CompTIA began the RFID+ certification program because many of the DoD, Wal-Mart, and Target suppliers were looking for someone who knew about RFID to help meet a mandate. Most of those suppliers are looking for information about UHF. So I'll devote the next section to what both CompTIA and many of the large users of RFID are focusing on: the UHF Generation 2 (Gen 2) reader.
RFID was branded as overhyped in 2003 when Wal-Mart first made its announcement requiring suppliers to use RFID tags. Unfortunately, much of that ballyhoo still continues, and that is part of the reason CompTIA is trying to create a base-level standard of understanding. It is clear that Gen 2 far outperforms Gen 1 and creates a unified global specification. This is particularly true now that ISO standard 18000-6 interoperates with Gen 2. However, some RFID manufactures are overstating the true performance of Gen 2 quite unnecessarily. End users are generally forced to rely heavily on media sources and marketing from hardware vendors when shaping their opinions about RFID. Unfortunately, this approach seldom produces winning results.
There is a particular mode of Gen 2 that allows many readers operating in close proximity to not interfere with each other as much as they were before this new dense reader mode was written into the protocol. Gen 2 dense reader mode does a nice job of addressing the problem of reader-to-reader interference within a dense reader environment. To illustrate, reader A cannot use the same channel that reader B is using to receive a tag response if both readers are set up for dense reader mode.
Gen 2 is leading the industry toward improved security, friendlier dense reader environments, and rapid tag interrogation. Unfortunately, the number of possible tag reads has been embellished by at least two fold. Today's readers far outpace their Gen 1 counterparts, interrogating as many as 50 tags per second. But don't make the mistake of relying on media sources and vendors claiming Gen 2 read rates of 1,600 tags per second, which is possible in theory only. Simply Google "1,600 tags per second" to see some of the vendors' claims.
Tag vendors would have you believe that the reader is only half the equation. Dense reader mode does not address the question, however, of interference at the tag. How does Gen 2 tag silicon (which employs relatively unsophisticated filtration) handle several readers hopping on the same channel? Some vendors have claimed that Gen 2 silicon can discriminate between competing readers and respond to only the strongest signal.
At ODIN Technologies, we decided to put this claim to the test by placing two readers at 4.5 feet and 9 feet away from the same tag at the same power level. Leveraging our licenses granted by the FCC, we programmed each reader to transmit at a single fixed frequency to see whether the tag would respond to only the strongest reader signal. Not only was the tag unable to identify the stronger reader, but it failed to respond altogether. The reader interference caused a disabling conflict at the tag, making it clear that silicon and Gen 2 design are not as advanced as some manufacturers claim.
There are five critical tests that can determine the performance of a reader:
Power output analysis
Receive sensitivity testing
Interference rejection testing
Tag acquisition speed
After years of testing readers at ODIN Technologies, we learned that these five critical parameters need to be isolated in a scientific methodology and compared before doing actual use-case analysis. All of these tests should be performed in an experienced laboratory, so that the isolation of each variable being tested can be assured. An anechoic chamber is often where much of the testing takes place. However, you need to make sure that some tests are purposely performed in a real-world environment to find out how the reader responds outside of a lab.
ODIN Technologies has performed hundreds of thousands of tests and published the results in our RFID Benchmark Series. This series goes into much greater detail than this book about the tests performed and how they were done.
Distance testing is a simple metric that end users have used for years to compare reader performance. By setting up two readers in the same location and slowly backing a tag away from the readers, an end-user can determine which reader has the best read distance.
This sort of testing, however, offers limited value, because it is impossible to decouple the core drivers of performance over distance (transmit power accuracy and receive sensitivity, for example), which are covered in the other tests. Additionally, in tons of deployments, I have yet to deploy a dock door of greater span than 10 feet, making read performance at 15 feet of limited value. Still, distance testing is helpful for comparing the performance of today's readers with those of several years ago and provides an easy-to-understand method of discussing reader performance.
To perform this test, place 20 tags in a test grid format on a corrugate substrate. The grid should be placed at four distances: 3, 5, 10, and 15 feet from the interrogation antenna of the reader under test (RUT). The total number of tags read at each distance should be recorded as well as the number of reads for each tag over a 30-second time interval. Make sure no other readers are active during the test and make sure that you don't compare results based on read rate, rather on relative performance, because each reader vendor uses its own software to record read rates.
This simple yet important test is often overlooked by integrators deploying RFID, since they tend to trust the reader manufacturers. However testing the level of power that each reader port emits is critical prior to any RFID deployment.
Power output analysis is at the root of potentially putting an RFID reader company out of business if they violate the FCC rules by broadcasting more energy than legally allowed. The issue of power management is often overlooked by end users who are unfamiliar with regulations and do not have the scientific equipment required to take power measurements at the port or in the far-field.
Effective power management is critical to RFID reader performance because it determines the range and consistency of tag capture. Overdriving or going above legal limits of reader transmit power is tempting for reader vendors because extra power often achieves better results in a single portal situation. For those readers providing Gen 2 compliance, power management is particularly important because dense reader mode (DRM) has specific power management requirements that contribute to better performance.
To create this experiment, you need to have a real-time spectrum analyzer (RTSA). The RTSA provides highly precise measurements and spectral triggering not common to other less-sophisticated instruments. Standard cabling fabricated for this testing should be used. Power measurements at the reader port are taken by using a fixed 10 dB attenuator to avoid over-loading the RTSA. Radiated power measurements are then captured at 5 feet from the interrogation antenna by using a low-gain antenna.
The first power measurement you investigate should focus on peak RF power output at the reader port. These measurements should be taken with the reader in a factory default setting; no special configuration commands should be used. The FCC requires that power emitted by an "intentional radiator" not exceed 30 dBm. Some vendors have interpreted this to mean that the power output at the reader port shall not exceed 30 dBm. Others have assumed this to include losses due to cabling, so they drive the port above the 30 dBm mark based on the assumption that several decibels will be lost in the cabling. In practice, it seems that the FCC is ultimately concerned that the radiated power not exceed 36 dBm, which includes a 6 dBi gain antenna.
In addition to peak power, end users should understand frequency hopping and the channels of reader transmit power. The FCC regulation (part 15.247) states that the reader needs to hop from channel to channel pseudo-randomly over at least 50 channels within a 10-second time period. Transmit power is one of the most important factors in defining success between a reader and tag(s) in almost any use case. Readers that enable higher power output to compensate for losses due to cabling will be more successful in the field if they are to be operated at full power. Although this capability does require that power measurements be taken at the end of each cable to ensure FCC compliance, it guarantees that the best performance possible will be realized.
The receive sensitivity takes into account one of the other big factors in RFID system performance: the tag. Tag consistency has a profound impact on performance of an RFID network. Unreadable tags often lead to expensive exception-handling processes and slow supply-chain throughput, so getting the best read rate of weak tags can be a big time and money saver. Until production quality improves, it's up to the readers to make up for the shortcomings inherent in the poorer-quality tags. Tag quality is a common and known problem, but it is only part of the equation; the reader's receive sensitivity or how well it can "listen" for tag responses is another key factor.
Every reader design team faces the formidable challenge of designing a system that transmits 1 watt (continuous wave) to power a tag, while simultaneously listening for a faint tag response that is 1 millionth of that power as strong. Success is driven by the ability to carefully filter the tag response from the transmission signal so that these two signals do not interact destructively within the reader. It's kind of like trying to hear a pin drop at a KISS concert.
As you might have guessed, because this is one of the more formidable challenges that a reader overcomes, the testing of receive sensitivity measurements is nontrivial. Many readers today transmit and receive on the same RF port making the process even more difficult and less accurate than having a dedicated send and receive port. This characteristic is known as "mono-static" antenna sequencing and requires specialized RF engineering equipment to separate received tag responses from the transmission signal to determine the receive sensitivity. This is one of the tests that is easiest and best to outsource to an RFID laboratory.
If you do have a lab perform the test, knowing the interpretation of the results is germane to designing a great-performing RFID system. A 30 dB difference in receive sensitivity is identical to saying that one reader is 1,000 times more sensitive than another. The stark contrast displayed in testing the cross section of today's readers demonstrates how different one reader is from the next, and those types of variations routinely have been recorded in the ODIN labs.
Keep in mind that a high receive sensitivity can be both a blessing and a curse. The reader may be optimized for interrogating a single tag, but how effective is the reader at filtering out additional readers and tags, which is common in a field deployment? Given the option, however, the more sensitive the reader, the better the performance.
The next area to test is the level of interference that a system with many interrogation zones is capable of withstanding. This type of system is usually called a dense reader mode (DRM) environment. DRM was created to deal with implementations that will continue to scale as the industry unfolds over the coming years. End users are beginning to understand the value associated with deploying RFID throughout an enterprise, migrating RFID implementations beyond the warehouse into manufacturing, and raw-materials tracking. Other companies are simply adding readers to their existing RFID network as their volume of tagged products and assets grows.
When additional interrogation zones are brought online, existing RFID reader infrastructure faces growing interference pressure. Many end users are overlooking an important question as they select RFID technology components: "Are the readers I'm deploying today sufficiently robust to reject the interference generated by my planned and unplanned RFID infrastructure 2–5 years from now?" I always advise end users to consider this impact when designing and deploying RFID solutions at the beginning-design with the end in mind. You do not want to be forced to replace your whole RFID network because it cannot scale to meet your or your clients' needs over time.
As I mentioned, DRM has been designed as part of the Gen 2 specification to help readers and tags address the problem of growing numbers of readers in the same environment. Although many readers are dense reader mode compliant, some implement the protocol far more effectively than others. Interference rejection testing is designed to study the effects that interfering reader signals have on the performance of a RUT. This test should be performed with a single tag attempting to be read, and an increasing number of interfering readers added to the environment. By pointing the interferers directly into the RUT, the full interference rejection capability of that reader can be determined.
Dense reader mode is often thought of as the most important feature of a Gen 2 reader, especially by retailers who are actively deploying 40–60 units per store and 100+ units at distribution centers. Product manufacturers will quickly face these same issues as they continue to scale their RFID footprints to meet increased shipping volumes and the need to track tagged goods at more read points. There are also many examples of RFID installations in manufacturing environments that include dozens or even hundreds of RFID readers in a single facility. As many of these implementations migrate to UHF RFID technology, dense reader mode will be a critical component of project success.
The final test of the five scientific evaluations is for acquisition speed. This refers to how fast tags can be read by a reader.
The ability to interrogate large tag populations is critical to many applications, but not all. As I've been alluding to (unfortunately, for unsuspecting end users), not all readers are created equal. Some of the people who designed the EPC protocol-including Dr. Daniel Engels of the University of Texas, who contributed to this book-have said there is a theoretical maximum of 1,600 tags per second. In reality, this number is just not achievable with today's technology. However, reading large-enough populations of tags for most business uses is possible today.
Real World Scenario-Nothing Runs Like a Well-Tuned RFID Network
A few of our engineers at ODIN were hired to deploy a work-in-process tracking system for a well-known U.S.-based industrial machine manufacturer. A primary concern when we designed the system was using a reader that could scale as the manufacturer moved from a handful of readers up to hundreds of readers. The testing we did around interference rejection will allow them to add readers without degrading performance as they expand to meet the needs of their business.
If you think of an application for which tag acquisition is critical to success, it is reading cases on a pallet coming through a dock door such as a common dock door portal, or aggregating cases to a pallet and associating them to a specific pallet tag (for example, at the end of a manufacturing process), or reading runners at the start or finish of a race. If you're concerned about runners, you don't need 15 feet of read range, but you do need acquisition speed.
At ODIN technologies we've deployed systems which are designed to read 320 cases on a pallet. To get those kinds of results, careful component selection, design, and implementation are critical.
All readers implement an "anticollision" algorithm to define how a reader will interrogate more than one tag in the field. Although the Gen 2 air interface protocol defines an anticollision sequence known as "slotted random anticollision," many design variables are left to firmware engineers regarding its implementation. The speed of tag interrogation is driven primarily by the unique anticollision algorithm developed by each vendor.
To establish the limitations of each reader's anticollision algorithm, it is best to place a tag grid composed of 400–500 tags in the interrogation zone of the RUT. The RUT should then be placed in continuous read mode for five seconds and the total number of unique identifiers captured. This simple but effective test should be repeated five to ten times for statistical significance. It is important to the results that you normalize the data so that you will not be thrown off by the reader manufacturer's software. I've mentioned this in other sections of the book, but one manufacturer's "read" can often be very different from that of another manufacturer.
Now that you have the basics of scientific testing and you may know the top two or three readers that will meet your needs, the best thing you can do is set up a mock-up of the specific environment you'll be testing, or actually test those readers in that environment. If you are planning on reading all the cases you put on a pallet, think of the best place you can capture all those reads. The amount of time each case sits on the pallet while being stretch-wrapped and the change in tag orientation relative to the antenna all make the stretch wrapper one of the best locations. This should be one of your use cases.
Each time I design an RFID solution, I leverage scientific data similar to the results outlined in this section. This provides an easy-to-defend scientific rationale for supporting technology-selection decisions. The scientific tests are specifically designed to isolate a single variable (power output, receive sensitivity, and so forth) so that readers can be compared. This points to a short list of viable reader options. The next step is to test the top readers in the use case that will be deployed as a final proof of concept.
Scientific testing does the heavy lifting when it comes to testing anything RFID. However, a little peace of mind verifies your testing methodology and can confirm a thesis based on the scientific testing. It can also help put one reader ahead of another if the scientific testing results are similar.
The most common uses for UHF RFID today fall under three simple applications, or use cases:
Certainly no one industry, such as fast-moving consumer goods, is going to have a use-case scenario that fits everyone's needs. But given the focus of many of today's RFID adopters, it at least makes sense to start your design process thinking about these parameters. The three have very different requirements largely based on the dwell time (amount of time the tag is in the interrogation zone) and the number of tags being read at any one time.
The conveyor use case is one of the most widely deployed to automate internal sorting processes as well as to conduct inline verification of applied RFID tag performance. Even more commonly, it is used in the process of capturing material receipt information, particularly among major retailers. Although not the most challenging use case, it is among the most important. This testing should be conducted by using either a fast-moving conveyor (up to 600 feet per minute, or fpm) or, for superior control over object orientation relative to the interrogation antenna and to ensure scientific accuracy, a device that can accelerate the product above 600 feet per minute in the interrogation zone while ensuring dwell time is around one second. What you will find is that even the middle of the speed range (400 fpm) is fast by most standards, and there are only a couple of readers that can actually perform at 600 fpm. The readers that do not perform well at this speed are likely to be those that conduct heavy filtering on board. This is telling, because if a tag is read even once, it appears in the results. The reader's activity and reporting to the applications can sometimes not be in real time, this is said to be a synchronization issue. It is a discrepancy between the time at which the reader is searching for a tag and the time that the tag is actually present.
Next is one of my favorites for deployment, the stretch-wrapper use case. It has become an important point in manufacturing and warehouse operations to verify tagged cases and associate them to a specific pallet. An interrogation zone at the stretch wrapper is often employed to save processing time and maintain throughput levels after RFID is introduced into operations. Figure 5.2 is a sample ODIN Technologies' stretch-wrapper deployment with the rack cover removed to show the three-antenna configuration.
Figure 5.2: Stretch-wrap test setup
This testing is quite simple to conduct. A standard pallet is placed on a stretch wrapper and rotated three times. The number of reads captured for each unique tag is recorded. The trial should be conducted five times for accuracy, and the data should be normalized to make up for reader vendors' software differences in interpreting a tag read. This analysis makes it easy to identify how each reader compares in this use case and whether each reader has an affinity for a particular tag type (which they usually do, so having your tag testing done first is essential). If tags are being shipped to a third party such as a distributor or retailer who will also be attempting a tag read, then it will be important to select tags that are likely to perform well with that reader. This is in your best interest because you directly benefit from higher read rates and better data quality at your downstream supply-chain partner. Finally, if you are a distributor or retailer or do not have control over which tags are used, you will want to look at readers that perform consistently across tag types. This will allow you to maximize your read rate performance regardless of what type of tags you start receiving into your facility.
The adjacent dock door use case is where end users from many industry verticals are harvesting business value from fixed RFID portals. The value proposition ranges from automated receiving to ship verification to more-efficient put-away processes. Yet with this value comes complexity. Distribution centers outfitting every dock door with RFID leads to a serious level of RF noise that leads to a high likelihood of two readers activating the same channel simultaneously and missing read opportunities. One size does not fit all when it comes to designing and deploying a dense reader environment.
In addition, adjacent dock doors lead to confusion of a different type called cross talk. Cross talk occurs when a reader in portal A reads case tags moving through portal B. Resolving this issue, while ensuring a robust use case, requires special tuning tools to optimize reader configuration, leading to the best performance possible. The other way to approach this testing is manually, by employing a signal generator and spectrum analyzer.
Testing for adjacent dock doors requires at least five of the same type of reader. The RUT should be placed in the center portal and four interfering readers placed on the outside of the two adjacent dock doors. This setup will simulate the effects of three adjacent dock doors. This testing is similar to what has been conducted by major retailers as they have assessed RFID reader performance.
What you will find is that even if a reader is Gen 2 compatible, only a handful actually perform adequately in this test, for a large-scale RFID deployment. Bear in mind that as additional interfering readers are brought online in your environment, reader performance will continue to degrade. Scaling an RFID implementation is by definition expensive and difficult to reverse. If ever there was a time to leverage physics in your decision-making process, this is it. I always recommend that testing be conducted in accordance with the scale of your implementation prior to making a final reader selection decision. I have conducted testing similar to the preceding for clients who wanted the piece of mind that comes with measuring twice and cutting once.