Aim for the Worst

By Ari Juels

The performance of tags—even those of the same make and model—can vary greatly. To achieve high read rates, design systems to read the weakest tags.

  • TAGS

By Daniel Deavours

As a scientist, I like predictability. I’d like to be able to tell RFID end users they can run tests and get results that can be quantified and reproduced. But after spending several hundred hours running tests on UHF RFID tags for the RFID Alliance Lab, a nonprofit test center based at the University of Kansas, I’ve concluded that the first law of RFID is that there’s precious little predictability.

The lab’s tests revealed that not all Class 1 Electronic Product Code tags perform the same. Some Class 1 tags are three times faster than others. And while all Class 1 tags can be read faster than Class 0 tags, the slow-performing Class 1 tags were read only slightly faster than the Class 0 tags. The best-performing tags in free air (ideal conditions) are Class 0 tags, but not all Class 0 tags outperform Class 1 tags. In general, tags can be read more reliably as they get closer to the reader antenna, but some Class 1 tags actually show a decline in read rate when they get too close.

So here’s my advice. In addition to taking into account the tag’s orientation to the reader, the materials your product is made of and packaged in, and the tag’s placement on your product, you also have to design your system to account for variations in the performance of seemingly identical tags. In particular, consider the following issues.

Variation: Tags of the same model number from the same vendor are not identical. It’s not clear why, but we’ve found variations among tag models of more than 10 dB, which means that read distances may vary by as much as a factor of 3. You need to place your readers close enough and move your tagged products slow enough to be able to read even the worst-performing tags in a batch.

Yield: It’s possible that 5 percent or more of the tags you receive from a vendor will be dead on arrival (can’t be read, period). Some vendors have begun removing or marking dead tags, but we still find tags that are nonfunctioning. Also, depending on the type of tag and the manufacturer, 10 percent or more of tags can be “quiet” or “weak.” That is, they are not dead, but they are readable only at very short distances in ideal conditions.

Durability: Tags not only have to function when put on a case or pallet, they also have to survive the shipping process. Friction, mechanical shock, electrostatic shock, extreme temperatures and other factors can cause a tag to degrade or stop functioning. (The RFID Alliance Lab plans to study this issue in the near future.)

If you’re planning an RFID pilot test to determine which tag to use, focus on tags that work best near the materials your products are made of and packaged in. Make sure you manually remove dead or quiet tags from rolls or use a label applicator that can reject these tags before they are applied to products. And set up your readers to read the worst-performing tags in a group, not a typical or average tag. Otherwise you may find that as you begin tagging products, 5 percent or more of the tags won’t be read.

Daniel Deavours is research director of the RFID Alliance Lab, an RFID Journal–supported test center that produces unbiased reports. To comment on this article, click on the link below.