Optimizing Large-Scale Deployments

In RFID networks, there is strength in numbers. The best way to optimize a large-scale deployment is to look at the common configurations and start collecting performance data on those various scenarios.

Even after years of deploying the most complex RFID networks in the world, I have found that there are usually four or five basic reader configurations that are indicative of everything you will deploy. Keeping data on what the various configurations are and how they perform is the key to optimizing large-scale deployments. This is something that we used to do manually at ODIN, but now have automated tools to do. If you are building your own RFID network, it's something you can easily track by using Microsoft Access or even Excel. Think of your RFID deployment as a football team. There are probably 25 specialty positions that a typical professional football team may define, but at the end of the day it comes down to three or four key functions-really big guys who need to stop other people, lightning fast guys with good hands, a guy who can throw something, plus the occasional guy who can kick that same thing. Now Bill Belichick of the World Champion New England Patriots probably has 200 pages detailing a left inside guard vs. a right defensive tackle, but at the end of the day the requirements can be sorted out into those four functions.

RFID networks should be sorted out by base configurations. The most common ones are standard dock doors, conveyors, personnel doors, and stretch-wrap machines. Keep track of how each one of those are configured and set up. If you have 30 or 40 dock doors set up, even though there may be some differences, you should have a very high correlation as to how each one of those readers performs. You can use configuration changes to determine the optimization of a particular portal class. You can do this by creating a baseline performance for each interrogation zone type, and experimenting with new configurations and hardware by comparing with the existing performance. Just remember to test to statistical significance with the reader under test, so the data are meaningful.

The other way of optimizing a large-scale RFID deployment is to look at dependencies in the business process and set performance standards based on those dependencies. Think of a typical distribution center as an example. After items go through a dock door read zone, they usually are loaded onto a conveyor and pass another read zone, and then are often run through a sorting system read zone. Each one of these interrogation zones should have a very similar read rate of the same tag numbers. There should be a high correlation of specific tags in this business process flow. If there isn't-if one particular interrogation zone is significantly lower than the others in that same work flow-there is an opportunity to optimize that performance.