFF x2 ∈ AM

1 1 [0, 0, 1, 1]T OFF OFF x1 ∈ AM x2 ∈ AM

Example:

Let 010011 denote the information bit sequence

1st two bits choose activity pattern

2nd two bits form one 4-QAM symbol

3rd two bits form another 4-QAM symbol

Tx vector is x = [1 + j, 0, −1 − j, 0]T



SSK, SM, GSM

Parameters and spectral efficiencies of SSK, SM, GSM

Modulation # Tx antennas # RF chains Spectral efficiency

(nt ) (nrf ) (bpcu)

SSK 2m, m ∈ {1, 2, · · · } 1 m

SM 2m, m ∈ {1, 2, · · · } 1 m + log2 M

GSM ∈ {1, 2, · · · } ∈ {1, · · · , nt}⌊

log2

(ntnrf

)⌋

+ nrf log2 M



Achievable rates in GSM

0 5 10 15 20 25 300

10

1720

30

35

40

50

6064

70

80

Number of transmit RF chains, n rf

Ach

ieva

ble

ra

te,

R (

bp

cu

)

nt=4

nt=8

nt=12

nt=16

nt=22

nt=32

nrf

=13

nrf

=24

nrf

=16

Achievable rate R as a function of nrf in GSM for different values of nt and 4-QAM.

————————-* T. Datta and A. Chockalingam, On Generalized Spatial Modulation, IEEE WCNC’2013, Shanghai, Apr. 2013.



NAVA-Plus

GSM extension to NAVA

Modulation nt nrf Spectral efficiency

4-QAM 16 16 32 bps/Hz

4-QAM 20 20 40 bps/Hz

4-QAM 20 16 44 bps/Hz



Book

Release by January 2014



4 P’s

Papers

Patents

Prototypes

Products



Concluding remarks

Gigabit wireless – a reality now

can trigger interesting and new use cases and applications



Concluding remarks



Large MIMO is a key enabling technology

major technological bottlenecks have been cleared

under various stages of development and testing worldwide

being considered as the technology for 5G, HEW



Concluding remarks







India needs to

get active in the standardization efforts of 5G and HEW



Concluding remarks







India needs to


push for India-centric use cases and interests



Concluding remarks







India needs to



promote product companies and the eco-system



Concluding remarks







India needs to



promote product companies and the eco-system

grow the technology base and exploit the Indian market



Thank you


Documents

Gigabit Access in Wirelessworkshop.nkn.in/2013/images/presentation/2nd... · A large MIMO technology demonstrator project Goal Demonstrate Gigabit transmission over-the-air Joint