Reproduction permitted for personal use only. For reprints and reprint permission, contact email@example.com.
On its surface, radio frequency identification technology may seem simple: A data reader gathers information from an information-embedded chip, with an antenna enabling the transfer of data via the radio frequency.
The physics related to the antenna is where things can get complicated and challenging. Overcoming those challenges is a key to bringing the price of transponders or tags down to the often-mentioned five-cent level - a level that market analysts say is necessary before RFID will become truly ubiquitous.
Understanding the fundamentals of signal-to-noise ratio (SNR) is important to uncovering ways to create less expensive tags and more reliable tag-to-reader communication.
In general, SNR at the tag level will be improved with more efficient antennas, since the tags communicate with readers by their chips modulating the radar cross section (RCS) of the antennas. Illuminated by a reader's pulse, a tag's chip powers up and begins to cycle through its unique code, modulating the load across the tag's antenna terminals, which in turn appears as a signal at the reader's antenna.Modulation, efficiency are important
The strength of that modulation and the absolute signal returned in large part determine the SNR of the read. Think of a signaling mirror held by a person on a distant mountaintop. The degree to which the signal can be seen is proportional both to how much the mirror is being turned (modulated) and by how clean it is (its efficiency). A tag works the same way.
Like a signaling mirror, in general, a larger antenna gives a larger RCS. Of course, there are caveats, such as the need for the antenna to be resonant at its wavelength of operation, but at 915 MHz, that wavelength in free space is well beyond a foot long, so most antennas are resonant at a fraction of that (half or less).
Furthermore, the antennas can be either electric- or magnetic-field dominated, and that will affect their performance on various substrates, particularly metals and liquid-filled containers.
Finally, interference among tags as well as from surrounding objects can cause the equivalent of fading, an effect much like what you hear when your FM radio is tuned to a weak station as you drive through a city.
For some applications, larger antennas, embedded into labels or wrapped around bottles, might be feasible, if a bit unconventional. Special designs using conductive ink are already making their way from laboratories to test benches, and those designs will also lower production costs over copper-foil antennas, the current tag standard. Conductive inks are not yet as efficient as copper foil, however, and measurements made at the University of Wisconsin-Madison (www.uwrfidlab.org
) confirm the greater resistive losses of ink vs. copper that one would expect. Improved formulations (province of the ink companies) and better antenna designs (our own lab's goal) will begin to close the gap in performance.
Performance in RFID systems is not simply due to the tag's design and construction, however; one of the most significant variables is the application (package, placement, packing) itself. In contrast to the current approach, optimal RFID systems will not be "one size fits all" in the future but, rather, customized to fit the particular application. A tag on a can of soda will be very different from a tag on a gallon of milk; both will be distinct from placement on a garment or a package of frozen meat; those all assume that tags can be made cheap and durable enough to justify item-level deployment.
Thus, to fully analyze tag performance, we need to study both the antenna design and its environment. Computer-based electromagnetic (EM) simulations of standard tag antennas on a variety of substrates done by Jorg Yen, a UW-Madison Ph.D. student, show the dramatic effect of antenna orientation and loading, what the substrate does to the EM fields.
Still, we must make sophisticated measurements of those antennas to confirm our predictions. Such measurements are done using RF instruments like network analyzers and performed in anechoic chambers that both shield out other interfering signals and dampen reflections from the test antennas to approach the ideal of simplicity from which we can always make things more complex.
We can now measure antenna performance on single products or in arrays of packaged ones, studying their mutual interference. We can apply special measurement fixtures to determine the effect of antenna placement and environment independent of the chip being employed. That helps us gain quantitative and, thus, predictive power in the sometimes-murky world of real-life RFID applications.
Indeed, the goal of our work in RFID systems is to understand at a more scientific level what the various effects of antenna design, placement and construction are, so that we have an improved ability to predict the optimal configuration for any given need. Without that rigorous approach, improvements in RFID systems at the physical level will be gained through empirical "cut and try" methods that are unlikely to translate across product lines or tag generations.
General guidelines for tag placement are being developed, notably by MIT's Auto-ID Lab, and those can be consulted in the absence of specific measurement data and detailed analysis from our UW-Madison lab.
RFID systems have indisputable promise, and their widespread deployment is no longer a question of "if" but, rather, "when."
The conceptual simplicity of identification from a distance without the line-of-sight requirements of barcodes opens up a dizzying array of applications, yet that superficial simplicity belies an underlying complexity that arises in large part from the nature of RF wave propagation itself: absorption, reflection and interference that can confound reads in otherwise valid applications.
It is the goal of our laboratory to understand the physical layer, the antennas, RF propagation and the RF environment, so that we can advance the state of RFID systems. We want to turn the clock forward as rapidly as we can without too much expensive blind experimentation.
Daniel van der Weide, Ph.D., is a professor in the Department of Electrical & Computer Engineering at the University of Wisconsin-Madison. He can be reached at firstname.lastname@example.org