Embed Size (px)
Citation preview
Design of a battery free wireless identification and sensing
platform
Alanson Paul Sample
A thesis submitted in partial fulfillment of the requirements for the degree of
Master of Science in Electrical Engineering
University of Washington
2008
Program Authorized to Offer Degree: Department of Electrical Engineering
University of Washington Graduate School
This is to certify that I have examined this copy of a master’s thesis by
Alanson Paul Sample
and have found it complete and satisfactory in all respects, and that any and all revisions required by the final
examining committee have been made.
Committee Members:
_________________________________________
Alexander Mamishev
_________________________________________
Joshua Smith
_________________________________________
Brian Otis
Date: _________________
In presenting this thesis in partial fulfillment of the requirements for a master’s degree at
the University of Washington, I agree that the Library shall make its copies freely
available for inspection. I further agree that extensive copying of this thesis is allowable
only for scholarly purposes, consistent with "fair use" as prescribed in the U.S.
Copyright Law. Any other reproduction for any purposes or by any means shall not be
allowed without my written permission.
Signature_______________________________
Date___________________________________
i
Table of Contents
Chapter 1 Introduction.................................................................................................... 1
1.1 Motivation.......................................................................................................... 1
1.2 Objective ............................................................................................................ 2
1.3 Potential Impact ................................................................................................. 3
1.4 Scope.................................................................................................................. 4
Chapter 2 Background .................................................................................................... 6
2.1 RFID system ...................................................................................................... 6
2.2 RFID Reader ...................................................................................................... 7
2.3 RFID Protocol: EPC Class 1 Gen 2 ................................................................. 11
2.4 RFID Tags........................................................................................................ 12
2.4.1 Passive Tags................................................................................................. 13
2.4.2 Active Tags .................................................................................................. 16
2.5 State of the Art - Sensor Enhanced RFID Tags ............................................... 17
2.5.1 Prior Work ................................................................................................... 17
2.5.2 Prior Work on WISP.................................................................................... 19
Chapter 3 Wireless Sensing & Identification Platform ................................................ 21
3.1 Analog Front End............................................................................................. 23
3.2 Demodulation and Modulation ........................................................................ 25
3.3 Digital Section and Power Conditioning ......................................................... 26
Chapter 4 Firmware & Power Management Algorithm ............................................... 28
4.1 Power Management Algorithm........................................................................ 28
4.2 Communication and Application Layers ......................................................... 30
Chapter 5 Power Budget............................................................................................... 32
5.1 Turn On Power Requirement ........................................................................... 32
5.2 Duty Cycle ....................................................................................................... 34
ii
5.3 Active Energy Consumption............................................................................ 35
Chapter 6 Communicating with the WISP ................................................................... 37
6.1 Overloading the Identifier................................................................................ 37
6.2 Gen 2 Read Command ..................................................................................... 38
6.3 Gen 2 Select Command ................................................................................... 38
6.4 Gen 2 Write Command .................................................................................... 39
Chapter 7 Experimental Results ................................................................................... 40
Chapter 8 Sensors and Peripherals ............................................................................... 43
Chapter 9 Conclusion ................................................................................................... 49
Chapter 10 Future Work ............................................................................................. 52
iii
List of Figures
Figure Number Page
1 Progression of work on wireless sensing and computing based on passive
RFID technology........................................................................................................ 5
2 Block Diagram of an UHF RFID System. ................................................................. 7
3 EPC Generation 2 RFID readers.............................................................................. 10
4 915 MHz RFID reader patch antennas with covers. ................................................ 10
5 915 MHz RFID reader patch antennas without covers. ........................................... 11
6 EPC RFID tag class structure (source www.epcglobalinc.org). .............................. 13
7 Block Diagram of a typical commercial RFID tag. ................................................. 15
8 Alien RFID tag, Model ALN-9540 "Squiggle" ....................................................... 15
9 Alien RFID tag, Model: ALL-9460 "Omni-Squiggle" ............................................ 16
10 The α-WISP uses two tilt switches orientated in opposite directions as a
simple one bit RFID accelerometer. ........................................................................ 19
11 Block Diagram of the WISP. ................................................................................... 22
12 Wireless Identification and Sensing Platform (WISP). ........................................... 22
13 Data logger version of the Wireless Identification and Sensing Platform............... 23
14 Schematic of the Analog Front End......................................................................... 24
15 Oscilloscope plot of the demodulated data extracting data from the RF
waveform transmitted by the RFID reader. ............................................................. 26
16 Block diagram of the power management algorithm for the WISP......................... 29
17 Oscilloscope scope plot of the WISP responding to EPC queries along
with its rectified voltage........................................................................................... 31
18 Rectified voltage (left scale) and output power at 1.9V (right scale) are
plotted verse input power measurements using multi-meter and network
analyzer for RF signal insertion into the antenna ports. .......................................... 34
19 WISP performance: harvested voltage, Uplink Packet Errors, and
Responses per Query as a function of input power.................................................. 42
iv
20 Cold impulses are applied to WISP and a Fluke thermal probe and
plotted over time. ..................................................................................................... 44
21 Light level measured by WISP in a 13 hour period................................................. 45
22 Received measurement of the acceleration of earths gravity long two
axes using the WISP enabled with a 3D accelerometer........................................... 46
23. WISP data logger with operational Super capacitor. ............................................... 48
v
Dedication
To my wife
Brittany
and my parents
Michael, Melinda, Karen, & Mani
1
Chapter 1 Introduction
1.1 Motivation
Over the last decade, advances in integrated circuit design and RFID
manufacturing techniques have enabled a cost effective method for producing passive
Radio Frequency Identification (RFID) tags, capable of reporting a unique ID when
queried. These devices typically consist of a printed antenna bonded to an integrated
circuit (IC), and are completely powered from the RF energy transmitted by an RFID
reader. Once the tag is powered on, the IC’s digital logic analyzes data from the RFID
reader and responds with a unique ID when necessary [10].
To date, industrial efforts have focused on the long standing goal of
manufacturing a passive UHF RFID tag, which sells for five cents. This price is widely
viewed as the tipping point at which wide-spread adoption of RFID technology will
occur. Unfortunately, there are numerous obstacles facing the industry and, for the
foreseeable future, this target is unobtainable [26][29].
On the other hand, by adding a small amount of functionality to passive RFID
tags, it has been shown that profitability can be significantly increased. For instance,
Intermec sells a rigid RFID tag that can be placed on metal objects and is capable of
withstanding extreme temperatures and exposure to hazardous environments. This
product sells for around five dollars, but still uses a standard RFID chip [14]. Another
product, from KSW-Mitochon, incorporates a temperature logger into a near field active
RFID that can be used to detect dangerous temperatures in food products during transit
[22].
Both of these examples suggest that, although the price of traditional RFID tags
may not decrease sufficiently for all potential tracking applications, enhanced tags can
provide increased functionality, which is more valuable to the customer then wireless
identification alone. Thus, enhancing RFID tags and developing new applications that
can create innovation in the RFID industry is one motivation for this research.
2
While the functionality of commercial passive RFID tags is extremely limited,
today's tags are essentially a wireless micro-computing platform with a RF transceiver
and a nearly unlimited lifetime. As these tags become more pervasive in our work places
and homes, they present an untapped resource that is continually waiting in the
background for power and are able to independently execute instructions.
A second motivation for this research is to capture the truly wireless, micro-
computing nature of passive RFID and explore applications beyond simple barcode
replacement. This new paradigm of transmitting both data and power to an object
represents a fundamental shift in the way people use technology. Just as the advent of
wireless communication technology enabled an expansive array of mobile products and
gadgets, cutting the last cord allows for a new set of devices and usage models to
become feasible. Due to advances in the semi-conductor industry it should be now
theoretically possible to create a class of devices, enhanced with sensors and capable of
doing computational work, that never needs to be physically wired for data and power,
nor ever needs to have their batteries replaced.
1.2 Objective
The objective of this thesis is to investigate the building blocks and algorithms
necessary to create a general purpose Wireless Identification and Sensing Platform.
Referred to as a WISP, this device will be an enhanced passive RFID tag, with sensing
and computing capabilities, that remains completely wirelessly powered. In particular,
this thesis aims to accomplish the following:
• Investigate methods for harvesting RF energy from standard UHF RFID
reader
• Integrate sensors and a general purpose microcontroller into a passive
RFID platform
• Determine the limitation of powering large loads from harvested power
• Communicate data using the standard a RFID protocol
3
• Design a PCB-based, flexible RFID tag that can be reused to investigate a
wide variety of applications
1.3 Potential Impact
The Wireless Identification and Sensing Platform (WISP) proposed in this thesis
is intended to be a research vehicle that allows people inside and outside the RFID
community to explore new applications and usage models for RFID. Traditionally, RFID
tag designers have been specialists in integrated circuit design. They have generally
focused on innovating CMOS circuit blocks, such as RF rectification, power
management, and low power state machines, with the goal of increasing tag read range.
The process of manufacturing these custom Integrated Circuit (IC) tags presents a
significant barrier to entry when considering the high cost of software, servers, chip
fabrication, and specialized testing equipment, not to mention the long fabrication
cycles.
Consequently, when researchers do add additional functionality, such as ADC
and light sensors, the focus is on the device and is not driven by any particular
application. It is important to note that custom IC tag design will undoubtedly offer
longer range, better performance vs. power consumption, and lower manufacturing costs
in large volumes. However, it is difficult under this design paradigm to develop new
applications that will take RFID beyond simple item tracking and identification.
In contrast, WISPs are PCB-based flexible platforms that allow a relative novice
to prototype both hardware and software RFID designs in a bench-top setting. The full-
featured microcontroller allows for fast code development with debugging support.
Sensors and peripherals can be easily added via the exposed headers or by using an
optional daughter board. Testing equipment generally consists of an RFID reader and an
oscilloscope. When compared to IC tags, probing and debugging circuit elements is easy
and straightforward, as many of the signal lines are exposed by the PCB design.
The WISP fundamentally lowers the barrier to entry and allows people from a
wide variety of fields to develop RFID technology. Whether it is students as part of a
4
class project, security specialists, consumer electronics designers, or even artists, it is
believed that a diverse group of people will be able to advance RFID technology and
find new application spaces and usage models. The hope is that when compelling ideas
are discovered, IC tag designers will be able to draw upon the lessons learned from the
WISP implementation when designing their custom tags.
1.4 Scope
In an effort to increase the capability of RFID based sensing, the author of this
thesis has investigated topics which have culminated in the creation of the Wireless
Identification and Sensing Platform (WISP) [24][25][28]. This thesis focuses on the
integration of sensing and computational capability into a UHF passive RFID tag. The
hardware design of the PCB-based WISP is presented. In particular, RF harvesting
techniques, communication, and power management circuitry are described, along with
performance analysis.
Figure 1 shows an overview of efforts on RFID based sensing and computing
made by members of Intel Research Seattle and the department of Electrical Engineering
at the University of Washington over the last several years. Significant contributions by
the author are marked with green dots. Yellow triangles indicate team efforts. Green dots
and yellow triangles represent efforts towards problems with iterative solutions where
the author made contributions as well as worked with team members during the design
process.
A description of the power management algorithm implemented by the MCU and
in conjunction with the hardware features is discussed. However, a detailed discussion
of the code used to implement the EPC protocol (Gen1 and Gen2), developed by Dan
Yeager with assistance form Polly Powledge, is not discussed. Although, the usage of
the EPC protocols for transmitting data to and from the WISP is presented for
completeness.
Finally, since the completion of the WISP, a number of researchers have been
able to implement new RFID applications using the platform. Their work ranges from
5
medical to security applications, and are described in [31][12][13]. Additionally, a low-
cost, touch interface for passive RFID tags, which is compatible with modern
manufacturing processes, has been designed and implemented by the author.
Initial RFID Sensing Prototypes
Alpha-WISP: used mercury switches to select one of two
commercial RFID ICs
Pi-WISP: A MCU used a digitally controlled RF mux to switch
between two commercial RFID ICs
Wireless, Battery Free
Computational &
Sensing Platform
RFID
Transceiver
UHF RFID
Energy
Harvesting
EPC
Demodulation
(ASK)
Back-Scatter
Modulation
Microcontroller
(MSP430)
EPC Gen 1
EPC Gen 2
Wireless
Power
Managment
Sensor
Integration
Light
3D
Acceleration
Temperature
UHF Rectifier
Antenna
Design
Wireless Identification and Sensing Platform
Power
Management
Algorithm
Capacitive TouchA Low-Cost Capacitive Touch
Interface
for Passive RFID Tags
Passive Data LoggerWirelessly-Charged UHF Tags for
Sensor Data Collection
Neural WISPAn Energy Harvesting Wireless
Brain Interface with 1m Range
SecurityMaximalist cryptography and
computation on the WISP UHF
RFID tag
RFIDs and Secret
HandshakesDefending Against Ghost-and-
Leech Attacks with Context-Aware
Communications
RFID Enhanced DevicesPacemakers and implantable
cardiac defibrillators: Software radio
attacks and zero-power defenses
RFID Applications using the WISP
Figure 1 Progression of work on wireless sensing and computing based on passive RFID technology.
6
Chapter 2 Background
2.1 RFID system
In recent years, rapid development of Radio Frequency Identification (RFID)
technology has resulted in a wide variety of applications and devices used for
identification and tracking purposes. RFID systems typically consist of small, low-cost,
battery-free devices called “tags”, which use the radio signal from a specialized RFID
reader for power and communication. When queried, each tag responds with a unique
identification number by reflecting energy back to the reader though a technique called
backscatter modulation. Tags are application specific, fixed function devices that have a
range of 10-50cm for inductively coupled devices and 3-10m for UHF tags.
Traditionally, RFID tags have been used as a replacement for barcodes in applications
such as supply chain monitoring, asset management, and building security [10].
Figure 2 shows a block diagram of a typical UHF RFID installation scenario. A
RFID reader queries a set of tags by transmitting energy and data through its antenna.
This energy travels through the air in the “far field” regime, meaning that energy is
radiated away from the reader as a coupled electromagnetic wave. If the tag is within the
interrogation field of the reader, an alternating RF voltage will be induced on the
transponder’s antenna. This RF signal is then converted into DC power by the tag’s
rectifier and used to turn on its state machine and digital logic. Once activated, the RFID
transponder will analyze the data from the reader and reply with its ID number when
selected by the reader. The RFID reader implements a thorough protocol to identify all
tags in its interrogation field and reports the data back to a host computer or database.
7
UHF RFID
Reader
Tag
Tag
Tag
RF Power & Data
Reader
Antenna
Host
Computer
Figure 2 Block Diagram of an UHF RFID System.
2.2 RFID Reader
In typical applications, the RFID readers act as a middle man between the RFID
tags and the application layer. The role of the reader is to provide power to the tags,
identify individual tags out of a population, and report those IDs to the host. Figure 3
shows three popular EPC Gen2 UHF RFID readers. UHF readers generally drive linearly
or circularly polarized patch antennas due to their good antenna gain and small size.
Three types of patch antennas are shown with and without covers in Figure 4 and Figure
5, respectively.
All readers must perform the following primary functions:
• Provide power to the RFID tag in the form of a RF carrier wave
• Transmit data to the tag by modulating the RF carrier wave
• Detect and decode the backscatter signal from the transponder
• Singulate individual tags out of a population
• Report the ID of the tags in the interrogation field to a host application
8
RFID readers are highly specialized forms of two-way radios, which transmit
data as well as power to the tag. In order to deliver power to the tag, the reader emits a
continuous wave that is received by the tag’s antenna and rectified into DC power. To
transmit data, the reader modulates the amplitude of its carrier. Clearly, the more power
that is transmitted the longer the read range of the tag. However, there are two main
factors that limit the amount of energy that actually reaches the tag.
First, there are regulatory limitations on the amount of RF power that can be
safely transmitted. In the United States, the Federal Communication Commission (FCC)
limits the power transmitted in the Industrial, Scientific, and Medical (ISM) radio band
(902 MHz – 928 MHz) to 4W (EIRP) [9]. Effective Isotropic Radiation Power (EIRP),
also known as Equivalent Isotropic Radiation Power, refers to the peak power density
that a theoretical isotropic antenna (which distributes power evenly in all directions)
would produce for a given input power. Basically, the FCC wants to limit the amount of
power per unit area to insure safety. However, power density is not easily measured.
Thus, equation (1) is used to normalize the power output of antennas with different gains
so that the power density in any one direction is equivalent to an isotropic antenna.
cat LGPEIRP −+= (1)
Pt is the transmitter power measured in dBm, while Ga is the antenna gain
expressed in dBi. Often times, it is important to take into consideration the losses of the
cable that connects the transmitter to the antenna. This is done with the term Lc, which
has units of dB.
The second factor that limits tag reader range is the fundamental propagation of
the electromagnetic wave through space. In the ideal case, the power radiating from an
isotropic source travels on a uniform sphere that grows in size as the wave travels away
from the RFID reader. Since the surface area of the sphere increases with distance, the
power density at any one point on the sphere will decrease with distance. This
phenomenon is described in the Friis path loss equation (2), shown here in logarithmic
form.
9
RTTR GGd
PP ++
−=λπ4
log20 (2)
In this equation, the power transmitted by the reader is PT (dBm) and the tag’s
received power is PR (dBm). The wavelength λ is measured in meters. The transmitter
and receiver antenna gains are GT (dBi) and GR (dBi), respectively. Finally, the term d
represents the distance between the tag and the reader measured in meters. If the power
to turn on a tag is known, then the use of equations (1) and (2) can be used to estimate
the range at which the tag will respond. Although not readily apparent in the logarithmic
form of the Friis equation, the power received by the tag diminishes quadraticaly with
distance.
One of the key features that separate RFID readers from traditional radio
transceivers is that the RFID reader receives backscattered signals rather the broadcasted
RF signals. The term backscatter (or alternatively “backscatter radiation”) refers to the
communication method used by a passive RFID tag, where data is encoded in the RF
energy that is reflected off of the tag’s antenna and sent back to the RFID reader.
This is analogous to how a stranded hiker would use a signal mirror to alert
rescuers to his or her location. The light from a source, such as the sun or search lights,
is reflected off of the signal mirror and bounces to the rescuers. If the hiker chooses to,
he or she can encode data in the reflected signal by moving the mirror back and forth.
Since there is potentially very little energy that is backscatter, the RFID reader must be
sensitive enough to identify and decode the signal while simultaneously transmitting a
high-power, continuous wave.
One significant challenge to designing RFID systems is identifying tags out of a
large population quickly and efficiently. This problem has largely been solved for static
ID tags with protocols developed by the RFID industry. Implementation varies from
system to system; as an example, the EPC Class-1 Generation-1 protocol used a binary
tree search method [7], while the EPC Class-1 Generation-2 uses a slotted aloha
algorithm [8]. However, it is still an open research question if either of these methods is
10
sufficient for effectively handling senor data and other higher level communication
traffic. Chapter 2.3 describes an overview of the EPC Class-1 Generation-2 protocol.
Figure 3. EPC Generation 2 RFID readers.
Figure 4. 915 MHz RFID reader patch antennas with covers.
11
Figure 5. 915 MHz RFID reader patch antennas without covers.
2.3 RFID Protocol: EPC Class 1 Gen 2
As mentioned earlier, the EPC Generation-2 protocol [8] is based on the Framed
Slotted Aloha algorithm [23], where each frame has a number of slots and each tag
responds in one randomly selected slot per frame. The number of slots in a frame is
determined by the reader and can be varied on a per frame basis. Before starting a frame,
a reader can optionally transmit a Select command which limits the number of tags
eligible to respond by providing a bit mask and a memory location, as only tags with IDs
(or memory locations) that match this mask will respond in the subsequent frame.
To begin a frame, the reader transmits a Query command which indicates the
number of slots. Upon receiving a Query, each tag randomly chooses a slot in which to
reply. If a tag chooses zero for its slot counter it responds immediately with a 16 bit
random number (RN16). The reader echoes this RN16 in an ACK command and the tag
responds with its ID. At this point, the tag is singulated. When a tag is singulated the
reader can read and write tag memory.
After singulating a tag, the reader transmits a QueryRepeat command which
indicates the end of the slot. This signals to the tag that its ID has been read successfully
and it should not respond in subsequent frames. Additionally, all other tags decrement
12
their slot counter and transmit an RN16 if their counter reaches zero. Of course, tags
may choose the same initial value for their slot counter. In this case, their transmissions
will collide, the tags will not be singulated, and they will remain selected in the next
frame. A series of frames are conducted, each with a decreasing number of selected tags,
until all tag IDs have been read. This mechanism enables the rapid identification of tags.
The Gen 2 standard specifies a memory architecture that includes banks for
storing tag configuration information and the tag identifier, as well as user memory with
a size bounded only by the device hardware. In the case of the sensor enhanced tags, this
user memory can be used to store sensor data. For instance, data for a particular sensor
could be written to a given memory location or a time series can be written to a
sequential range of memory locations.
2.4 RFID Tags
The term “RFID tag”, which is synonymous with the term “RFID transponder”,
is an often overused, catch-all term that refers to a device that performs some type of
wireless identification, and may or may not include: a multi-bit ID, wireless power
capabilities, batteries, a microcontroller, GPS receivers, Wifi transponders, sensors, etc.
Therefore, when discussing RFID tags, it is important to clearly identify what type of
device is under consideration and its inherent limitations. Generally, RFID devices can
be placed in one of two categories:
• Passive Tags - Wirelessly power by the reader (battery free)
• Active Tags - Batteries are used to power part or all of the system
13
Figure 6. EPC RFID tag class structure (source www.epcglobalinc.org).
EPCglobal, which is the organization that defined the Electronic Product Code
(EPC) specification, has further subdivided RFID tags into the sub-classes shown in
Figure 6. Class 0 and Class 1 tags contain a write once memory for storing an Electronic
Product Code identifier. Class 2 tags are loosely defined as tags with additional
functionality. This has typically meant a tag with additional memory that can be changed
frequently for the purpose of data storage. Class 3 tags add batteries for longer read
ranges and higher reliability, but still use passive backscatter modulation for
communication. Class 4 tags are active tags that can communicate with other Class 4
tags as well as readers. Class 5 tags are essentially battery powered, wirelessly
networked readers. These tags have the ability to interrogate all classes of tags,
communicate with readers, and report data directly to the host computer.
2.4.1 Passive Tags
Passive RFID tags are fixed function devices that are powered and read by a
standard RFID tag. UHF tags consist of a thin 2D printed antenna and a CMOS
Application Specific Integrated Circuit (ASIC). The absence of an onboard battery
means that the device can be quite small and is suitable to be used as a label or sticker.
14
Although their read range is limited to roughly 3-10m, the lack of batteries means that
the tag has an extremely long lifetime that can be measured in decades. Furthermore, the
simplicity in design makes them suitable for mass manufacturing and has become a
valuable method for asset tracking and access control.
Figure 7 shows a basic block diagram of an RFID tag. The design goal is to
maximize read range while providing compliance with the protocol. Read range is
primarily limited by the amount of RF power that can be transmitted by the reader, and
the rate at which this RF power is attenuated as it travels through free space. Since the
tag is entirely passive, it relies on the energy provided by the incident radiation to power
up. The electrical current induced in the antenna by the incoming radio frequency signal
is converted into DC power by the rectification block. IC tags typically use diode
connected MOS transistors configured in a multi-stage, voltage doubling configuration.
The regulation block powers the rest of the circuit once the minimum threshold of
rectified voltage has been met. The demodulator thresholds the amplitude shift key data
transmitted from the RFID reader into logical ones and zeros. This serial stream of data,
along with a clock signal from the ring oscillator, is passed into the digital state machine.
The state machine is responsible for interoperating the RFID protocol and responding
with its ID when directed. RFID chips usually contain non-volatile memory, so that a
unique ID can be programmed after fabrication.
When it is the tag’s turn to transmit data, the RFID reader will broadcast a
continuous RF waveform. The tag will then modulate the impedance of its antenna
between two states using a RF transistor. This impedance mismatch will cause a
reflection of RF energy which will travel back to the RFID reader. Since the RFID tag
only has to toggle a transistor to encode data, the power requirement and complexity of
the design is far less then that of traditional radios. Conversely, since the signal sent to
the reader is simply a reflection, there is potentially little signal strength available for the
reader to decode. Thus, in order to have inexpensive, long range RFID tags, much of the
burden of RF communication has been transferred to the reader, which must be able to
detect and decode the backscatter signal.
15
Rectifier
Demodulator
Modulator
Regulator
& POR
Digital
Logic
/
State
Machine
Ring
Oscillator
Figure 7. Block Diagram of a typical commercial RFID tag.
Figure 8 and Figure 9 show modern EPC Class 1 Gen2 RFID tags. The tags
shown consist of printed copper antennas bonded to a chip. Modern UHF tags, such as
the Atmel ATA5590 [1], consume less then 12 µW of power when active and have a
read range of 10 meters.
Figure 8. Alien RFID tag, Model ALN-9540 "Squiggle"
(antenna with label backing is 10mm x 1.5mm).
16
Figure 9. Alien RFID tag, Model: ALL-9460 "Omni-Squiggle"
(antenna with label backing is 7.6mm x 7.6mm).
2.4.2 Active Tags
Active RFID tags run the gamut from simple battery assisted identification
devices to full embedded systems, such as wireless sensor network nodes. The defining
factor is that the addition of a battery makes it an “active” device, which allows the
possibility for extra functionality.
If you recall, for passive RFID tags, the primary engineering challenge is to
minimize the power consumption, thus increasing range. In contrast, the addition of a
power source changes the design constraints and allows designers to focus on specific
applications. Often times active RFID tags do not need to have any custom silicon.
Instead, they are constructed of multiple printed circuit boards and can be quite large in
size.
Active RFID tags generally fall into three categories: identification with
increased reliability, sensing and monitoring, and localization. As is the case with any
RF device, interference and multi-path effects can cause significant performance
problems. Thus, one of the advantages when considering active tags is increased
17
readability in difficult RF environments where metal, liquids, or RF noise is prevalent.
Since active tags can transmit a signal rather then simply using backscatter modulation,
they generally have much longer read ranges with higher reliability.
As active RFID tags transition into embedded RFID devices, additional features
such as microcontrollers, buttons, sensors, and large data storage modules are often
added. Since they are battery powered, active tags can have much higher activity and
data transmission rates. Additionally, for applications that need real-time localization,
there are now tags that come equipped with WIFI and GPS [11][6].
Unfortunately, active tags can cost $20 to more than $150 each. Users are often
forced into using a proprietary protocol that does not integrate into the well established
EPC passive tag framework. Finally, the need for batteries limits the lifetime for active
tags to a few years and restricts them from being embedded into objects.
2.5 State of the Art - Sensor Enhanced RFID Tags
Industrial efforts in the development of RFID technology have produced a robust
physical layer, capable of wirelessly powering and querying a tag. This core technology
enables a new class of wireless, battery-free devices with communication, sensing,
computation, and data storage capabilities. Unconstrained by batteries, these devices
have the potential to operate for years, if not decades.
2.5.1 Prior Work
The authors in [21] present network architectures for an RFID-enhanced
environment where objects are seamlessly tracked and monitored. The implementation
of an environment augmented with RFID to enhance the quality of life and independence
of elderly citizens is discussed in [19]. In this example, participants wear small RFID
reader bracelets that report interaction with tagged objects. Activities can be inferred
from this data and reported to caregivers. Specific applications for sensor-enhanced
RFID tags are identified in [30] and include infrastructure and object monitoring,
18
automatic product tamper detection, identification of harmful agents, and biomedical
devices for noninvasive monitoring.
Conventional RFID applications are also benefiting from sensor-enhanced RFID
tags. A commercially available RFID tag for detecting dangerous temperatures in food
products during transit is reported in [30]. This product suggests the possibility that
although the price of RFID tags may not decrease sufficiently for all potential
applications, sensor-enhanced tags may provide a substantial increase in functionality for
the same price as conventional RFID tags.
To date, there are several approaches for incorporating sensing capabilities into
RFID. Active tags, a subclass of RFID tags, use batteries to power their communication
circuitry, sensors, and microcontroller. Active tags benefit from relatively long wireless
range (approximately 30 m) and can achieve high data and sensor activity rates.
However, the batteries required by active tags are disadvantageous for device cost,
lifetime, weight, and volume.
In contrast, passive sensor tags receive all of their operating power from an RFID
reader and are not limited by battery life. There are several examples of application-
specific, non-programmable passive tags with integrated temperature and light sensors,
as well as an Analog to Digital Converter (ADC) [4][16]. One attractive feature of
passive sensor tags is the prospect of permanently embedding them in objects for
structural, medical, or product monitoring. Another advantage is their suitability for
applications in which neither batteries nor wired connections are feasible, due to weight,
volume, cost, or other reasons. One limitation of purely passive sensor tags is the
required proximity to an RFID reader. However, other methods such as solar, thermal, or
kinetic energy harvesting could be used as a secondary power source, if needed.
A further consideration is the configurability and computational power of RFID
sensor tags. Existing devices are generally fixed-function with respect to sensory inputs
and they lack computational capabilities. A commercially available RFID tag with some
additional functionality is described in [18]; however, this device can only transmit one
bit of sensor data in addition to its ID. Furthermore, it is limited by a short read range,
due to its 125 kHz operating frequency.
19
2.5.2 Prior Work on WISP
The general-purpose WISP described in this thesis, was preceded by several less
capable devices (also called WISPs) that were described in earlier publications by our
group. The first venture into sensor-enhanced RFID was the α-WISP shown in Figure 10
and published in [20]. With this device, one bit of sensor data was encoded by using
anti-parallel tilt switches to multiplex one of two RFID tag ICs to a single antenna.
Thus, a reader could infer three states about a tagged item (tag right side up, upside
down, or not present). This simple example of overloading the EPC ID to encode sensor
data allowed inference of very coarse orientation information. However, the use of
commercial RFID tag ICs restricted our ability to control the RFID communication
channel and in turn our ability to configure WISPs for new applications.
Figure 10 The α-WISP uses two tilt switches orientated in opposite directions as a simple one bit RFID
accelerometer.
The π-WISP [27] used a microcontroller powered by harvested RF power to
activate a GaAs RF switch, which multiplexed two commercially available RFID ICs
into one tag antenna. This device could transmit at most one bit of sensor data per query,
and used two separate antennas for communication and power harvesting. The
significant difference between the previous work and the WISP presented in this paper,
is that the microcontroller is now implementing the EPC protocol and no commercially
available RFID ICs are used in the design. This gives the WISP the ability to control all
20
64 bits of the ID for data encoding, versus 1 effective bit for the previous approaches
based on enabling and disabling commercial tag ICs. Furthermore, the device described
in this thesis uses a single antenna for power harvesting and communication, while the
approach of [27] required separate antennas for these two functions.
21
Chapter 3 Wireless Sensing & Identification Platform
The WISP is manufactured as a printed circuit board (PCB), which offers a
number of benefits when compared to traditional Integrated Circuit (IC) tag designs. A
few of these advantages include low development cost, fast design cycles, and easy
debugging and measurement of circuit parameters. The PCB implementation allows the
flexibility to physically add and remove sensors and/or peripherals to create devices for
new applications. In contrast, IC implementations offer the ability to customize
components and decrease power consumption (yielding better range), as well as creating
devices with a smaller form factor and at a lower cost when manufactured in high
volume.
A block diagram of the WISP is shown in Figure 11 and is similar in function to
traditional IC RFID tags. The antenna is balanced by an impedance matching network
and is fed into the RF power harvester. The Radio Frequency (RF) signal transmitted by
the RFID readers is rectified into DC voltage to power the rest of the tag. The
demodulator block converts the Amplitude Shift Keyed (ASK) data that is superimposed
on the RF carrier into a logic level stream of serial data. This extracted serial data is
parsed by the MSP430 microcontroller (MCU) to receive downlink data from the reader.
Uplink data is sent via the modulator circuit, which “back-scatters” the signal by
changing the antenna impedance. Finally, the microcontroller’s internal temperature
sensor, as well as any external sensors, are powered and measured by the MCU.
Since the power consumption of the microcontroller, sensors, and peripherals are
much greater then that seen in traditional passive RFID technology, the WISP duty
cycles between active and sleep mode. In sleep mode, the WISP shuts down and reduces
its current consumption to a few micro-amps and energy is accumulated by the
harvesting RF power over multiple EPC queries. Once sufficient voltage is obtained, the
WISP polls sensors and communicates with the RFID reader.
22
TI MPS430
Microcontroller
Flash
Memory
Temperature
Sensor
Modulator
Power
Harvester
Impedance Matching
Demodulator
Power
Management
Sensors and
Peripherals
Figure 11. Block Diagram of the WISP.
Figure 12 depicts the WISP platform, made of a four layer FR4 PCB with
components on both sides and an integrated dipole antenna. The WISP in its base
configuration has several onboard sensors: a circuit for measuring the rectified supply
voltage, a temperature sensor, and a 3D accelerometer. Small header pins expose all
ports of the microcontroller for expansion to daughter boards, external sensors, and
peripherals. Finally, a low current surface mount LED is included in the design. Figure
13 shows the data logger version of the WISP which has additional features, such as a
larger microcontroller, a real time clock, external EEPROM, and an optional 0.1 Farad
super capacitor for extended lifetime.
Figure 12. Wireless Identification and Sensing Platform (WISP).
23
Figure 13. Data logger version of the Wireless Identification and Sensing Platform.
3.1 Analog Front End
The defining characteristic of far field RFID systems is that tags can be read at a
significant distance, generally on the order of 2-10 meters. For passive RFID, this
requires that the RFID reader transmits sufficient energy to power the tag at large
distances. However, due to regulatory limits on the amount of power that can be
transmitted and the path loss associated with electromagnetic propagation, there is very
little power that actually reaches the tags. Therefore, the power harvesting circuit must
maximize the operating distance by converting the very limited incoming RF power to
DC power with sufficient voltage to activate the tag.
The RF power received by the WISP’s dipole antenna is fed to the analog front
end depicted in Figure 14. A discrete matching network is used to provide the maximum
power transfer from the antenna to the rectifier. RF Schottky diodes, specifically
designed for 915MHz low power application, were selected to make a five-stage voltage
doubling circuit. This circuit converts the AC input signal to DC power which is fed into
a storage capacitor.
24
RF Rectifier
Rectified DC Power
Figure 14. Schematic of the Analog Front End.
For RF rectifiers of this type, the input and output impedances are not well
isolated. Further confounding the problem, the output impedance of the rectifier is fairly
high; an undesirable trait for any power source. This means that as the load on the
rectifier changes the input impedance also changes, resulting in the analog front end
becoming mismatched to the antenna. This leads to the problem of selecting values for
the impedance matching network when it is not possible to guarantee constant input
impedance.
To determine the correct values for the matching network the operating cycle of
the WISP must be taken into account. First, the WISP is most effective at storing
harvested energy when it is in sleep mode, as the current consumption is minimal.
Second, the WISP will spend most of its time repeatedly charging up to 1.9v and then
discharging to approximately 1.8v. Thus, to determine the correct values, the WISP is
placed into sleep mode and the impedance matching network is swept with a variable
capacitor until 1.9v is produced for the lowest possible input power. Stated another way,
the key parameter for maximizing the read distance of the WISP is minimizing the
quiescent current consumption so that the minimum operating voltage of 1.9V
(supervisor threshold) can be rectified with the lowest possible input power.
25
3.2 Demodulation and Modulation
The EPC Gen 2 standard defines that reader-to-tag communication uses ASK
modulation on a carrier wave in the range of 902-928 MHz. When not transmitting data,
the carrier waveform remains at a constant amplitude; when bits are transmitted, the
amplitude of the carrier drops to at least ten percent of its normal value and the phase of
the carrier may be reversed. The duration of the continuous waveform between these low
amplitude pulses indicates logical “ones” or “zeros.”
Figure 14 shows a schematic of the WISP’s demodulator circuit. The output of
the harvester is fed through the diode, which supplies power to the comparator and acts
as a reference for the level shifter. A capacitor is used to filter out transients while
allowing proper biasing at varying distance and received power levels. When activated,
the current consumption of the comparator functions as a constant-current source,
pulling current through the diode. In this way, the voltage drop across the diode is used
as a detector, where current supplied by the harvester (high amplitude RF modulation)
results in positive voltage, and a lack of current (low amplitude RF modulation) yields
negative voltage. The comparator is used to generate a rail-to-rail logic level waveform,
and the level shifter converts the unregulated logic level to the regulated logic level. It is
important to optimize current consumption and speed when choosing a comparator.
Further savings can be achieved by disabling the comparator when there is insufficient
voltage to start up the MSP430.
An example of a demodulated signal is shown in Figure 15. This oscilloscope
plot shows the 915MHz RFID waveform and the resulting demodulated signal. Note that
time frame is 20µs per division and thus, the individual cycles of the 915MHz carrier are
not visible. However, the ASK modeled data is visible as gaps in the carrier and
enveloped signal.
RFID tags do not actively transmit radio signals. Instead, they modulate the
impedance of their antenna which causes a change in the amount of energy reflected
back to the reader. This modulated reflection is typically called backscatter radiation. In
order to change the impedance of the antenna, a transistor is placed between the two
branches of the dipole antenna. When the transistor conducts current, it short-circuits the
26
two branches of the antenna, changing the antenna impedance. In the non-conducting
state, the transistor has no effect on the antenna and thus, the power harvesting and data
downlink functions occur as if it were not present. This impedance modulation is
currently implemented with a 5 GHz RF bipolar junction transistor, which allows for
effective shunting of the 915 MHz carrier wave.
Figure 15. Oscilloscope plot of the demodulated data extracting data from the RF waveform transmitted
by the RFID reader.
3.3 Digital Section and Power Conditioning
Since the power available to RFID tags is extremely limited, careful component
selection must be made to minimize current consumption. As advances in IC
manufacturing now allow discrete components with less than 1 µA of current
27
consumption and operation at 1.8 V, it is now possible to construct working, wirelessly
powered RFID tags with discrete components.
The general purpose computation capabilities of WISP are provided by an ultra-
low power microcontroller. This 16-bit flash microcontroller, the MSP430F1232, can
run at up to 4 MHz with a 1.8 V supply voltage and consumes approximately 600 µA
when active at those frequency and voltage settings. Of particular interest for low power
RFID applications, the MSP430 has various low power modes. Its minimum RAM-
retention supply current is only 0.1 µA at 1.5 V. The device provides over 8 kilobytes of
flash memory, 256 bytes of RAM, and a 10-bit, 200 kilo-samples-per-second Analog to
Digital Converter (ADC). The low power consumption of this relatively new device is a
critical factor in enabling use of a general purpose microcontroller in passive RFID
systems.
Another critical design consideration is operation with uncertain power supply
conditions. Because the available RF power varies greatly throughout device operation,
supervisory circuitry is necessary to wake and sleep the device based on the supply
voltage level. WISP uses a 1.9 V supervisor and a 1.6 V power-on-reset to control
device state and reset the microcontroller, respectively. The supervisor provides roughly
100 mV of headroom on the storage capacitor above the 1.8 V of regulator voltage. This
serves to buffer the supply voltage from dropping below 1.8 V, due to the large power
consumption of the microcontroller in active mode.
28
Chapter 4 Firmware & Power Management Algorithm
The WISP is essentially a software defined RFID tag, which uses the MSP430 to
implement the EPC Class-1 Generation-2 protocol and performs sensing and
computation tasks. There are significant challenges when developing applications on the
WISP as compared to battery powered embedded systems. Primarily, there is no
guarantee that a given task can be completed before running out of power. Although the
voltage supervisor provides headroom above 1.8 V, the rate at which the energy stored
in the supply capacitor is consumed is directly affected by the design choices of the
programmer. Failure to properly manage sleep cycles when the WISP harvests energy or
inefficient coding practices can result poor performance.
The WISP software can be described on three levels. At the lowest level is the
power management algorithm, which is responsible for managing the device state,
including sleep vs. active modes. Built on that is the communication layer, which
enables bi-directional communication by sampling downlink data bits, implementing a
Gen 2 state machine, and generating uplink data bits. The third level is the application
layer where users implement custom function and encoding data in the appropriate EPC
packets.
4.1 Power Management Algorithm
Meeting the low power requirements of passive RFID tags requires that the MCU
consumes, on average, as little power as possible. As mentioned previously, this is
achieved by duty cycling between active and low power sleep states. The key is that the
WISP receives a constant amount of power as defined by Friis’ path loss equation 1 for a
set distance. When the WISP is in active mode the power consumption far exceeds the
power harvested. However, when the WISP is in sleep mode, the total current
consumption of all the circuits is a few micro-amps and there is a net power gain which
charges the storage capacitor. Therefore, duty cycling does not simply yield lower power
consumption; it represents two different states, power harvesting and active operation.
29
Figure 16. Block diagram of the power management algorithm for the WISP.
The state diagram for the power management layer is shown in Figure 16. State
transitions are primarily driven by hardware interrupts from the voltage supervisor,
which indicate if there is sufficient energy stored for operation. Initially, the WISP is
away from a RFID reader and is in a power down state. When the WISP is brought
within range of a reader, it begins to harvest power and the voltage across the storage
capacitors begins to rise. At approximately 1.6 V the MSP430 powers up in a reset state
and begins executing code. Since this event is not driven by the supervisor, it is
important that the code enters sleep mode (LMP4) as quickly as possible in order to
repeatedly avoid browning out on start up. Once in LMP4, the WISP waits for sufficient
voltage (1.9 V), as indicted by the supervisor interrupt. Next, the state machine
transitions to the application layer, which performs user defined functions, such as
sensor measurements. Here, an EPC packet is generated and the WISP sets up and waits
for a commutation interpret which indicates the beginning of an EPC packet. In the
communication layer, the WISP processes the incoming data, executes the EPC Gen 2
protocol, and transmits its response. While not shown in Figure 16, the communication
layer often reports the same data twice to increase communication reliability.
30
4.2 Communication and Application Layers
A considerable challenge when programming the MSP430 involves meeting the
timing constraints of the EPC protocol while still maintaining a low clock frequency.
RFID tags that have custom state machines are designed at the hardware level to receive
and send using the EPC protocol. The general-purpose MSP430 must be carefully tuned
to perform EPC communication, both for receiving and transmitting data. In particular, a
mix of C and assembly language is used where the C code maintains ease of
configurability for the firmware for different sensor applications and the assembly code
allows fine-grained control of the timing of the MSP430 for EPC communication.
As previously described, the demodulator envelops and thresholds the Phase-
Reversed Amplitude Shift Keyed (PR-ASK) signal from the reader into a serial data
stream representing the data bits 1 and 0 as long and short pulses, respectively. To
interpret data from the reader, the MSP430 uses the periodic edge of the waveform as a
hardware interrupt, and then during the interrupt service routine re-samples the bit line to
detect a 1 or 0 during the differentiated part of the waveform. This data is quickly shifted
into memory before repeating this process. To detect the end of a transmission, a timer is
refreshed during each bit. When bits are no longer received the timer expires, the packet
is interpreted and, if appropriate, a response is sent to the reader. A detailed description
of how the WISP uses and implements the EPC specification is described in 2.3.
Figure 17 shows a set of EPC queries and responses along with the
charge/discharge cycle of the WISP. Since the operating voltage range of the WISP
occurs between 1.9v-1.8v the rectified voltage appears to be nearly constant. In actuality,
the WISP enters active mode at 1.9v, consumes the energy in the storage capacitor until
approximately 1.8v, then enters a sleep state and harvests power until 1.9v is reached.
This duty cycling can be seen in the packet transmitted plot. Here, the WISP does not
respond to every packet sent by the reader, instead it spends most of its time in a sleep
state.
31
Figure 17. Oscilloscope scope plot of the WISP responding to EPC queries along with its rectified
voltage.
Performing application level tasks, such as sensor measurement, is generally
done in tight conjunction with the EPC protocol. In this scenario, the completion of a
receive/transmit cycle triggers the application layer to immediately take a sensor
measurement, generate the desired EPC packet, and setup for a Query. This protocol
centric approach works well for sensor driven applications where data is requested from
the RFID tag at regular intervals. However, applications which leverage the wirelessly
powered computing capability of the WISP benefit from a loose coupling with the
communication layer.
32
Chapter 5 Power Budget
One of the significant challenges of incorporating microcontrollers, sensors, and
peripherals into passive RFID technology is the ability to manage the large power
consumption of these devices. For example, the MSP430F1232 running at 3 MHz
consumes approximately 470 µA at 1.8 V. The resulting power consumption is
significantly larger then typical passive RFID tags. Under these conditions the harvester
cannot continuously supply power to the WISP during a single reader query.
One method to overcome this challenge is to use a large storage capacitor (on the
order of ten microfarads) to accumulate charge over multiple EPC queries. Once
sufficient voltage is obtained, the WISP can operate in a burst mode, polling sensors and
communicating with the RFID reader. This approach of duty cycling is often used in low
power applications; however, this presents a challenge for RFID networks when the
WISP is not necessarily able to respond to each reader query.
The next section examines the issues related to powering the WISP from three
perspectives. First is the received RF power required to turn on the device, the second is
the operating duty cycle based on input power, and the last is the energy needed in the
storage capacitor for active operation of the microcontroller and additional sensors.
5.1 Turn On Power Requirement
In the presence of the RFID reader, the WISP’s RF rectifier will charge the
storage capacitor until the power input to the device equals the power lost due to
quiescent current.
in loss rectified lossP P V I= ≡ × (3)
Thus, a key parameter for maximizing the read distance of WISP is minimizing
the quiescent current consumption so that the minimum turn on voltage of 1.9V
(supervisor threshold) can be rectified with the lowest possible input power. In order to
33
characterize the system, a network analyzer was used to inject a continuous 915 MHz
waveform into the antenna ports of the WISP. Figure 18 shows the resulting plot of
rectified voltage and output power vs. input power when the WISP is in sleep mode.
Rectified voltage was measured with the WISP in sleep mode (only quiescent current
draw), and shows the minimum input power needed to start operation. After the 1.9 V
supervisor threshold has been met, the rectified voltage continues to increase with input
power, until the over-voltage protection diode activates at 5.4 V. In the actual
implementation of the WISP, the MPS430 activates at 1.9 V and starts consuming
power. Thus, the rectified voltage never rises above the supervisor threshold.
Using the minimum input power needed for activation from Figure 18, the
expected operating distance for the WISP can be calculated with the logarithmic form of
the Friis equation (3) for path loss, with a term for polarization loss included.
PRTTR LGGd
PP −++
−=λπ4
log20 (4)
The transmit power of the reader PT = 30 dBm (which is equivalent to 1 Watt).
Its center frequency is 915 MHz, corresponding to wavelength λ = 0.33m. The transmit
antenna gain GT = 6 dBi (this yields an effective isotropic radiated power of 4 WEIRP,
the United States’ regulatory limit for this ISM band). The receive antenna gain GR = 2
dBi (the standard gain figure for a dipole antenna), and the polarization loss LP = 3dB.
Lossed LP occurs because only half of the power transmitted from the circularly-
polarized transmit antenna is received by the linearly-polarized receive dipole antenna.
Using the operating thresholds of -9.5 dBm from Figure 18, equation (4) predicts a
maximum operational range of 4.3m.
34
Figure 18. Rectified voltage (left scale) and output power at 1.9V (right scale) are plotted verse input
power measurements using multi-meter and network analyzer for RF signal insertion into the antenna
ports.
5.2 Duty Cycle
While rectified voltage (rather than power) determines the maximum achievable
range, the operational duty cycle (percentage of the time WISP can be active), is
determined by the amount of rectified power. In practice, the rectified voltage will
typically remain near the threshold voltage (1.9 V). This is due to the operation of the
supervisor, which transitions the WISP from sleep to active mode, resulting in the
consumption of power whenever the stored voltage exceeds this operating point.
Therefore, it is important to characterize the output power of the harvester at 1.9 V.
Figure 18 shows the result of output power verse input power at 1.9 V. This is
accomplished by fixing the output voltage at 1.9 V using a power supply and measuring
35
the amount of current that is supplied by the WISP. Then, the duty cycle of WISP
(percentage of the time in active mode) is estimated as the ratio of rectifier output power
to WISP active power consumption.
cycleDutyTT
T
P
P
sleepon
on
active
out =+
= (5)
In this equation, Pout is the output power of the WISP, Pactive is the active power
consumption, Ton is the time in active mode, and Tsleep is the time in sleep mode. For
example, the power rectified at 0 dBm (310 µW) divided by the active power
consumption (1.8 V * 600 µA = 1.12 mW) yields a duty cycle of 27%. This agrees well
with experimental values, which are presented in Chapter 7.
5.3 Active Energy Consumption
Since the rectifier cannot supply enough power for continuous operation, it is
important to quantify the amount of energy that needs to be stored in order to power the
WISP during active periods. During one EPC Gen 2 communication cycle, the complete
WISP (not just the microcontroller) consumes on average 600 µA * 1.8 V = 1.08 mW. A
single query takes 2 ms including reader and tag communication. Using the expression
for the energy stored in a capacitor (E=½CV2, with C=10 µF), the amount of voltage
headroom needed above 1.8 V is 116 mV, resulting in a total minimum voltage threshold
of 1.91 V for a complete packet transmission. It should be noted that the MSP430 will
operate down to 1.7 V, even though this value is below the specified supply voltage.
However, operation is not guaranteed; it has been observed that the Digitally Controlled
Oscillator (DCO) can begin to slow down. Thus, it is not recommend that the designer
rely on the extra 100 mV of headroom below 1.8v. In the case of the previous example,
the use of 16 mV out of specification headroom (1.90 mV - 116 mV) has proven to give
reliable results.
The same method for calculating the required stored energy can be used when
selecting sensors for the WISP platform. Sensor tasks and packet generation are
generally done prior to the EPC query. However, it is reasonable to assume that when
36
performing sensor applications the MCU will exhibit similar current consumption.
Inequality (6) expresses an energy feasibility condition for a particular sensor; the
energy required to read the sensor must not exceed the usable stored energy. This
expression can be used to calculate the capacitor size and voltage headroom required to
operate a particular sensor, which in turn determines the range at which the sensor can
be operated.
( ) ( )22
2
1ddrecWSdd VVCTIIV −≤+ (6)
The current consumption for the sensor and WISP are Is and Iw, respectively; C is
the capacitance of the storage capacitor and T is the total time of active operation. The
rectified voltage is Vrec and Vdd is the required operating voltage. Assuming that the
sensor has the same voltage supply as the WISP, Vdd = 1.8 V. The left hand side of
inequality (4) represents energy consumed by the sensor and WISP during one
measurement. The right hand side represents usable stored energy above Vdd, the
minimum operating voltage of WISP. Inequality (6) makes it clear that the limiting
factor when selecting sensors is not only the current consumption (which determines
power) but, also the total required execution time of the sensor and WISP (energy, rather
than power).
37
Chapter 6 Communicating with the WISP
The EPC Class-1 Generation-2 protocol (henceforth referred to as Gen 2) was
designed to rapidly identify tags with static IDs. However, when implementing sensing
applications with the WISP, it is necessary to transmit and receive higher order data. The
Gen 2 protocol provides several mechanisms that can be used to implement a two way
communication layer. First, the WISP can send uplink data by overloading the identifier
and by using Gen 2 Read command. Secondly, applications that need to transmit data to
the WISP (e.g. to actuate its behavior), can use the Select command and the Write
command.
6.1 Overloading the Identifier
The Gen 2 protocol efficiently reads tag identifiers and by overloading the
identifier to include sensor data, a collection of WISPs can also report data efficiently. In
our initial applications, the identifier was replaced with the sensor data of interest.
However, when using more than one WISP, data from different devices cannot be
differentiated. Additionally, this approach breaks the semantics of the protocol and limits
the interoperability of WISPs and standard tags.
The Gen 2 specification allows for the transmission of up to 496 bits of identifier,
while current tags generally have an identifier of only 96 bits. Hence, up to 400 bits of
sensor data can be piggybacked along with the ID, enabling data from different devices
to be differentiated while at least partially maintaining the original semantics.
Unfortunately, by sending sensor data along with the identifier the read time per tag is
increased and the time required to read data from a particular tag can be prohibitively
high. For many sensing applications, particularly those that use a large number of
devices, reading all sensor data from every tag will be undesirable and overloading the
identifier may be insufficient to meet the application requirements.
38
6.2 Gen 2 Read Command
After singulating a tag, the reader can issue a series of Read commands to read
the contents of tag memory, with each command eliciting up to 512 bytes of data. Before
issuing a Read command, the reader requests a temporary, random 16 bit handle from
the tag. This handle is used in the Read command to address the tag, and an arbitrary
number of Read operations can be issued in sequence. Using this mechanism, a reader
can selectively read sensor data stored in the user memory of a single WISP.
Using the Read command to gather sensor data has drawbacks with respect to
efficiency and flexibility. First, to read new data from a WISP the device must again be
singulated and a new handle must be obtained. With a large number of tags, singulation
is time intensive. Even in the best case, where a single device is selected with the Select
command, the singulation process must still be conducted, albeit with only a single tag
responding and a new handle must be obtained; only then can data be read from the
WISP. This results in a large amount of the WISP’s active time being spent on protocol
overhead. Additionally, the identifier of the device with the desired data must be known
prior to the read event, along with detailed knowledge of the memory layout with respect
to sensor data location.
By overloading the identifier or using the Read command, basic sensing
applications can be implemented using the WISP. However, when deciding which
technique to use, the energy cost must also be considered. Specifically, using the Read
command consumes more energy than returning the data with the ID. This presents a
trade-off between range and speed, with the proper balance being largely application
specific.
6.3 Gen 2 Select Command
The Select command is intended to limit the number of tags that respond in a
Query round. For example, a collection of retail items may have identifiers that indicate
their model number and the Select command can be used to inventory only items of a
given model by providing a memory pointer and bitmask which matches only that
39
model. However, this mechanism can be repurposed to function as a general purpose
broadcast channel, with the pointer and mask being interpreted by the WISP software as
opcodes and data. As an example, we have implemented software for the WISP which
interprets Select commands as instructions to blink LEDs.
6.4 Gen 2 Write Command
Along with the general purpose broadcast facility of the Select command, the
Gen 2 Write command can be used for unicast down-link communication. After a tag is
singulated, the reader can write arbitrary memory locations on the tag in 2 byte words.
Additionally, the BlockWrite command can be used to write up to 256 words at a time.
This mechanism can be used to transfer data to the WISP; for example, to store location
information on the tag as it moves through a supply chain. Additionally, a WISP could
be programmed to look to certain memory locations for parameters that affect its
operation. For example, to modify the sampling rate of the WISP, the Write command
could be used to transmit the desired rate to a known memory location and the WISP
would refer to this value when setting its sampling rate.
40
Chapter 7 Experimental Results
Figure 19 shows experimental results of the WISP performance: rectified output
voltage, tag responses per reader query, and the rate of tag-to-reader packet errors are
plotted vs. received power (dBm). The experimental set up consisted of an EPC Gen 1
RFID reader driving a 6 dBi circularly-polarized patch antenna. The reader’s antenna
and WISP were placed one meter apart and one meter above the ground to minimize
multipath effects. An adjustable attenuator inserted between the reader and its antenna
was used to vary the power transmitted to the WISP.
Finally, equation (4) is used to calculate the path loss over the one meter
separation between the WISP and RFID reader. Thus, the WISP received power is
defined as reader transmit power (1 Watt), minus variable attenuator, minus transmission
path loss. It should be noted that the 1 watt source represents peak output power of the
RFID reader, while the average output power (not considered here) is highly dependent
on reader transmission rate and the specific implementation of the EPC Gen 1 protocol.
To measure rectified Output Voltage, the WISP is placed in its low power state
and voltage is averaged over a ten second interval using an oscilloscope. This is
necessary to account for the variation in output power as the reader implements the EPC
protocol. The resulting plot shows the WISP turns on with a peak received power level
of -5.9 dBm, which is significantly more than the average power level of -9.5 dBm
measured with the network analyzer in Figure 19. In order to verify that this difference
in turn-on threshold is caused by lower average power in the experimental setup, the
RFID reader was replaced with a 915MHz, 1 W continuous wave source and the turn on
power was found to be -8.7 dBm. The 0.8 dBm difference between the continuous
source and the network analyzer is thought to be due to impedance mismatch between
the dipole and the analog front end of the WISP as well as antenna non-idealities. The
2.8 dBm difference between the continuous source and the RFID reader is then due to
lower average power output by the reader.
The plot of tag Responses per Query shows the number of successful tag
responses received by the reader normalized over the total number of queries made. This
41
is equivalent to the operating duty cycle of the WISP and, as expected, is proportional to
received power. The response rate drops to zero at -7 dB because there is insufficient
voltage for operation. At 0 dBm input power, section IV.B predicted an operational duty
cycle of 27% using equation (3), which is close to the experimental value of 25% from
Figure 19. The reason that duty cycle (unlike turn-on voltage) is not diminished by the
lower average power of the RFID reader is because duty cycle is normalized to the query
rate of the reader. In other words, Responses per Query excludes times in which the
reader is not transmitting.
The Uplink Packet Error represents the percent of query responses made by the
tag that are not correctly received by the RFID reader. Due to the limited data interface
with the RFID reader selected for the experiment, the number of reader rejected uplink
packets is not directly available. To collect this data, the WISP counts the number of
query responses it has made and reports the current tally as data encoded in each uplink
packet. When the RFID reader application software receives gaps in the running tag
response tally an error is recorded. Figure 19 shows that as received power decreases to
the point at which sufficient voltage can no longer be rectified for operation, the uplink
packet error rate increases. It is theorized that this system instability is due to the brown
out state of the MPS430, along with the ring oscillator, used as the system clock,
becoming detuned as the 1.8 V regulator drops out.
42
Figure 19. WISP performance: harvested voltage, Uplink Packet Errors, and Responses per Query as a
function of input power.
43
Chapter 8 Sensors and Peripherals
Several types of sensors have been successfully integrated into the WISP
platform: light, temperature, push-buttons, rectified voltage, and 3D acceleration. As a
rule of thumb, sensors that operate as resistive transducers are good candidates for the
WISP. These devices typically require a small amount of current and when placed in a
voltage divider configuration they can be easily be measured with the MSP430’s Analog
to Digital Converter (ADC). One example is the measurement of rectified voltage, which
is easily accomplished with a voltage divider that scales the signal down to the range of
the 1.8 V ADC. Alternatively, active sensors are much more demanding in terms of
current consumption and required operating voltage.
Real-world environmental noise often requires band-limiting the sensor signal to
a few kilohertz or less. Such a filter may have a long time-constant. Energy
considerations should also be examined using equation (6) to choose a sufficient
capacitor size for the system. For example, powering a 500µA sensor for 10 ms requires
a great deal more energy than RFID communication; a 50µF capacitor charged to 1.9 V
would be needed to provide enough energy to measure the sensor.
In order to enable environmental monitoring applications, the WISP was
enhanced with temperature sensors. The MSP430 does have an on-chip temperature
sensor. However, its accuracy and power consumption is poor compared to signal chip
solutions and the LM94021 low power analog temperature sensor from National
Semiconductor was added. Figure 20 shows a time series of temperature measurements
reported by WISP using the external temperature sensor. An inverted can of compressed
air was used to generate a low temperature impulse. After the WISP’s temperature
sensor had recovered for about 30s, a heat gun was used to generate a high temperature
impulse. The LM94021 has 1.8 °C accuracy and a Fluke thermal probe was used as
reference for the system. The maximum error recorded between -20 °C and 50 °C by the
WISP was 2 °C.
44
Figure 20. Cold impulses are applied to WISP and a Fluke thermal probe and plotted over time.
Another example of environmental sensing has been demonstrated using the
WISP with a photo resistor as a light sensor. The WISP was mounted by suction cup to
the inside surface of an exterior window in an office environment, with the sensor
oriented inward. Figure 21 shows the light level measured by the WISP over a 13 hour
period. The measurement began at 5:20 pm and the initial part of the measurement
shows the light level dropping due to sunset. After sunset, the light level remains
constant for about three hours. We believe that all the interior lights were on during this
period. Then the interior lights nearest the WISP were extinguished, with lights further
away remaining on for about one hour. Next, the light level dropped to a minimum
45
when all the remaining interior lights were extinguished. This low light level persists
until sunrise begins the next morning.
Figure 21. Light level measured by WISP in a 13 hour period
Figure 22 shows data collected using the three dimensional accelerometer on the
WISP. The Analog Devices ADXL330 MEMS accelerometer draws 200uA at 1.8V. Due
to the relatively high current consumption of these devices, continuously powering them
would cripple the range of WISP. To overcome these high power requirements, the
sensor is only powered for a short period of time, just long enough to take a
measurement. Provided that the sensor and conditioning electronics can stabilize
sufficiently quickly, this allows for a wide range of sensors to be measured over UHF
RFID. Powering this accelerometer, the WISP is able to provide accelerometer
measurements at rates of approximately 10 to 20 samples per second, depending on
46
range. After the measurement is taken and the data packed into the EPC ID, the WISP
calculates the correct CRC. Then the “ID” is reported to the RFID reader and the
information is then decoded in real time by the computer. Although the WISP’s
accelerometer data rate is too low for many industrial applications, it provides good
absolute orientation data that is suitable for gaming and input devices.
-1.5 -1 -0.5 0 0.5 1 1.5-1.5
-1
-0.5
0
0.5
1
1.5
X acceleration (G)
Y acceleration (G)
Wirelessly-powered acceleration measurements
collected while WISP is rotated with respect to gravity
Figure 22. Received measurement of the acceleration of earths gravity long two axes using the WISP
enabled with a 3D accelerometer.
There are many examples that demonstrated novel uses of passive RFID
technology, which benefit from wireless, battery-free operation. However, these
systems are inherently limited by the requirement of tag proximity to a reader for power
47
and finite wireless range, due to RF path loss over distance. One particularly interesting
class of applications involves a tagged item that travels between two reader-equipped
locations but does not have reader proximity during transit. For example, this situation
occurs during cold chain transport of food and chemicals between warehouses. One may
be interested in tracking the temperature or vibration of goods during transit where there
is no reader coverage.
To enable these applications, the authors in [31] have proposed a new tag device
called a passive data logger (PDL). A PDL is a battery-free RFID tag with a large
capacitor for energy storage. The PDL seamlessly recharges its capacitor when it is near
a reader and uses the stored energy to measure attached sensors and log data to non-
volatile memory (NVM) when it is away from a reader. As a proxy for cold chain
monitoring, a refrigerated milk container was instrumented with a WISP-PDL and
monitored throughout its consumption. For this study, the WISP-PDL sampled and
logged data in 10 second intervals and consumed 1.8 µA on average from a 1.8 V
supply. Over the course of 24 hours, the temperature and fill level of the carton was
measured and written to memory. At the end of the study, the data was read from the
WISP-PDL using the Gen 2 Read command showing the complete history of the milk
carton. As the refrigerator acted as a faraday cage, the WISP-PDL harvested energy only
when removed from the refrigerator but continued to sense when not directly powered
by a reader.
Building off of the work in [31], a fully integrated WISP-PDL has been
implemented by the author of this thesis and is shown in Figure 23. The 0.1 Farad red
super capacitor is clearly visible on top of the WISP. Additional features, such as a 1.8v
external 8k EEPROM, and a 350 nA Real Time Clock have been added to expand the
capabilities of the PDL. Presently this platform is still under development.
48
Figure 23. WISP data logger with operational Super capacitor.
49
Chapter 9 Conclusion
In order to explore RFID applications beyond simple barcode replacement, this
thesis argues the need for a reconfigurable, computation and sensor enhanced device,
capable of being wirelessly powered and interfacing with standard RFID technology. It
is believed that such a device will lead to the discovery and rapid implantation of
innovative RFID applications that can not be developed by traditional CMOS tag design
alone. Furthermore, the cost structure of traditional RFID applications is focused on
manufacturing tags that can be sold at the lowest possible price, as a replacement for
barcodes. However, it has been shown that by adding even a small amount of increased
functionality to RFID tags, new markets can become available and the profitability of
the tags can be substantially increased.
This thesis presents the design and performance of the Wireless Identification
and Sensing Platform, a programmable, sensor enhanced, passive UHF RFID tag. The
WISP is powered by a standard, commercially available UHF RFID reader and
implements the Electronic Product Code (EPC) Class 1, Generation 2 protocol.
The first half of this thesis describes the implementation of the major functional
blocks of the WISP: RF harvester, downlink demodulator, power management,
microcontroller, and uplink modulator. Additionally, the power management algorithm
is presented along with a general overview of how the WISP uses the EPC Gen2
protocol to communicate data to and from the reader. This is followed by a detailed
analysis of the power budget and performance of the WISP, which shows an ideal range
of 4.3 m. Finally, several low power sensors that have been integrated into the WISP are
presented: light, temperature, and acceleration.
More generally, WISP has proven the feasibility of powering devices with
relatively large power consumption (such as a microcontroller and sensors) using only
the RF energy transmitted by a standard RFID reader. To the author’s knowledge, the
WISP is the first of a new class of battery-free, wireless sensing and computational
devices.
50
In the introduction, the author argued that implementing the WISP as a flexible,
PCB based platform, instead of a higher performance silicon intergraded circuit, would
allow people from a wide variety of fields to develop RFID technology. Since the
completion of the first few prototypes, there have been a number of independent
research efforts using the WISP to create new RFID applications. The following
passages summarize two research projects, first the usage of the computational power of
the WISP for encryption, and the second, which uses the WISP’s sensing capability as a
user input device.
Conventional wisdom states that strong cryptographic algorithms are unrealistic
for RFID considering the computational constraints and power issues of IC tags. As a
result, various lightweight cryptographic protocols have been proposed and implemented
[15]. However, many of these protocols have serious vulnerabilities and were
subsequently hacked or exploited [17]. However, the computational power and
flexibility of the WISP enables the realization of stronger, more conventional
cryptographic techniques designed to enhance both privacy and security.
In [3], the WISP was used to demonstrate RC5 based symmetric cryptography
for use on UHF RFID tags. The particular RC5 variant implemented uses a 32-bit word,
12 rounds, and a 16-byte secret key, which is stored in flash. While there were practical
challenges in implementing RC5 on such a resource-constrained platform, the authors
showed that with careful implementation strong cryptography is within the scope of
UHF RFID. Additionally, their choice of RC5 was partly because RC5 can be efficiently
implemented in both hardware and software, so their work can be used as a basis for IC
implementations.
Even when strong cryptography is used, RFID is still susceptible to “man in the
middle” attacks. For instance, RFID is widely used for access cards where an RFID
enabled employee badge uses a cryptographically strong challenge/response mechanism
to open doors to a secured building. An attacker in this scenario does not need to break
the encryption but, only needs to generate the correct response to the RFID reader
challenge. On the other hand, RFID enabled credit cards have no encryption and the
information transmitted via RFID is virtually identical to that printed on the card.
51
Gathering this information no longer requires the conscious act of removing the card
from a wallet and swiping the card through a magnetic reader. Consequently, thieves can
steal the card information wirelessly, even while the card remains securely in the
cardholder’s wallet or purse.
In [5] ,the 3D accelerometer on the WISP was used to implement a “secret hand
shake” based authentication system to protect against “ghost and leech” and skimming
attacks. When the user wants to authenticate a transaction or gain building access, they
first perform a gesture with the card which unlocks the card and enables communication.
The gesture could be a figure eight or any unique movement that the card would not
experience in everyday activity. Only if this handshake is correct will the WISP unlock
and transmit its ID to the reader. This approach leverages not only the computational
power of the WISP, but also its sensing capabilities to provide a level of security that is
not possible using standard IC tags.
52
Chapter 10 Future Work
Up to this point, WISP has been viewed in the context of a single device
interacting with a RFID reader. Now that the hardware and software design of the WISP
is robust enough for a variety of applications, there is the potential to deploy a large
number of WISPs through out an environment.
In [2] we proposed to extend RFID beyond simple identification to in-depth
sensing using WISPs to form RFID Sensor Networks (RSNs). This combines the
advantages of RFID technology with those of Wireless Sensor Networks (WSNs). The
WISP demonstrates the technical feasibility of building small, battery-free devices that
use the RFID PHY and MAC layer to power themselves, sense, compute, and
communicate.
WSNs, based on “mote” sensing platforms, have been applied to many real-
world problems. Remote monitoring applications have sensed animal behavior and
habitat, structural integrity of bridges, volcanic activity, and forest fire danger, to name
only a few successes. These networks leveraged the relatively small form-factor
(approximately 1”x 2” x 3”) of motes and their multi-hop wireless communication to
provide dense sensing in difficult environments.
While the feasibility of WISPs has been established by this thesis, how to harness
many such devices to create RSNs is an open question. An RFID sensor network
consists of multiple WISPs and one or more readers. Consequently, realizing full-scale
RSNs will require development of both the WISP and the reader, as new protocols and
techniques must be developed unlike those of either RFID or WSNs. Since the
traditional RFID usage model is very different from that of WSNs, RSNs face
substantial challenges when trying to integrate the two technologies. For example, unlike
WSNs, RSNs must cope with intermittent power, and unlike RFID, must support sensor
queries rather than simple identification. Some of the key research challenges are
outlined below.
53
Tasks and Intermittent Power
There are two fundamental differences when developing application on WISP as
compared to battery powered systems; first, the available power is dependent on the
range of the WISP form the reader. Secondly, the time available to execute instruction is
rigidly enforced by the amount of energy stored in the supply capacitor and it’s
discharge rate. Thus, completion of a task can not be guaranteed because the application
does not have complete control of the duty cycling schedule. New methods for dealing
with these types of power constraint systems will have to be investigated.
Unpowered Operation
In many cases, WISPs may need to gather sensor data when they are not in the
proximity of an active RFID reader. For example, the temperature of blood plasma
should be monitored while it is out of the refrigerator; even though these periods should
be relatively short they are crucial for the application. One approach is to provide a small
amount of energy storage as was done with the Passive Data Logger WISP. However,
the large storage capacitor simply increases the RC time constant of the system (and thus
decreases the rate of voltage discharge). A more complete approach to dealing with
intermittent power in needed. Alternatively, RSN topologies that can deal with sensor
black outs while still being effective at gather data may also be a viable solution for
some applications.
Sensing Protocols
Currently, WISPs with new sensor data must wait until they are interrogated by a
reader. This increases the likelihood of many devices wanting to use the bandwidth
limited channel at the same time. Techniques to perform data preprocessing within the
network (on each RSN device) can help to some extent. However, the standard RFID
strategy of identifying and then communicating with each device is wasteful as most
devices may not have relevant data. A more dynamic strategy based on the value of the
sensor data would be more effective.
54
Repurposing EPC C1G2
As previously mention the EPC Gen 2 protocol is not particularly well suited to
transmitting sensor data, which would be necessary for practical implementation of
RSNs. However, the Gen 2 protocol does provide mechanisms such as the Select
command and the option to overloading the ID field which have proven successful for
initial applications. Furthermore, commercially available EPC RFID readers provide a
valuable piece on infrastructure that can not be readily replace. Plus keeping RSN
development tethered EPC provides a standard interface for research to collaborate
around. Further experimentation is needed to see is current EPC RFID readers are
flexible enough to allow researcher sufficiently control in implement real world RSNs.
55
References
[1] Atmel Cooperation, "ATA5590 Datasheet." September 2004, Available at the following URL: <www.atmel.com>.
[2] M. Buettner, B. Greenstein, A. Sample, J. Smith. “Revisiting Smart Dust with RFID Sensor Networks”, Seventh ACM Workshop on Hot Topics in Networks (HotNets-
VII), Alberta, Canada, Oct 6-7 2008.
[3] H. Chae, D.J, Yeager, J.R. Smith, K. Fu, "Maximalist Cryptography and Computation on the WISP UHF RFID Tag," in Proceedings of the Conference on RFID Security, 2007.
[4] N. Cho, et al, "A 5.1-µW 0.3-mm2 UHF RFID Tag Chip Integrated With Sensors for Wireless Environmental Monitoring," IEEE European Solid State Circuits
Conference, September, 2005, Grenoble, France. pp. 279- 282.
[5] A. Czeskis, K. Koscher, J.R. Smith, T. Kohno, “RFIDs and Secret Handshakes: Defending Against Ghost-and-Leech Attacks and Unauthorized Reads with Context-Aware Communications.” 15th ACM Conference on Computer and Communications
Security, October 27-31, 2008.
[6] Engadget "Fujitsu unveils GPS receiver with integrated RFID tag." Engadget. 27 Dec 2006. <http://www.engadget.com/2006/12/27/fujitsu-unveils-gps-receiver-with-integrated-rfid-tag/>.
[7] EPC Global, "EPC Generation 1 Tag Data Standards." Version 9 1.1 Rev.1.27, 10 May 2005 <http://www.epcglobalinc.org/standards>.
[8] EPC Global, "EPC Radio-Frequency Identity Protocols Class-1 Generation-2 UHF RFID Protocol for Communications at 860 MHz - 960 MHz." Version 1.2.0, 23 Oct 2008 <http://www.epcglobalinc.org/standards/uhfc1g2>.
[9] Federal Communications Commission, "Radio Frequency Devices." Title 47 of the Code of Federal Regulations Part 15, 10 July 2008 pp. 101-106, 30 Nov 2008 <http://www.fcc.gov/oet/info/rules/>.
[10] K. Finkenzeller, “RFID Handbook - Fundamentals and Applications in Contactless Smart Cards and Identification” (2nd Edition). (pp. 1-7). John Wiley & Sons.
56
[11] L. Frenzel, "Low-Power Wi-Fi Breakthrough Offers Active RFID And Location Services." Electronic Design. 30 March 2006. <http://electronicdesign.com/Articles/Index.cfm?AD=1&ArticleID=12208>.
[12] D. Halperin, T.S. Heydt-Benjamin, B. Ransford, S.S. Clark, B. Defend, W. Morgan, K. Fu, T. Kohno, W.H. Maisel, “Pacemakers and implantable cardiac defibrillators: Software radio attacks and zero-power defense,” In Proceedings of the 29th Annual IEEE Symposium on Security and Privacy, May 2008.
[13] J. Holleman, D. Yeager, R. Prasad, J. Smith, and B. Otis, "Neural WISP: An energy harvesting wireless brain interface with 1m range," IEEE Biological Circuits and Systems (BioCAS) Conference, November 2008.
[14] Intermec Technologies Corporation, "Intermec RFID Tags & Media," 2007 <http://epsfiles.intermec.com/eps_files/eps_brochure/RFIDTagSelectionGuide_brochure_web.pdf>.
[15] A Juels, “Minimalist cryptography for low-cost RFID tags”. The Fourth International Conference on Security in Communication Networks. Volume 3352 of LNCS., Springer-Verlag (2004) pp 149–164.
[16] F. Kocer, M.P. Flynn. "A new transponder architecture with on-chip ADC for long-range telemetry applications," IEEE Journal of Solid-State Circuits, vol. 41, no. 5, May 2006, pp. 1142-1148.
[17] MBTA vs. Anderson, et al (United States District Court - District of Massachusetts August, 2008)
[18] Microchip Technology Inc., 125 kHz Passive RFID Device with Sensor Input, August 25, 2005
[19] M. Philipose, K.P. Fishkin, M. Perkowitz, D.J. Patterson, D. Fox, H. Kautz, D. Hahnel, "Inferring Activities from Interactions with Objects," IEEE Pervasive Computing ,vol. 03, no. 4, October-December, 2004, pp. 50-57.
[20] M. Philipose, J. R. Smith, B. Jiang, K. Sundara-Rajan, A. Mamishev, S. Roy, “Battery-Free Wireless Identification and Sensing” IEEE Pervasive Computing, vol. 4, no. 1, Jan 2005, pp. 37-45.
57
[21] D.C. Ranasinghe, K.S. Leong, M.L Ng, D.W. Engels, P.H. Cole, "A Distributed Architecture for a Ubiquitous RFID Sensing Network,” Proceedings of the
International Conference on Intelligent Sensors, Sensor Networks and Information
Processing, 2005, pp 7-12.
[22] RFID Journal LLC, “New Low-cost Temperature Sensor” RFID Journal, July 19, 2002 http://www.rfidjournal.com/article/view/28.
[23] L.G. Roberts, , “Aloha packet system with and without slots and capture” SIGCOMM Comput. Commun. Rev., 5(2):28–42, 1975.
[24] A.P. Sample, D.J. Yeager, P.S. Powledge, A.V. Mamishev, J.R. Smith, "Design of an RFID-Based Battery-Free Programmable Sensing Platform," Instrumentation and Measurement, IEEE Transactions on , vol.57, no.11, pp.2608-2615, Nov. 2008
[25] A.P. Sample, D.J. Yeager, P.S. Powledge, J.R. Smith, "Design of a Passively-Powered, Programmable Sensing Platform for UHF RFID Systems," RFID, 2007. IEEE International Conference on , vol., no., pp.149-156, 26-28 March 2007
[26] S. Sarma, "Towards the 5¢ Tag." Auto-ID labs, 10 Nov 2001 <http://www.autoidlabs.org/uploads/media/mit-autoid-wh-006.pdf>.
[27] J.R. Smith, B. Jiang, S. Roy, M. Philipose, K. Sundara-Rajan, and A. Mamishev, “ID modulation: Embedding sensor data in an RFID timeseries,” Proceedings of the Seventh Information Hiding Workshop Barcelona, Spain, June 2005.
[28] J.R. Smith, A.P. Sample, P.S. Powledge, A.V. Mamishev, S. Roy., “A wirelessly powered platform for sensing and computation” Proceedings of Ubicomp 2006: 8th International Conference on Ubiquitous Computing, Orange Country, CA, USA, September 17-21 2006, pp. 495-506.
[29] B. Trebilcock, "What ever happened to the 5-cent RFID tag?" Modern Materials
Handling Company Briefings, 05 June 2008. <http://www.mmh.com/blog/1000000500/post/1210027721.html>.
[30] R. Want, "Enabling Ubiquitous Sensing with RFID," Computer ,vol. 37, no. 4, April, 2004, pp. 84-86.
58
[31] D.J. Yeager, P.S. Powledge, R. Prasad, D. Wetherall, J.R. Smith, "Wirelessly-Charged UHF Tags for Sensor Data Collection," RFID, 2008 IEEE International Conference on , vol., no., pp.320-327, 16-17 April 2008.
