Reading between the lines: RFIDs confront the venerable bar code
Three decades ago, a 10-pack of Juicy Fruit gum and a cashier at a Marsh supermarket in Troy, OH, were the participants in the first successful test of what we now know as the UPC (universal product code) bar-code system. Evolution of the bar-code system continues; for example, US and European standards-group representatives recently agreed on a common 14-digit format that, beginning in January 2005, bar-code readers worldwide must support. But, all in all, bar codes today are mature, pervasive, and well-understood. (Some ex-presidents may beg to differ on that last point, though. Remember George HW Bush's befuddlement when, on the 1992 primary-campaign trail in New Hampshire, he unsuccessfully attempted to use a bar-code scanner in a grocery store?)
Technology marches on, though, and an up-and-coming contender to the product-identification throne has emerged: the RFID (radio-frequency-identification) device. Ironically, RFID technology is almost as old as bar codes, which in 1934 received their first patents. Great Britain's Royal Air Force employed RFID-like techniques to distinguish between friendly and enemy incoming airplanes during World War II, and Harry Stockman's October 1948 treatise, "Communication by Means of Reflected Power" in The Proceedings of the IRE (Institute of Radio Engineers) first detailed the theory and implementation of RFID. Prolific inventor Charles Walton in 1973 received the first RFID patent for a passive RFID-based door-lock reader. Walton coincidentally shares the same last name as the late Sam Walton, the founder of Wal-Mart, which, along with the US Department of Defense, has played a leading role in spurring current RFID deployments.
If RFID is such a timeworn technology, why then has the interest in it accelerated so dramatically in the past few years? Part of the reason is chip capability; thanks to Moore's Law, passive RFIDs sell for less than 50 cents in high volumes, and analysts predict they'll sell for less than five cents in high volumes by the end of this decade. Adequate infrastructure capability is also important; the dot.com explosion of the late 1990s fueled the development of networking equipment and powerful servers with speedy CPUs and I/O connectivity and containing ample memory and hard drives, and the subsequent dot.com implosion has resulted in copious underused network bandwidth begging for someone to harness it.
The final piece of the interest-in-RFID puzzle comes from customers' needs. Manufacturers, distributors, and retailers all want to as much as possible automate their systems to eliminate expensive and unreliable human beings from the process, and they aspire to have timely and accurate insight into the location of individual products at a given time and into various product-staging locations' inventories. If possible, they'd like to extend their insight beyond the store, to link each product with an individual consumer, and, in combination with other collected data, to ascertain the means by which they can lure that consumer into buying even more (see sidebar "Privacy concerns"). Governments, too, have an interest in learning as much as is legally possible about what their countries' citizens and residents are up to.
As is the case with nearly every budding application in its infancy or adolescence, a diversity of incompatible options has emerged to address the various challenges that invariably arise (Reference 1). With RFID, these differences begin with the fundamental means by which the RFID tag communicates with its reader. A passive RFID tag contains no power source of its own. Instead, the reader powers, or "harvests," power using either inductive coupling or electromagnetic capture in a process in which the reader "excites" the RFID tag. An active RFID tag, conversely, includes a battery, substantially increasing its cost but also potentially enhancing its functional capabilities, along with its operating range. A semipassive tag, an intermediary approach, runs the chip's standby circuitry from a battery but draws power from the reader during active communication sessions.
RFID readers and tags also broadcast and receive using a diversity of frequency ranges. Low-frequency RFID systems operate at 125 to 134 kHz, for US and international use, respectively, and 13.56 MHz, another international standard, is the most common high frequency. UHF (ultra-high-frequency) RFID systems range from 866 to 960 MHz, and microwave systems operate at 2.4 to 5.8 GHz. With all other factors being equal, high-frequency RFIDs have longer range than their low-frequency counterparts, fundamentally because near-field effects don't degrade high-frequency RFIDs' signals. If a tag is less than one wavelength away from a reader, the signal decays with the cube of the distance; beyond one wavelength, the signal decays with the square of the distance. High-frequency RFIDs can also more quickly transmit and receive data.
Conversely, high-frequency tags and readers are more expensive and burn more power than their low-frequency peers, and environmental factors, such as packaging, moisture, and nearby metallic items, adversely attenuate high-frequency devices' signals more than their low-frequency counterparts. You also need to be aware, when specifying and designing RFID gear, that a frequency freely usable in some parts of the world may not apply in others or may require an expensive and time-consuming licensing process. Unlicensed frequency bands are also subject to spectrum corruption, a factor that anyone who has tried to simultaneously operate a microwave oven, a cordless phone, a Wi-Fi access point and client, and a Bluetooth-paired set of equipment has experienced (Reference 2). How does the RFID tag modulate its data on the carrier frequency it sends back to the reader? Again, there's no consistent strategy. AM (amplitude modulation)—specifically ASK, or amplitude-shift-keying, FM (frequency modulation), phase modulation, and PWM (pulse-width modulation) are all possibilities. To minimize the probability that two tags may simultaneously broadcast, thereby corrupting each other's signals, manufacturers sometimes employ TDMA (time-division-multiple-access) algorithms. Infineon's 13.56-MHz, PJM (phase-jitter-modulation) RFID tags employ modulation techniques the company licensed from Magellan Technology. The techniques reportedly enable the tags to read and write at rates as high as 848 kbps—approximately 25 times faster than conventional 13.56-MHz tags; they also implement FTDMA (frequency- and time-division-multiple-access) and eight-channel frequency-hopping for anticollision-interference avoidance (Table 1).
CRC (cyclic-redundancy check) or other checksum codes can determine whether the reader correctly received a tag's transmission, and a plethora of ECC (error-correcting-code) schemes can correct bad bits and prevent the need for time-consuming rescanning. Currently, no consistency exists about whether the broadcast data is encrypted, although large implementers, such as Wal-Mart, are driving de facto standards and beginning to bring some order to the morass. Whether you choose to implement encryption in your design and how robust that encryption is can significantly affect the tag's cost, size, power draw, and other key factors.
One key advantage that RFIDs have over bar codes is that individual units can contain unique identifying data sequences; UPC codes, conversely, are generic to all units of a manufacturer's product. And, if rewritable memory, such as EEPROM, flash memory, battery-backed RAM, FRAM, or MRAM, contains the EPC (electronic product code), you can alter and append the EPC as the item goes through its manufacturing, distribution, sales, and usage life. How much data the RFID should store is a topic of much debate, and is to some extent application-driven. STMicroelectronics' XRA00 UHF RFIDs, for example, support EPCglobal Class 1 specifications. They contain a 128-bit memory organized as eight blocks of 16 bits each. The first block stores a 16-bit CRC value. The next six blocks store the 96-bit product code that the device uses during the inventory sequence, and an 8-bit "kill code" and eight lock bits that protect the memory contents share the last block.
Under the EPCglobal scheme, the RFID reader, after reading the EPC code from the tag, queries ONS (Object Naming Service) servers whose databases VeriSign administers. These servers, conceptually analogous to the DNS (Domain Name Servers) that translate URLs (uniform resource locators) into IP (Internet Protocol) addresses, return the IP address of a server that contains detailed information on the RFID-tagged item. EPCglobal, an outgrowth of the earlier Auto-ID Center, is one of two primary RFID standards-setting groups; the ISO (International Organization for Standardization) is the other. Clearly, the two groups need to collaborate their data-format-conforming efforts for RFID to become truly ubiquitous. On the other end of the memory-density spectrum are companies such as Boeing, which is testing much larger, 10-kbit RFID tags that enable the storage of long serial numbers, detailed parts information, and repair histories. The large-RFID-tag approach also applies whenever an RFID reader is not connected to a network and, therefore, cannot access the ONS in real time.
Another important advantage of RFIDs over bar codes is that laser line-of-sight orientation between a tag and a reader is unnecessary. Any alignment between tag and reader is also potentially irrelevant because proximity requirements depend on which antenna technology you employ. Circular-polarized antennas, like omnidirectional microphones, emit and receive radio waves in a circular pattern. Using them provides a better chance that the destination will receive a broadcast signal in situations in which the sender cannot precisely control the orientation between the transmitter and the receiver. Conversely, the operating range of a circular-polarized antenna system is less than that of a linear-polarized antenna, which is analogous to a unidirectional microphone.
US journalist, attorney, and motivational writer Napoleon Hill (1883 to 1970), the so-called Founder of The Science Of Success, stated that "Every adversity, every failure, every heartache carries with it the seed of an equal or greater benefit." Keep that quote in mind as you survey today's seemingly irreconcilable RFID landscape; sooner or later, it will inevitably sort itself out, and the companies that guess right will greatly benefit from that consolidation (see sidebar "Application teasers"). Aside from speed-boosting modulation schemes, such as Infineon's PJM, several other RFID-tag-differentiation opportunities exist. They include size, as Hitachi's 2.45-GHz µ-Chips exemplify. The chips, which contain embedded antennas, measure only 0.3 mm sq (Figure 1). Other differentiators include power consumption and communication robustness. (One factor currently limiting the adoption of RFID is that, in some field trials, it's no more reliable to read than are bar codes.)
Companies are also investigating various means of supplementing traditional-RFID functions with environmental sensors that can report such factors as tire pressure; temperature; humidity; the presence of various biological agents to determine contamination, spoilage, and the like; and whether someone has previously tampered with or mishandled the item by using excessive force or vibration, for example (Reference 3). The tag can report real-time data; alternatively, it can provide a simpler indication that the measurement has at some point exceeded a threshold value. Cost is, of course, perhaps the most important improvement factor that will broaden RFID's applicability. It makes little sense to attach an RFID tag to an item whose cost is comparable with or even within one or two orders of magnitude of the cost of the tag itself. For example, although a retailer may today be interested in RFID-tagging large cases of paper towels, RFID costs will have to significantly drop before the retailer will consider tagging individual paper-towel rolls. This cost-driven brake on adoption will, at least in the short term, act as a natural means of addressing RFID-privacy concerns.
The long-term potential for RFID ubiquity is bright, but the size and cost pressures that will increasingly affect tags lead to uncertain prospects that they'll fill many semiconductor fabs or that they'll be wildly profitable for their manufacturers. Why, then, are so many people so excited about RFID? The reason is simple: The worth of all of that data streaming off RFID tags and readers, in and of itself of little value, emerges when other hardware and software on the Internet or a company's Intranet stores, transports, and manipulates the data. As a result, many will benefit—CPU vendors, such as AMD, IBM, Intel, and Sun; their sibling systems divisions; systems partners, such as Apple, Dell, and HP; networking-equipment vendors, such as Cisco; and enterprise-software suppliers, such as Microsoft, Oracle, and SAP. The storage, transportation, and manipulation of that data also drives the fact that much of the recent media coverage of RFIDs has appeared in IT publications (see sidebar "Additional insights"). This data explosion will, of course, also enrich the fortunes of DRAM, hard-disk-drive, Ethernet, and other system-building-block suppliers.
For those designing RFID readers or implementing readers that others developed, the diversity of frequencies, formats, modulation, interference-suppression schemes, and other variables may motivate you, if your customers' cost expectations allow, to make those readers as flexible as possible. For the readers' digital subsystems, you can ensure flexibility primarily by enabling updatable firmware using code storage in flash memory or a small-form-factor hard-disk drive instead of ROM and by enabling updatable hardware using FPGAs and PLDs rather than ASICs. With the RFID readers' analog subsystems, you might consider implementing programmable analog arrays from companies such as Anadigm, Lattice Semiconductor, and Zetex, instead of hard-wired circuits (Figure 2).