Aug 02, 2010In Japan, delivery services handle more than 320 million packages annually, mostly for consumers. After pickup, the packages go to a central distribution center for sorting and grouping. From there, they go to a destination distribution center, where they're sorted into smaller groups, then on to branch distribution centers for delivery. During this process, it's not unusual for packages to be separated from their designated groups.
We could RFID-tag each package to identify it accurately and swiftly, but an operator or driver would need to check every item against a master list to know if a group were missing a package—or contained an extra one. Such a list would typically be available on an electronic data interchange (EDI) system that runs on a network. But there are many situations—in branch distribution centers or on the road, for example—where there is no network connection or workers don't have access to the list for other reasons.
At the Auto-ID Lab Japan, we wanted to develop a convenient way to check whether a group of packages is complete—for instance, should it include just 99 packages, or was one of 100 packages lost or misplaced? We developed a method called group coding of RFID tags. The first step is to create a unique identification for the group, based on each item's RFID tag ID or Electronic Product Code number. Then we write the group ID onto the user memory of every RFID tag in the group. A package may belong to more than one group, so, for example, we can establish a group for a destination center and one for a branch center.
To verify the integrity of a group, a worker could use a handheld reader to check each tag's unique ID and group ID. An estimated group ID would be computed automatically from the unique IDs of packages that have the same group IDs. Then, the estimated group ID could be compared with the original group ID, to determine immediately if there are any missing or extra items. By extending group coding, we could also determine the number of missing items. (This technology is similar, in principle, to forward error correction in wireless communications.)
Group coding could be applied to other RFID applications, such as ensuring the correct items are loaded onto trucks or received at distribution centers. It also could prevent shrinkage. In addition, instead of trying to achieve perfect read rates at fixed RFID portals, we could determine that a group of items is valid if we can identify, say, 97 percent of its expected contents.
Jin Mitsugi is an associate director of the Auto-ID Lab Japan at Keio University and an associate professor in the faculty of environment and information studies. Yuki Sato is a junior at Keio University.