Is Laboratory Testing Relevant?

By Ari Juels

It's getting a bad rap lately, but testing RFID products in a controlled environment is the best way to learn how things work and gain important insights.

  • TAGS

By Daniel Deavours

At a recent conference, I saw a video presentation by a vendor that showed a high-speed conveyor carrying tagged products, a forklift carrying a palette of tagged product (paper towels, of course) through an RFID reader portal, and several people in long, white lab coats wearing safety glasses and scribbling notes on their clipboards. They were apparently testing RFID products in laboratory conditions. We all know that factories and distribution centers aren’t nice, clean places, that factory workers don’t wear long, white lab coats, and that most products are far more difficult to tag than paper towels.

This kind of image gives laboratory testing a bad reputation. If you prove a tag works in those conditions, what does it say about working in the real world? Not much. But don’t be so quick to discredit laboratory testing.

The purpose of scientific experiments is to understand the relationship between input variables to a system and measured outcomes. Let’s say you want to know the relationship between tag performance and the distance between the tag and reader. You could set up an experiment in a manufacturing plant where you would vary the distance between the tag and reader and hold everything else the same. But if you change the distance and at the same time the background RF noise changes by some unknown amount, what can you learn from the results?

Laboratory testing is simply an attempt to control as many variables as possible. If you can’t control the variables, such as RF noise, there’s little ability to learn anything, and there’s a chance you may actually arrive at wrong conclusions. If you can control the variables, you can understand the relationship between the variables and measurements and gain important insights.

When people refer to performance “under laboratory conditions,” what they often mean is “under ideal conditions.” They are not the same thing. In the lab, I can create lots of nasty conditions for RFID. But I can carefully control those nasty conditions, so I can measure their impact on performance, while keeping everything else the same.

The real limitation to lab testing is that you can’t anticipate all the variables, and you probably can’t test all the permutations of the variables found in the real world. Lab tests can take you only so far. In the end, the viability of an RFID system must be proved in real-world conditions. At the RFID Alliance Lab, we do most of our testing in the lab, but we also acknowledge its limitations. That’s why we complement our laboratory tests with testing in real-world environments, to make sure the results we see in the laboratory are relevant.

The best approach to RFID testing is to do a mix of both laboratory and real-world testing. Laboratory testing is where you can test hypotheses and learn the cause-effect relationships. Real-world testing is where you can find what variables need to be tested and validate that the real world works the way you think it does. If things don’t work in the real world the way you expect them to, you can try to simulate those conditions under carefully controlled laboratory conditions to understand what is happening, form hypotheses about how to fix the problem and test the hypotheses to see if you’re right.

Daniel Deavours is research director of the RFID Alliance Lab, an rfid journal-supported test center that produces unbiased product reports.