Related to:
Aug. 28 2007 04:07 PM

If you are like me, you read news items hoping to find at least one little nugget of practical wisdom that you can apply to your own projects; at least one hard-earned solution to a problem that a team discovered during those long nights and weekends that are always necessary to finish large projects.
 
Here are a few such tips that we have run across in our business of helping large manufacturers, distributors, logistics firms and retailers to add RFID technology to improve their operations. Perhaps one of these assorted (but hopefully interesting) tidbits can be applied to one of your own projects. If so, drop me a note of thanks, or better yet: tell me one of your own.
 

The 10 Solution

Here's a tip that can be used to stretch the capabilities of inexpensive tags, readers and antennas. By stretch, I mean that within the space of a couple of seconds, you are trying to read a number of tags (let's say over 100) while:
>> At the far edge of the tag's practical range
>> The items are moving quickly
>> The items are piled up on top of each other in a jumble.
 
When such a system is tested for the first time, you are likely to see read rates of about 95-98.5% (before tuning). It wasn't so long ago that such a result was thought to reflect well on the state of RFID technology, but now that we're using it to help us do things like get our shipping accuracy up to the 99.97% (five sigma) level, we need a much more reliable production signal.
 
At that point, you can begin to examine the choice of tag, the reader (and its settings), the antenna (type, number and orientation) and changes to handling procedures. The cost of the solution will then depend on which of these you end up changing and how long it takes to do so.
 
We recently came up with a 10 solution: the application of an additional tag to each item. The second RFID tag was placed on an orthogonal surface to the original one, and read rates jumped up beyond 99.98%. The application in this case required that the system read 300 items haphazardly piled together in a confined space in fewer than five seconds. No tuning of the reader or antennas was needed.
 
Your application obviously has to be able to afford the extra tag, but don't forget that the tags on containers, totes, dollies and cages are in recirculation, so their cost is divided over many uses.
 

The World's Worst Network

Are you deploying a distributed solution, one that needs to work across several or many locations? This could involve tracking the on-hand location of shipments to each of your customer locations around the country or spare parts throughout your operations.
 
When you test your solution, what should you assume for worst case behavior for the network? Normal testing procedures involve simulating extreme latency, data rate and packet rejection rates. You should do this sort of testing, but I'd also advise you to add one additional test case: The Worst Network in the World.
 
We recently deployed a distributed system for a $60B company in a site where network traffic leaving the building stopped for a full hour each day. Stopped. As in, no traffic. We discovered the hard way that, even on a gigabit LAN in a modern facility, reasonable assumptions for network performance were not enough. The system, which was built to eventually link over 2,000 similar locations, had to gracefully react to a network that disappeared for an hour at a time.
 
The good news here: if you design with The Worst Network in the World in mind, even your cellular-linked applications will be quite reliable.
 

Three Lines of Text

You don't see a lot of discussion about user interface when it comes to RFID, but it can be very important to the success of a project. The purpose of many RFID systems is to detect discrepancies of one sort or another  such as a package being loaded on the wrong truck. An RFID system can compare the items passing through a dock door to an electronic manifest. Then what?
 
You can dramatically improve the impact and effectiveness of your system with big, clearly labeled user buttons and simple, clear text instructions. Don't just light up a red light or sound a buzzer when a mistake is detected: tell people how to fix the problem. Three lines of simple text on a big highly visible LED panel can eliminate the need for rework and hours of worker training.
 
My company recently RFID-enabled the newest distribution center for a major retailer that has made it a mission to employ workers with disabilities. In doing so, the company has made a surprising discovery: the same investments in process automation and clear user interface design that assist workers with disabilities have also been a tremendous help to workers of average abilities. In fact, this particular distribution center is 20% more efficient than any other distribution center in the company very impressive for a company that is already an industry leader in efficiency as an organization.
 

Reverse Tagging

If you need to track the exact location of relatively small number of items (expensive or critical assets, for instance) consider a reverse tagging approach. This is an approach where you mount an RFID reader on the item to be tracked and you put RFID tags everywhere that you want to consider as a tracking location. This is the opposite of the usual approach in which the reader stays fixed and the tags move.
 
In the reverse approach, the reader has to be mobile, of course, and needs a wireless link to the network. Unless the object being tracked has its own power source (e.g. a forklift), the reader will also need its own battery. A variety of such devices are on the market today from Motorola and others, with cost ranging from $1,500 to $5,000.
 
The tags need to be appropriately protected for the environment. In some cases they can be plastic laminated, in others they will need to be molded into rugged pucks. In any case, they will cost between $2 and $10 per finished tag.
 
One nice characteristic of this approach is that the infrastructure cost of the system is more directly matched to its capacity than typical RTLS approaches. If you have one item to track, you have one mobile reader. Two items? Two readers. The entire infrastructure cost of other approaches is typically incurred on the first day of use, even if the business needs to track just one unit.
 
Another way that this approach more closely matches reality is its means of customization. A new tracking location can be typed into the system for a tag (Aisle 27) and then that tag is stuck where it belongs. From then on, when an asset passes by the marker, its location will be reported as Aisle 27. Position-based RTLS systems typically either display coordinates or highlight position on a map that requires a fairly sturdy investment of design, setup and calibration. This can be cumbersome and awkward to maintain. If a new aisle is squeezed into a distribution center, a good deal of work is needed to redo the software graphics and mappings.
 
The map might look nice in a product demo, but you're probably better off with a big three line display that says Aisle 27, aren't you?
 
John Beans is the vice president of marketing for Blue Vector Systems. Contact him atjohn@bluevector.com.

Follow