Generation 2

A User Guide
 Summary

EPC Generation 2 is a new standard for RFID tags, specifying the operation of the tag
and the communication protocol for interoperability with EPC readers. It was developed
by a collaboration of leading RFID users and vendors, working through EPC Global, a
non-profit trade group. EPC Global is part of the UCC/EAN organization, which has
long administered bar code and other standards around the world.

RFID technology is constantly evolving, and Generation 2 marks a milestone in the

development of RFID for supply chain management applications. EPC Generation 2
was developed by the EPC Global community, including ThingMagic, to be a ‘best-of-
breed’ approach to RFID, and is intended to deliver improved performance in a number
of areas, particularly how quickly large numbers of RFID tags can be read. Many RFID
vendors have committed to support EPC Generation 2 – more than currently support
EPC Generation 1. This will increase competition in the RFID market, which in turn will
drive innovation, leading to reduced prices and increased performance over time.

Because of the inevitable learning curve surrounding the introduction of any new tech-
nology, 2005 will be a development year for Generation 2 RFID tags. Throughout the
year, there will be regular announcements from vendors about prototypes, plans, part-
nerships and production milestones. Tag samples and prototypes will become available,
especially to large end users. Performance and interoperability testing are also sched-
uled to occur in 2005, once the test methodologies and performance standards have
been defined and agreed to by the EPC Global community.

High volume, general availability of Generation 2 tags is expected some time in late
2005 or early 2006. However, Generation 1 tags will not disappear from the sup-
ply chain: instead Generations 1 and 2 will co-exist for 6 – 12 months while the
Generation 2 learning curve ramps up. Depending on Generation 2 price and perfor-
mance, and market conditions, this ‘Generation 1 Shadow’ could last until early 2007,
when Generation 1 tags finally disappear from the supply chain. Tag technology will
continue to evolve; around the time Generation 2 becomes established around 2006/7,
end users and technology vendors expect work to begin on developing a ‘Generation 3’
system with further improvements to Generation 2 technology.

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide i
 This constant evolution of tag standards is a major consideration for buyers of RFID
infrastructure. Tags are the consumables of the RFID world, but the permanent infra-
structure is made up of RFID readers, bought as capital equipment and installed and
maintained at an ongoing additional expense. Simply put, tags may come and go, but
readers are forever. To avoid painful obsolescence or costly retrofits, deployment-scale
RFID readers must be able to read any tag, regardless of generation or vendor. They
must be easily upgradeable to account for new tags, and to accommodate changes to
existing tags as tag vendors change silicon revisions. For these reasons, EPC readers
must always be able to read at least three generations of tags – the current one, the
previous one, and the next one.

The technology best suited to this complex task of interoperability among EPC genera-
tions is Software Defined Radio (SDR). SDR is an advanced radio technology in which
the modulation and demodulation of radio signals is performed exclusively by software.
In the RFID context, this means that new communication protocols can be added with
straightforward software upgrades, and an SDR-based reader can easily handle multi-
ple protocols simultaneously. ThingMagic has long been the leader in SDR-based RFID
readers, since its Mercury1 SDR-based reader was introduced in 2001. It seems inevi-
table that after the introduction of Generation 2 most readers will be made this way, but
ThingMagic’s mature SDR technology base will allow it to continue to lead the market in
capabilities and performance.

In this white paper, we will explore the detailed infrastructure requirements for reading
EPC Generation 2, explain how ThingMagic’s SDR technology easily manages the tran-
sition from Generation 1 to future protocol generations, and show how SDR provides
effective future-proofing against the future EPC generations and tag vendor iterations to
come.

ii Generation 2
A User Guide
 Contents

1 1.A Multi-Protocol
EPC Generation 2
Protocol
12 3. Beyond EPC Generation2: Other
Classes, Other Generations

i. A Brief History Of Generation 2 i. The EPC Class Structure

ii. The Specification in Overview ii. Drivers Of Generation3

Development
iii. Generation 2: A ‘Multi-Protocol
Protocol’
14 4. Software Defined Radio: Flexible,
Future-Proof RFID Infrastructure
iv. The Generation 2 Tag Value Chain
i. Solving The Problem Of Tag
v. Companies Involved In Producing Variation & Iteration
Generation 2 Tags
ii. A Definition Of SDR
vi. Possible Variations of In
The Generation 2 Specification iii. Brief History of SDR

vii. Special Features: Higher Bit Rate iv. Benefits of SDR in RFID Readers

viii. Special Features: Dense Reader v. SDR and Generation 2

Mode
16 5. Case Study: How Mercury4’s
9 2. EPC Generation 2
Market Timings And Development
Software Defined Radio Delivers
Native Generation 2 Performance

i. Steps To Market i. Mercury4’s SDR-Architecture

Overview
ii. Likely Lead Times & Timings
ii. Mercury4’s Native Generation 2
iii. Generation 1 Likely Sunset Timings Capability

iv. Generation 1 Shadow: Managing iii. Adding Native Generation2

The Overlap & Transition
iv. Other Benefits of SDR For
vi. Iterations: How Generation2 Will Generation2 Transition
Change Over Time
19 6. Frequently Asked Questions

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide iii
 1 EPC Generation 2:
A Multi-Protocol Protocol
i. Brief History Of Generation 2

EPC Generation 2 has its origins in a meeting held by the Massachusetts Institute
of Technology’s Auto-ID Center in Newport, Rhode Island on October 2, 2002. The
Center wanted to add RFID protocol features contributed by Matrics, Inc. to its existing
EPC Generation 1 Class 1 draft specification. In addition to the directors of the Auto-
ID Center, other attendees included representatives from Matrics, Alien, Philips and
ThingMagic. This exchange of ideas in Newport led to new thinking on EPC tag design
and the evolution of EPC technology.

Rather than delay the ratification of the EPC Generation 1 specification, the Center
decided to use these new ideas to begin work on a new generation of EPC tags,
intended to reach the market in 2005. This new generation would serve three purposes:
first, it would aim to deliver performance improvements over Generation 1 technol-
ogy. Second, it would drive convergence between Generation 1’s Class 0 and Class
1 protocols. Third, it would provide a point of market entry for RFID vendors who had
not been closely involved with the Generation 1 specifications. When the Auto-ID
Center transitioned to EPCGlobal a year later, the Generation2 development process
transitioned too. During 2004, under the leadership of EPCGlobal, a growing num-
ber of users and vendors developed and agreed upon the final specification for EPC
Generation 2 tags. The specification was ratified at the end of the year, and work began
on developing commercial products that met the Generation2 requirements.

ii. The Specification In Overview

The Generation 2 specification is a 94-page engineering document entitled “EPC™
Radio-Frequency Identity Protocols / Class-1 Generation-2 UHF RFID / Protocol for
Communications at 860 MHz – 960 MHz”. The full text is copyright EPCGlobal, Inc.
and is only available to EPCGlobal subscribers. It describes in considerable detail how
Generation2 RFID tags should communicate with RFID readers.

Some key points of the specification are:

• Tags must be able to communicate on any frequency between 860 MHz to 960

1 Generation 2
A User Guide
 MHz. To account for variations in regional radio regulations, readers may operate
using any permitted frequency within that range.
• Tags must be able to understand three different modulation schemes: Double
Sideband-Amplitude Shift Keying (DB-ASK), Single Sideband-Amplitude Shift
Keying (SS-ASK) and Phase-Reversal Amplitude Shift Keying (PR-ASK). Readers
will determine which modulation scheme is used, within the context of government
radio regulations.

• Tags must be able to transmit at several different speeds, or data rates: 80kbits,
160kbits, 320kbit and 640kbits. Readers determine what speed to use. For per-
spective, Generation1 protocols communicated at between 70 and 149kbits. In
theory, the higher data rates of Generation 2 will allow faster tag reads, but in
practice many other factors beyond raw transmission speed contribute to real world
read rates, and the higher data rates may sometimes result in lower reliability.

• Generation 2 tags support Electronic Product Codes up to 256 bits long, where-
as Generation1 tags support Electronic Product Codes up to 96 bits long.

• Generation 2 includes a method to support ‘dense-interrogator channelized

signaling’ (sometimes called ‘dense reader mode’) which is an attempt to reduce
interference between EPC readers, by making it less likely that reader signals will
drowning out tag responses. This mode is intended for use in locations where
multiple readers are in operation at the same time. Implementation of this part of
the specification is optional for reader makers. Again, real world performance will
depend on many factors, including external interference from other devices, such
as UHF cordless telephones, industrial equipment, and legacy UHF wireless LAN
equipment.

The most obvious conclusion to draw from the Generation 2 specification document
is that Generation 2 is something of a ‘Chinese Menu’ for RFID. The multiple options
for modulation, for example, provide a number of different ways to do the same thing,
which is to receive data from tags. The specification also provides for many optional
commands, as well as vendor-specific custom commands. To be fully compliant with
the specification, a tag must offer the full menu. The reader, on the other hand, can pick
and choose among these options by, for example, determining the what modulations
scheme and data rate the tag should use in the context of a particular communica-
tion attempt. For a variety of reasons, including achieving good performance in a wide
variety of end user applications, and to function effectively in the presence of external
interference, the reader is likely to dynamically choose among these many options to fit
each particular circumstance. This is why we call EPC Generation 2 a ‘multi-protocol
protocol’.

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 2
 iii. EPC Generation 2: A ‘multi-protocol’ protocol

EPC Generation 2 has a much higher potential for variation than any other RFID tag.
Its ‘Chinese Menu’ approach to the underlying communication protocol is one reason
for this, but there are other reasons too. For example, different tag vendors will under-
stand the specification in slightly different ways. Some vendors may implement only
some of the options due to schedule or cost constraints. Some will add proprietary
‘extensions’. Some tags will have two antennas, while others will only have one antenna.
Even among tags made by the same vendor, variations will be introduced over time, by
design changes for cost or manufacturability reasons, or attempts to deliver better per-
formance.

While compliance tests and adherence to the specification will deliver a basic degree
of interoperability between different ‘Generation 2’ flavors, RFID readers may need to
address each variation differently to achieve optimal performance from each one, and
among mixed populations of Generation 2 tags from different vendors. The unprec-
edented number of tag vendors competing in the Generation 2 market means that a
declaration of ‘Generation 2’ compliance may not, in itself, be an assurance of optimal,
or even acceptable, performance from all tags.

Until Generation 2, the RFID industry has never seen this degree of support for a single
platform before, and true interoperability will require a subtle, nuanced approach to
RFID infrastructure. The potential for broad variation means missteps in purchasing
RFID readers could cause significant and unforeseen problems for system integrators
and end users.

iv. The Generation 2 Tag Value Chain

The true extent of Generation 2’s potential for variation is a result of an ever-expand-
ing RFID tag value chain. Making an RFID tag requires many steps, conducted by many
companies. Each step has some impact on how tags behave, both in homogeneous
and heterogeneous populations, and greatly impact what readers must do in order to
work optimally with groups of tags.

The higher levels in the Value Chain (e.g. Specification, Chip Design) tend to introduce
the most variation. Lower levels (e.g. label conversion) introduce less variation, but can
greatly impact physical performance characteristics such as range and consistency of
behavior across tagged products.

3 Generation 2
A User Guide
 SPECIFICATION
MOST
VARIATION

CHIP DESIGN

CHIP FABRICATION

CHIP PACKAGING TAG ANTENNA DESIGN

INLAY MANUFACTURE

EACH STEP IN THE

GENERATION 2 TAG VALUE
CHAIN CHANGES HOW
EACH FINISHED TAG BEHAVES LABEL CONVERSION

LEAST VARIATION
PRINTING / ENCODING

FINISHED GENERATION 2 TAG

Figure 1: Each Step In The Tag Value Chain Changes Tag Behavior

v. Companies Involved in Producing Generation 2 Tags

While the potential variation introduced at a given step in tag production decreases as
the tag moves through the value chain, the number of companies involved – and there-
fore the quantity of variations - increases. The net result is that a great number of tag
variants can be introduced by any stage in the process.

This variation already exists in the Generation 1 world, which has far fewer participating
vendors. The situation will change with the much larger number of companies commit-
ted to Generation 2 technology. Some companies that have publicly announced their
participation in the Generation 2 value chain include:

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 4
 Given this large number of companies and organizations, it’s not surprising that there
will be a large number of Generation 2 variants!

vi. Possible Chip-Level Variations In The Generation 2

Specification

Full compliance with the Generation 2 specification means almost all options in the
‘Chinese Menu’ must be offered, but many chip designers are already considering mak-
ing cheaper products that offer a reduced number of options. Variations of this type that
may be expected include:

Excluding the special case of Proprietary and Custom commands, these options com-
bined allow for some several hundred variations among at least 10 chip vendors. While
a number of permutations this large is unlikely, it is quite likely that a significant number
of different combinations will be tried as the Generation 2 market matures through
2005 and beyond.

5 Generation 2
A User Guide
 vii. Special Chip-Level Features: Higher Bit Rate

One new feature of Generation 2 is a higher speed or data transmission (or ‘bit rate’)
than in Generation 1 tags. Generation 2 tags will be capable of sending data and
responding to commands at speeds up to 8 times faster than Generation 1 tags. In
theory, this higher bit rate will lead to a greater number of tags being read per second.
The increase in real-world tags read per second will not keep pace with the increase
bit rate, however, as higher speeds necessarily introduce higher error rates and greater
noise susceptibility, and longer EPC codes (up to 256 bits) will also consume some of
the extra communication capacity.

One analogy that for this situation is that buying a faster car does not guarantee your
commute will be quicker – total commuting time is strongly dependent on externalities
like traffic and road conditions. The same is true of tag data rate versus tag read rate. It
is likely that Generation 2 will deliver faster tags reads than Generation 1, in many situ-
ations. But the real-world performance increase will not be directly proportional to the
increased data rate provided by Generation 2, and in some situations the best course
of action may be for the reader to fall back on a lower data rate mode for communica-
tion reliability reasons.

viii. Special Chip-Level Features: ‘Dense Reader Mode’

Another new feature of Generation 2 is ‘dense reader mode’. Dense reader mode is
intended to help avoid reader interference with tag responses, by reserving certain sub-
parts of the government allocated radio spectrum for tags to use. As tags are much
weaker communicators than readers, this has some benefit in helping tags to work in
situations where multiple readers are operating in close proximity.

However, ‘dense reader mode’ is not a panacea for every problem that can arise when
many readers are operating in close quarters. There are several important aspects to
interference, not all of which are addressed by ‘dense reader mode’. These include:

• Other devices besides RFID readers may be operating in or around the UHF
spectrum. These devices are free to hop into channels being used by tags, and can
drown them out. In the US, for example, unlicensed FCC Part 15 devices, such as
RFID readers, cordless telephones, wireless LAN equipment, industrial equipment,
etc, are permitted the same priority in UHF spectrum use. The presence of even
one non-RFID device in an end-user facility may partially or completely negate the

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 6
 potential benefits of dense-reader mode, because non-RFID devices will most
likely not adhere to the channelization scheme required for dense-reader mode to
be effective.

• Not all Generation 2 RFID readers must support ‘dense reader mode. It is an
optional part of the Generation 2 specification. Even a single non-dense-reader
mode RFID reader, such as a handheld reader, a UHF active tag system, or a
legacy proprietary RFID system, may also drown out tag responses and negate the
potential benefits of dense-reader mode.

• Passive RFID tags, such as Generation 2 tags, are broadband backscatter devic-
es. As such, when they respond to one reader’s interrogation on a certain frequen-
cy, they are actually modulating a response across all frequencies simultaneously.
This backscatter modulation can be picked up by any reader within view of the
responding tag, which can lead to interference even at much longer distances than
those at which the tag can be effectively read. Thus, all tags being read by a given
reader are potential interference generators to all nearby readers. This effect will be
particularly pronounced in environments such as long rows of warehouse loading
dock doors in which large numbers of tags are being interrogated by large numbers
of readers, and all readers and tags are operating in close proximity.

• Most importantly, as several papers by Auto-ID Labs’ Director Dr. Daniel Engels
have pointed out, in addition to being broadband backscatter ‘transmitters’, tags
are also ‘broadband receivers’ – tags have no knowledge of reader channels. The
receiver on a tag is akin to a crystal radio set, in that its frequency selectivity is
limited only by the properties of the tag’s antenna. This is in fact enforced by the
Generation 2 specification, because the Generation 2 specification requires tags
to operate from 860-960MHz. Thus, a given tag will receive all transmissions being
sent at the same time by all readers in view, even if the readers are operating in
‘dense reader mode’ and are therefore using different channels. This is a funda-
mental problem with contemporary RFID, and is best solved by precise, time-based
reader synchronization that aims to prevent two readers trying to talk to the same
tag at the same time. Because of this, time-based reader synchronization is the
most important strategy for managing dense reader environments. Unfortunately, the
Generation 2 ‘dense reader mode’ does not address this important shortcoming of
the broadband receiver on Generation 2 tags.

• ‘Dense reader mode’ should therefore be thought of as a first step toward opera-
tion in dense reader environments. But it is not a cure-all for the limitations of
today’s passive RFID technology. The fundamental problems of dense reader

7 Generation 2
A User Guide
 environments, including the tag’s broadband transmission and reception, remain
to be solved in future generations of passive RFID technology. In the mean time,
ThingMagic’s team of RFID experts have worked to ensure that ThingMagic’s prod-
ucts support the fundamental principles of ‘dense reader mode’, and are always
working to improve the performance of our existing (and powerful) time-based
reader synchronization modes.

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 8
 2 EPC Generation2 Market
Timings And Development
i. Steps To Market

The development and manufacture of Generation 2 tags by tag vendors (described in

1.iv above) is only the beginning of the process of adding Generation 2 tags to the sup-
ply chain. Once the tag development and manufacture is finished, a number of other
steps must take place before a tag is consumed, or read for the last time – for example
by a retailer at the time a case or pallet is broken down.

SUPPLY TO
PRINTER- INSTALL IN APPLIED TO
FINISHED LABEL ENCODER PRINTER / PRODUCT IN
MAKERS / ENCODER PROCESS
USERs

ADDED TO ADDED TO
DISTRIBUTED PURCHASED /
MANUFACTURER CUSTOMER
INVENTORY TO CUSTOMER CONSUMED
INVENTORY

Figure 2. Generation 2’s Steps To Market

These steps between the manufacture of the tag itself and the tag’s eventual purchase
or consumption add additional ‘time to market’ to the Generation 2 deployment cycle.
While some tagged products will move through the supply chain very quickly, other tag
products spend long periods in inventory both with the manufacturer and their custom-
ers. In addition, there is another important choke point in the Generation 2 steps to
market: the printer / encoder. Generation 2 tags must be written with end-user mean-
ingful data before they can be used. This writing operation will typically be done by an

9 Generation 2
A User Guide
 RFID-enabled label printer, a print-apply machine, or in some cases a fixed RFID reader
integrated into a conveyor. At each Generation 2 customer site, these writing devices
must be Generation 2 capable before Generation 2 tags can be used.

In addition, many customers and their suppliers may have inventory of Generation 1
labels that must be consumed before they begin encoding their Generation 2 tags. The
general availability and purchase of encoders that can handle both Generation 1 and
Generation 2 tags is a pre-requisite – and therefore a rate limiter - for Generation 2’s
entry into the supply chain.

ii. Likely Lead Times & Timings

Given the above, we can estimate the spread of likely lead times and timings for
Generation 2 adoption.

Depending on supply chain complexity and specific manufacturer / customer / product

combinations and situations, the first production quality Generation 2 tags could arrive
in a live Supply Chain as early as September 2005. This will be the leading edge of
adoption, however. Many products will first get Generation 2 tags much later, and some
users may ship their first Generation 2 tags as late as Summer 2006.

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 10
 iii. Generation1 Likely Sunset Timings

In overlap with this process, users and vendors will be exhausting their inventory of the
more mature, field- proven Generation 1 tags. These tags will not simply disappear
once Generation 2 is released – they will have to work their way through the supply
chains of both tag vendors and tag users. While most Generation 1 tags may be con-
sumed by July 2006, some Generation 1 tagged products may persist for longer – for
example due to unexpected sales slumps, forward buying or seasonality. For these rea-
sons, Generation 1 tags may remain in some supply chains as late as the start of 2007.

iv. Generation1 Shadow: Managing The Overlap &

Transition

Managing this ‘Generation 1 Shadow’ will be a key consideration for purchasers of

RFID infrastructure. RFID Readers in many supply chains will need to be able to read all
forms of Generation 1, as well as all forms of Generation 2, well into 2006 and possibly
beyond.

vi. Iterations: How Generation 2 Will Change Over Time

Once Generation 2 is released and begins to be established in the marketplace, it

will evolve. The highly competitive market for Generation 2 tags will lead to continu-
ous improvement, as IC designers and tag / inlay makers in particular make changes
to retain their leadership with regard to the price and performance of their competitors.
This vigorous innovation will introduce additional variations to the ‘multi-protocol pro-
tocol’. For RFID reader infrastructure to be considered truly Generation 2 compatible,
reader systems must be able to adapt to these changes painlessly and cost-effectively
to avoid unnecessary restriction of innovation, limiting performance or costly obsoles-
cence.

11 Generation 2
A User Guide
 3 Beyond EPC Generation2:
Other Classes, Other Generations

i. The EPC Class Structure

The EPC Tag Class Structure is often misunderstood. ‘Class’ is not the same as
‘Generation’. Class describes a tag’s basic functionality – for example whether it has
memory or a battery. Generation refers to a tag specification’s major release or ver-
sion number. The full name for what is popularly called EPC Generation 2 is actually
EPC Class 1 Generation 2, indicating that the specification refers to the second major
release of a specification for a tag with write-once memory.

The full EPC Class Structure is:

Figure 3. EPC Class Structure

The Class 0 designation was added to the Generation 1 system long after the Class
1 specification was created, as part of the process described in 1.i above. Class 1

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 12
 tags, of which ‘Generation 2’ is an example, contain a write once memory for storing an
Electronic Product Code. Class 2 tags add additional memory that can be changed fre-
quently, for storing additional data – for example from an onboard sensor. Class 3 tags
add batteries for longer read ranges and higher reliability, but are fundamentally passive
backscatter tags. Class 4 tags are essentially active tags that can communicate with
other Class 4 tags as well as readers. Class 5 tags are not really tags at all - they are
essentially wireless networked readers.

The intent of the EPC Class Structure is to provide a modular structure that covers a
wide variety of possible types of tag functionality. For example, the communication pro-
tocol for a battery-powered tag should be the same as the protocol for a batteryless
tag, with only the addition of necessary commands to support the battery. This way,
protocols are kept simple, and if the battery on a battery-powered tag fails or dies, the
tag can simply behave like a batteryless tag and still have some utility to the end user.
This kind of modularity is far easier in theory than in practice – nonetheless technolo-
gies tend to converge over time, and there is desire within the EPC movement to deliver
the dream of a modular, multi-functional stack of tag protocols. This adds another
dimension to the variety of tags that a reader infrastructure must be able to adjust to.
Now that the disparity between Class 0 and Class 1 has been eliminated by Class 1
Generation 2, the next likely step is to add a battery-tag to the system. This may take
the form of a Generation 2, Class 3 tag – or it could point the way to a while new gen-
eration entirely.

ii. Drivers Of Generation3 Development

The EPC system will not stop at Generation 2. The mass market for RFID that
Generation 2 will create should result in new waves of innovation as vendors compete
and users become more sophisticated and demanding in terms of reduced cost and
increased functionality. As was the case with EPC Generation 1 (see 1.i above) these
competitive pressures will quickly lead to ideas and inventions that cannot be con-
tained within the Generation 2 specification. At this point, most likely around late 2006
or early 2007, the development of a Generation 3 tag will be introduced to the EPC
standards development process. Just as occurred with the Generation 1 to Generation
2 transition, successful Generation 2 incumbents will resist this transition in turn, while
entrepreneurs, start ups and large corporations coming late to the EPC market will jump
in. The momentum will build, and around 2008 or 2009 Generation 3 tags will appear,
starting the cycle described here all over again. The RFID reader infrastructure must be
ready for more change.

13 Generation 2
A User Guide
 4 Software Defined Radio:
Flexible, Future-proof RFID Infrastructure
i. Solving The Problem Of Tag Variation & Iteration

The problem of continuous change in the EPC market posed in this paper is a vitally
important for all RFID users, and especially those responsible for buying and installing
RFID reader infrastructure. While tags are the consumables of the RFID systems, con-
stantly varying, iterating and regenerating, the RFID reader infrastructure is a deployed
capital expense that cannot easily or cost effectively be replaced every time a new tag
variant appears. Further, the comings and goings of tags are not neatly synchronized.
Generation 1 will not turn into Generation 2 at the stroke of midnight – the two genera-
tions must coexist, perhaps for as long as a year. And once Generation 1 is has disap-
peared, the reality (and hype) surrounding Generation 3 will begin, as well as the intro-
duction of new classes of tag. All this change is good – in fact crucial – for the RFID
user. It will deliver ever-improving performance, and ever-decreasing costs. But it can
only flourish is the RFID reader infrastructure is not an impediment to change – if it can
endlessly and easily adapt and adjust as new variations of tag appear and disappear.
The technology that enables this endless adaptation is called Software Defined Radio,
or SDR.

ii. A Definition Of SDR

A software-defined radio (SDR) uses software for the modulation and demodulation
of radio signals. An SDR performs the majority of its signal processing in the digital
domain, most commonly in a digital signal processor (DSP), which is a type of micro-
processor specifically optimized for signal processing functions. The magic of an SDR
based RFID reader is that it can receive and transmit a new form of RFID communica-
tion protocol simply by running new software on existing SDR hardware.

A software-defined RFID reader consists of an RF analog front end (“AFE”) that con-
verts RF signals to and from the reader’s antennas into an analog baseband or interme-
diate frequency signal, and analog to digital converters and digital to analog converters
which are used to convert these signals to and from a digital representation that can be
processed in software running on the reader’s digital signal processor.

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 14
 Put simply, RFID readers based on SDR technology store all protocol information in
software, and use protocol-neutral hardware to generate and detect radio waves. This
in contrast to conventional RFID readers, where protocols are created by adding hard-
ware, and changing protocols requires changing components in the reader.

iii. Brief History of SDR

Software defined radio technology has long been important in the military context,
where new radio equipment must interoperate with legacy equipment, much of which
is used for many years beyond its design lifetime. Additionally, the US military is often
called upon to work together with allies that have old, outdated equipment that is
incompatible with the more modern US communication hardware. This is exactly analo-
gous to the Generation 1 to Generation 2 (and beyond) transition in RFID.

Military SDR projects date back to the early 1990s, and several were fielded in that
time frame. Aware of these developments, in 1999 the MIT Auto-ID Center began
exploring the idea of using SDR in RFID readers. This resulted in a development con-
tract to ThingMagic LLC in 2001. ThingMagic demonstrated the first SDR-based
RFID reader at the Auto-ID Center’s Board Meeting in Cambridge, Massachusetts
in November 2001. Commercial SDR-based RFID readers were first introduced by
ThingMagic in 2003.

iv. Benefits of SDR in RFID Readers

SDR-based readers offer significant benefits to RFID users. In particular, SDR-based

designs have a unique ability to adjust and upgrade to new protocols, and to be con-
stantly improved via firmware upgrades. With the addition of standard networking
capabilities such as remote management, large networks of deployed SDR-based RFID
readers can be upgraded in a matter of minutes from a remote console in a customer’s
Network Operations Center, instantly becoming capable of reading new types of tag

v. SDR & Generation 2

SDR will be especially important as the RFID industry moves to Generation 2. This
important transition will take place as leading users are well into their RFID deploy-
ments. For the earliest adopters, it will be necessary to upgrade the installed base to
Generation 2, while retaining the ability to read Generation 1 tags. For later adopters,
the frequent iteration of Generation 2 as the technology matures, and as tag vendors
compete aggressively to improve performance and gain advantage, will be painful and
costly unless their infrastructure is based on SDR and can easily be upgraded and
adjusted as tags change.

15 Generation 2
A User Guide
 5 Case Study: How Mercury4’s Software
Defined Radio Delivers Native
Generation 2 Performance

i. Mercury4’s SDR-Architecture Overview

ThingMagic’s Mercury4 is an example of an advanced SDR-based RFID Reader plat-

form. Mercury4 platform based devices are also marketed as the ‘Agile 2’ by Tyco/ADT-
Sensormatic, and as the ‘VR-740’ by Omron. This SDR argument also applies equally
well to the ThingMagic Mercury4e/Mercury4h embedded reader product, which shares
a common code base with the Mercury4. We will now address the specifics of how
ThingMagic’s SDR-based RFID Infrastructure addresses EPC Generation2.

The Mercury4 consists of analog front ends that are essentially ‘stateless’, in that their
hardware is designed to be as protocol and application neutral as possible. They
are, however, optimized by frequency band – e.g. 902 – 928 MHz – and to meet the
requirements of certain regulatory environments – e.g. the FCC’s Part 15 regulations in
the United States or the ETSI EN302-208 regulations in Europe.

These analog front ends are connected via high speed analog to digital and digi-
tal to analog converters operating at sampling rates up to 13.56Msa/sec to a high
speed Digital Signal Processor (DSP) which in the case of the Mercury4 is a TI
TMS320VC5502, which provides 600MIPS of processing on parallel multiply-accu-
mulate pipelines, while running at a clock speed of 300MHz. The analog front ends
are capable of fully general I/Q modulation and demodulation with bandwidths up to a
Nyquist anti aliasing filter limit of 5.0MHz (double sideband), allowing up to 10.0MHz
of incoming signal spectrum to be simultaneously digitized and processed. Protocol-
related software is loaded into the DSP, which then controls how the analog front ends
communicate with RFID tags. The DSP is, in turn, controlled by an Intel IXP420 network
processor running at 266MHz. The Intel processor provides functionality including back
end networking, and high-level developer interfaces.

The high speed DSP, coupled with the protocol neutral analog front ends, means
that Mercury4’s hardware is capable of reading any tag that falls within its frequency

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 16
 band and regulatory set-up, up to the limits of tag subcarrier frequency of 5.0MHz on
either side of the transmitted carrier. This is more than adequate to accommodate all
Generation 2 waveforms. Practical tag reading is a matter of software alone – provid-
ing ThingMagic’s engineers have developed the appropriate software, Mercury4 will
read the tag. As currently configured, Mercury4 reads EPC Generation 1 Class 1, EPC
Generation 1 Class 0, ISO 18000-6B, EPC UCODE 1.19 and Zuma tags. If necessary,
all types of tag can be read on a scheduled basis, apparently at the same time from the
user’s perspective, with no deterioration in performance.

ii. Mercury4’s Native Generation 2 Capability

As the Generation2 specification was finalized, ThingMagic evaluated Mercury4’s ana-

log hardware to check for Generation2 compatibility. The result of this evaluation was
positive: Mercury4’s analog front ends are capable of complying with all aspects of the
Generation2 protocol, including Dense Reader Mode. Adding native Generation2 capa-
bility to Mercury4 is a matter of software development alone, and this development is
already bearing fruit with first silicon from the Generation 2 tag vendors.

iii. Adding Native Generation 2 to the Mercury4

ThingMagic Generation2 software development has been underway for some time
now. As a result of close partnerships with leading tag vendors, the company’s lab in
Cambridge, Massachusetts already has silicon-based tag samples and tag emulators
from a multiple suppliers, and its engineers are working on new software that will incor-
porate these tags seamlessly, and completely interoperably, with Mercury4’s existing
Generation 1 (including Class 1, Class 0, Class 0+) and ISO protocol set. This work
extends into ThingMagic’s Mercury4e and Mercury4h products, developed for RFID-
enabled label printers and handhelds. This new software release for the Mercury4,
called MercuryOS 2.3 Tesla, will undergo final development and extensive field (beta)
testing during the Spring and Summer, and is expected to be publicly available in the
Fall, well ahead of general availability of Generation 2 tags.

Adding Tesla to installed Mercury4’s is a simple matter of a firmware upgrade. This

process has been field proven in large deployments since the Mercury4’s release in
2004. Mercury4’s advanced networking capabilities facilitate straightforward software
upgrades remotely across the network, with a typical reader system downtime of only
1 to 2 minutes, regardless of the number of readers being upgraded. Thousands of
Mercury4 readers can be upgraded this way at the same time, using a straightforward
upgrade process that has already been exercised many times in real deployments by
leading end users.

17 Generation 2
A User Guide
 iv. Other Benefits of SDR For Generation 2 Transition

Software Defined Radio provides other benefits to EPC Generation 2 users, beyond
ease of upgrade:

• Mercury4’s SDR-based hardware enables reading of Generation 1 and Generation 2

tags at the same time: for example, if a user received a mixed pallet of goods that was
tagged with a variety of Generation 1 Class 1, Generation 1 Class 0, and Generation 2
tags, a Mercury4 running MercuryOS 2.3 Tesla will have no difficulty in reading them all.

• As Generation2 tags change and improve, future software updates will allow Mercury4
to keep pace with new tag-vendor innovations, and end user needs

As ThingMagic engineers, in partnership with our network of system integrators and

end-users, gain more real world experience with more Generation 2 tags, software
updates can allow them to deliver continually improving performance. This process has
already brought industry leading Generation 1 performance to Mercury4 owners, and
will result in continued industry leadership for Generation 2.

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 18
 Frequently Asked Questions

Is it really possible to upgrade a reader to read Generation 2?

Absolutely. Not all RFID readers can be upgraded to Generation2 without costly hard-
ware changes, but Mercury4 definitely can be, with a simple software upgrade. We
can’t speak for other manufacturers or models.

What about dense reader mode? Will Mercury4 support that?

Mercury4 is completely capable of delivering dense reader mode through a software
upgrade. That said, ‘dense reader mode’ alone is not enough for dense reader environ-
ments. It’s much more important to be able to do precise time-based synchronization to
avoid readers talking to the same tag at the same time. Mercury4 already has a sophis-
ticated network-based time based synchronization capability that has been field proven
in large deployments.

What about the 640k bit rate? Will Mercury4 support that?
Mercury4 has no trouble demodulating this bit rate. This may only be necessary in niche
applications, however – for example high speed conveyors. Faster bit rates will not
mean faster tag reads in all circumstances. The Mercury4 will adapt its downlink rates
to yield optimal performance for particular end user scenarios.

What about Mercury4’s Generation2 performance?

Mercury4 delivers best in class performance on all protocols it currently supports.
Generation 2 performance will be no different.

Does all this power and flexibility make Mercury4 more expensive?
Because Mercury4’s power comes from sophisticated software, it is no more expensive
than conventional readers. Once total cost of ownership is taken into consideration,
Mercury4 is much cheaper, as many of those conventional readers have to be thrown
away or retrofitted as protocols change – Mercury4 only needs new software.

When will Mercury4 be reading Generation 2?

Mercury4 is already communicating with Generation 2 tags in ThingMagic’s test labs.
Once performance is fully optimized and stability testing is complete, the MercuryOS
2.3 Tesla upgrade that adds Generation2 to Mercury4’s protocol suite will become
commercially available. Expect this software upgrade to be widely available in the fall, or
sooner if Generation2 General Availability accelerates.

19 Generation 2
A User Guide
 How do I buy Mercury4?
Mercury4 is available through an extensive network of authorized partners and resellers.
See our web site at www.thingmagic.com for the latest information. You can also order
evaluation and developer kits from our web site.

Version 1.0, April 2005. Updates at www.thingmagic.com

Generation 2
A User Guide 20
 About ThingMagic

ThingMagic is a privately held RFID reader maker based in Cambridge,

Massachusetts. It was founded by five MIT PhDs in 2000. The company’s state
of the art EPC reader technology is derived from its legacy as a key player in the
EPC movement since the inception of EPC at the MIT Auto-ID Center. In 2001,
the Auto-ID Center awarded a contract to ThingMagic which resulted in the devel-
opment of a low-cost, multi-protocol reference reader for reading EPC-based
RFID tags. Codenamed Mercury1, it was the first multi-frequency, multi-protocol
EPC reader, and the first RFID reader to use Software Defined Radio technology.
Several successors to Mercury1 have been developed by ThingMagic and used in
the Auto-ID Center’s Field Trials, Evaluation Kits, and then in many end-user EPC
pilots worldwide.

Today, ThingMagic’s fourth generation platform, Mercury4, is the market-lead-

ing EPC reader, with large-scale infrastructure deployments around the world.
Additionally, ThingMagic brought the signal processing power of Mercury4 to
bear on the embedded and handheld RFID reader market, through its powerful
Mercury4e and Mercury4h products, which are market leaders in the label printing
and handheld RFID reader markets.

21 Generation 2
A User Guide
RFID Products powered by ThingMagic are available from ADT-Sen-
sormatic, Omron, Zebra, RFID Global Solution, Venture Research, CIT
& other qualified partners. ThingMagic, Mercury4 and ‘Reads Any Tag’
are protected marks of ThingMagic. Other marks may be protected by
their respective owners.

© 2005 ThingMagic, excluding SDR definition & history