Turning somersaults to make RFID work at UHF
Juin 17, 2005
In the world of RFID, Ultra High Frequency (UHF) means license free frequencies for RFID. In the US, the radio frequency regulation laws permit power levels for use of RFID at UHF that give it longer range than at other frequencies. Now most of the rest of the world is following suit in at least easing their radiation laws at these frequencies, though not necessarily as much as is the case in the US today. The result is that UHF gives or will give one of the longest ranges for RFID without recourse to a battery in the tag and all the problems of life, size, cost and reliability that batteries entail.
Because of military and cellphone rights for different UHF frequencies in different countries there is no one UHF frequency approved for RFID in all countries. The EC approves 868 MHz, the US approves 915 MHz, Japan 950 MHz and so on. That is not a major problem, because some companies have RFID interrogators that cope with all these frequencies with no significant increase in cost and no degradation of performance. True, where a territory such as Europe approves a very narrow range of frequency it limits the scope for high integrity reading but there are still good options available.
Proven in action
A few years ago, UHF RFID was moved to centre stage by the decision of leading retailers and military organisations to standardise on it for pallet and case tagging. Certain suppliers promoted it strongly, particularly because the range seemed very attractive for such relatively large items. Millions of dollars are now being spent by the retailers and military on UHF interrogators and infrastructure, tens of millions of UHF tags are being delivered this year to users of pallets and cases by UPM Rafsec, Avery Dennison, Alien Technology and some 40 others. However, original plans were to fit billions this year - indeed it is now two years since Gillette placed an order on Alien Technology for 500 million of these tags, so things are not going to plan.
UHF problems with pallet and case applications
In this article we look at the severe problems being encountered in making UHF tags work satisfactorily on pallets and cases and why they are not yet, if ever, becoming the favoured tag for item level tagging despite many forecasts to the contrary. However, it is important to point out at the outset that IDTechEx is not recommending that the world's pallets and cases be tagged at a standard frequency of HF even though millions are so tagged today. Reverting to HF would bring other problems and may or may not be sensible depending on the level of success that the industry has in conquering the - largely unforeseen - problems now being encountered at UHF.
These problems include:
- Too many tags dead on arrival. There is reason for optimism here because things are improving rapidly.
- Intellectual property. At other frequencies this would not be much of an issue and it was thought, wrongly, that at UHF the industry would happily donate its intellectual property for the common good as has happened with other open systems in the electronics industry. However, Intermec, for example, has very valuable key patents on use of UHF for RFID and it understandably wishes to benefit proportionately. Up front payments and royalties from users will likely be demanded and this would increase the cost of tagging pallets and cases, delaying the advent of tags and systems low enough in cost to be viable on the majority of pallets and cases.
- Inadequate success in reading. This is the most severe problem of all. In systems delivered to date, read success of zero to 64% is frequently encountered when all agree that anything less than 99% or so is unacceptable. This is in stark contrast to read success with UHF tags for non stop road tolling for example. This is because the tagged products vary in shape and large amounts of water and metal are commonly encountered close to the tag and interrogator. The system is also more complex, with road tolling using simple 24 bit read only but pallet and case tagging moving to 128 bits or more, read write and complexities of signalling etc. Metro has achieved 100% success by treating the tag much like a barcode and reading it at modest range with nothing in the way. Kimberley Clark has sometimes achieved 100% reads by having dry non absorbent products and avoiding bringing the pallet load in from the cold and reading it immediately because the condensation on the load can give zero reads. However, these are scarcely solutions for everyone. Indeed, the objective of pallet and case tagging is to read tags on every case in the load, including those completely out of sight, and also from within the pallet (skid) itself - something that can never be achieved with barcodes.
- Specification creep. The original concept of the MIT Auto-ID Center was for ruthless adhesion to the minimum tag specification because it was realised that anything more would seriously delay or even make impossible the five cent tag needed for viable tagging of 35 billion pallets/ cases yearly let alone the one cent (or less) tag needed for viable tagging of most of the trillions of items barcoded every year. The Auto-ID Center felt that 64 bits of data read only would get the show on the road and 96 bits read only at limited range and with modest sophistication of data handling might be the end point. Now EPCglobal has taken over and an increasing minority of players are becoming concerned that every committee meeting adds "nice if" features that bloat costs. Respond to the small number of privacy advocates with tags being killable (one bit read write). Respond to the relatively poor availability and integrity of today's networks with read write tags. Respond to security concerns with complexity of signalling. Respond to requests for all in the value chain to add their own data by making the tag 128 bits and preferably several hundreds of bits. Taken in isolation, each of these acts of "specification creep" are plausible, even desirable, but they have malign consequences. Read write tags are inimical to the necessary data synchronisation and high tag costs and often more complex interrogator requirements resulting can put back the day when costs are low enough for the necessary paybacks to be achieved. Facility to kill tags in supermarkets will lengthen queues - the opposite of what RFID is said to achieve. Fortunately, in the Metro trials the public are shunning the "killing" machine at checkout. They just want to go home. Nestle has said "killability is not an option". There is a need to get back to the original vision of MIT Auto-ID Center as networks become better and simple read only, modest data tags can be used in trillions, preferably directly printed on products and packaging. Sounds familiar? Yes. That is because barcodes are like that today and Tesco are right to call RFID labels in the retail environment "radio barcodes" because that is what they are. Leave the clever aspects of RFID to other sectors such as military and healthcare. It's tough enough to meet the price points with something very basic. MIT could see that. The partners in the Japanese Hibiki project to make a four cent UHF tag have approached EPCglobal to remove one complexity of the latest specification and they have some sympathy in the West. However, Hibiki is based on over 500 bits so it has its own escalation.
Let us try to summarise a very complex and difficult situation with an allegory. It is like Tam O'Shanter in the famous poem of Robert Burns. He rode on the moor at night and encountered many horrors he did not anticipate. He had no assurance that turning back was the right thing either.
Please excuse our wit. He will probably get to the other side of the moor, but no one foresaw that we were riding into this, even as an interim stage.
For now, in the RFID equivalent, we are even considering "nuclear" options such as Gillette's plan to redesign a lot of its packaging to make it "UHF" friendly including removing metallization and foil and spacing the packs in the multipacks further apart. That will be costly. Is this the equivalent of changing a light bulb by rotating the house?
For the full article, read the May 2005 issue of Smart Labels Analyst. To learn more about these issues and how companies are tackling them, attend the following conference:
- Smart Labels USA www.smartlabelsusa.com