From SCDigest's OnTarget e-Magazine
- Sept. 15, 2015 -
RFID and AIDC News: As RFID Tags and Readers Improve, Erroneous Reads become a Growing Problem Expert Says
Don't Think Smart Readers Alone Can Solve the Problem – It Takes Sophisticated Algorithms on the RFID Server
SCDigest Editorial Staff
The good news is that RFID tag and reader technology continues to improve. The perhaps bad and maybe even surprising news that comes from that progress: many times RFID readers are picking up the wrong tags in a read zone.
So says Tom O'Boyle, director of RFID systems for systems integrator Barcoding Inc., a frequent contributor to articles on RFID in Supply Chain Digest.
SCDigest Says: |
 |
O'Boyle also cautions potential RFID users to not fall for claims that today's increasingly smart readers can do all this work to validate a tag read. |
|
What Do You Say?
|
|
|
|
This article is a combination of some thoughts on this issue that recently appeared in a guest column by O'Boyle on the Manufacturing Business Technology (MBT) web site and subsequent additional insight that O'Boyle provided directly to Supply Chain Digest.
It problem starts with the significant improvements in RFID tags, readers and antennas that have led to greatly expanded read ranges.
"Now, a tag that could barely be read at 5 feet away from the antenna is read at 20 feet away," O'Boyle wrote on MBT. "At last, improved long-range reading was a reality. However, long-range reading produced a problem - a problem bigger than the problem of not reading the tag at all: reading the tag accidently."
It turns out that RF energy can't be controlled enough to limit the reading of unwanted tags simply by changing the reader power or limiting the receive signal strength of tags (RSSI). That's because read zone characteristics change regularly, making a fixed reduction in reader power impractical.
O'Bolye cited the example of a forklift as it drives through a dock door read zone. The metal characteristics of the forklift bounce the RF energy for a short period, creating read events from tags outside the reader's normal field of view.
So, what's a poor RFID technician to do?
"It is necessary to apply smart filtering in order to determine what a "valid" tag read looks like," O'Boyle says. "A "raw" tag read does NOT equate to a valid tag read; a valid tag read creates actionable data. Raw tag reads become valid tag reads if they fit the profile of a valid read event."
So if a reader picks up a tag that does not meet the criteria of a valid tag read then that read needs to be discarded by the system as simply noise, O'Boyle says. If that erroneous read goes undetected, naturally the consequences are not good: bad data that often requires manual manipulation to correct the error.
O'Boyle adds that in his experience, "Invalid read events that are passed as valid events essentially remove most of the ROI the RFID system was designed to provide."
Improving the filtering to discard erroneous reads takes a lot of work to develop the right algorithms for a valid read event for each specific read zone. Those algorithms must take into account many of the following parameters:
• Tag header: The first few characters of a Tag ID used to filter tags that only belong to a certain customer, group of customers, or class of product
• Read count by antenna: Number or read events that are capture by a single antenna
• Number of antennas: Minimum number of antennas that must see the tag to validate it is truly in the read zone
• RSSI: The larger, more positive the RSSI value, the stronger the signal, which is usually indicative of distance.
• Only movement from distinct zone to another distinct zone: One can ignore all transactions except when a tag moves from one distinct zone to another. The best way to describe this is a doorway. You might have a zone on the inside of the door and a zone on the outside. The system could ignore all tag read events unless it sees the tag move from inside and then outside in a likely timeframe. Then and only then is the transaction recorded.
(RFID and AIDC Story Continued Below)
|
O'Boyle provided SCDigest an example of how such a read validation algorithm might work, going back again to the example of a fork truck going though fixed dock door readers.