Interview with Industry leaders

Mike Marsh:The requirements for RFID systems in the retail environment will be the largest deployment of electronic systems in the history todate of mankind.
The volumes needed are far beyond what humans can imagine, with estimated volumes as large as 100 million million (10 power 14) transponders needed PER ANNUM. At present the world is producing an estimated 230 million per annum for other applications beyond retail, a capacity that has been built up over the past 20 years. This new capacity requirement would require nearly a one million times expansion in the current manufacturing capacity.

The transponder requirements have three major sections, namely the chip; the antenna; and the assembly of the chip and the antenna onto the goods being labelled. The chip comes from silicon foundries who have been running their plants for the past fifty years, understand mass manufacture, and are the area that is most likely to understand the volume increase problem. They will need to increase their capacity by purchasing more manufacturing equipment, building bigger premises and hiring more staff - a cycle that takes up to five years.
At times this sector has been in massive oversupply, with capacity at only 36% a short while after the collapse after Y2K currently rising to 94% currently after downscaling. They will therefore not be in a hurry to risk their entire business by expanding too rapidly, but they are the major indicator of whether the manufacturing industry accepts there is a sustainable business in RFID production.
Although companies like Intel are associated with RFID initiatives, the production at their plants could not be used for transponder production as this requires plants with different processes not used in microprocessor development.

The second challenge is the production in volumes of antennas. This will either be by extending the capacity of the printed circuit industry to etch/stamp/punch metal foils in high enough volume, or on the route of printed conductive inks where drying time is a major limitation on throughput.

To pick up a chip, attach it to the antenna and finish off the assembly requires mechanical movement of components in the assembly machinary. The inertia of those moving parts creates a practical limit of assembly rates of 7 per second (or 100 million per annum per machine). Higher throughputs are available by paralleling assembly modules and most probably the fastest available can do 400 million per annum. The throughput of these machines is also limited by the curing time of the glues and sealants used is the assembly. Only a couple of companies in the world have the experience of building these machines, although the market need, if RFID did take off, would be for about 1 million of the machines running in parallel.

Although there is a lot of enthusiasm among some small players in this phase of the RFID experience, their ability to produce in the required volumes is virtually non-existant and it will need the participation of an entire new industry to manufacture at these rates.

Transponder News:Your newsletter states that many of the saught after benefits of RFID are a mechanism of the underlying IT systems, achievable with traditional retail ID technology (barcodes) as well as RFID systems? 

Mike Marsh:The scanning of barcodes, or the reading of RFID transponders, generates data that is used in a software package to provide management or control information. The barcode/RFID transponder is the input information to the system indicating that an item is currently in front of the reading station. The management benefits from using barcode/RFID Technology are realised via the software that takes these inputs and makes it into a management system, whether it is a tracing system for JIT manufacture/retail, an ordering system for restocking or automatic invoicing system. Many of the promised benefits from the future RFID systems are realisable immediately by just using the software abilities immediately. For example in a retail store where the goods are checked out via scanning systems, the accounting software already could know the stock levels in the store if it is told what has been delivered to the store, as it has a record of what has been removed. There might be a shortfall via shrinkage but this is a minor adjustment.

Even track and trace can be realised by improving communication between the different participants in the chain, be it the manufacturer, the transport company and the retailer. It is not necessary to uniquely identify every box of Rice Krispies on the pallet, just actually to track the pallet of goods that is referenced by the invoice that will be issued for the goods. Once again this is a function of software and communication , and not of whether the goods are labelled by means of barcodes or RFID transponders.

RFID transponders do have many benefits over that of barcoding in the immediate vicinity of the reader, such as higher scanning rates, accuracy and the ability to read the identity of the goods while they are not in line of sight of the scanner. There are other benefits further down the line such as incorporating EAS features for automated checkout but the industry is not at that point in the process yet, bearing in mind the shortage of transponders.

At present the proposed RFID system gives every item in the world a unique identity number. As the solution matures, I believe the RFID system will change to be directly compatible with barcode systems and the numeric data currently used to describe manufactured items, so that the same numbering system can be used from the major retailers to the "mom & pop" store.

The barcode and numeric system require that all units of the same product have the same identifier in order to simplify the printing and packaging process. RFID needs to become compatible to this system so that one can just plug in an RFID scanner or a barcode scanner and run the same software. We also do not want to get to the point where the identify is such a long number that it has to be written down on an A4 sheet of paper, it must be appropriate for checking and use by manual systems as well as the electronic systems such as barcodes and RFID.

Transponder News:The most important criterea relating to the practical application of RFID in large volumes seems to relate to the radio interference issues which are embodied in the Tag-talks-first and Reader-talks-first versions of protocols. Please explain and the relevance to the future success of EPC Generation2?

Mike Marsh:When you are the only user of a cell phone in a city, you don't care whether your signal would interfere with other cell phone users as there are none to complain. However when you design a communication system which will only be effective if many people make use of the system, interference issues are a key design criterea. Retail stores would need many scanners to operate in close proximity. The RF protocol between the reader and the transponders allows many transponders to be identified on a single operating frequency. This involves the reader providing an energy field to power up the transponders in its reading zone which is also used as the control frequency for the communications that will happen between the reader and the transponders to manage the channel and allow many transponders to communicate in an ordered method through that channel. There are two classes of protocol to manage this ordered communication, namely reader-talks- first (RTF) or tag-talks-first (TTF).

In RTF protocols, the reader calls out into space a regular message, asking if a particular tag number is present. For example "Tag NO 1 are you there?". A receiver on the transponder decodes this message and if it is TagNo1, then it replies "Yes" and the reader moves onto Tag No2. In order not to call out every number in the sequence, a binary treeing algorithm is used so that the entire sequence can be covered in a short time. The issues around this is that the reader calls out whether or not transponders are present, many calls are transmitted for each tag detected, and that the reader message is carried on the higher power transmitter signal of the reader which is received by other readers within a one kilometer radius on that frequency. Due to the high amount of calling needed for each successful reception, the calling rate has to be high which uses up wider bandwidths, usually about 200Khz per operating frequency.

In TTF protocol, the transponder sends its identity to the reader as soon as it gets sufficient operating power from the energising field. This message is transmitted at power levels that are up to one million times smaller than that of the reader signals in RTF protocols, meaning that their zone of interference is very small. The transponder in the simplest form of TTF, can send out this message at random intervals as long as it is in the energising zone, or in more sophisticated TTF protocols, it can be instructed by the reader to cease transmission by the reader with a simple code as soon as its message is correctly received. If the message is corrupted due to many transponders talking at the same time, they will repeat their message until they are heard with no corruption. The advantage of TTF is that the message information is carried in the protocol by the transponder using a very low power transmitter which has to be less than the operating power available to the transponder, which can mean its transmitter power is as low as just 1microwatt. This message cannot be heard by the other transponders in the field due to its very low power, and will not be heard by other readers a few meters away. The bandwidth needed for communication can also be as small as 10kHz (20 times less than that of RTF) as the transponders on entering the field initiate the communication with the reader waiting in listening mode. In reality, TTF readers can be operated as close as 4 meters from each other.There are more sophisticated TTF protocols such as Trolleyponder which allow all transponders to have identical data, such as would be needed in a barcode replacement RFID technology, but this is an issue only once the RFID technology has matured.

As can be seen, both from the narrow bandwidth and from the signal strength used to convey the message, TTF reader systems can have operating densities in a limited RF bandwidth situation that are up to 1000 times higher than that achievable with RTF type protocols.

While there are only a few readers in existance the importance of RTF vs TTF for the EPC situation will not be apparent to the users. As the number of readers increase this will become the most important issue and will eventually force a protocol switch from EPC from their current RTF strategy.

Transponder News:Is the 2nd generation EPC standard doomed to failure due to inherent RF limitations?
The current RFID initiative is distorted by the push from the retailers to have a useful system in large volumes in a short time. The initiative is not coming from the companies that are developing technology, but in response to market edicts. This means that the normal time that technology would need to be tested and bedded down is not being allowed.

In reality RFID at present for retail is most probably very similar in development to where the computer industry for small systems was in the 1970s. At that time one had the Apple 2, Osborne, Sinclair, and the IBM PC was making its debut. Since that time there have been many developments which gives us the current situation of a computer in every home.

RFID is going to get its technical embedding time it needs as serious players are not going to get involved on a sufficient scale with something that might not work. Their investment is at risk, they are not going to commit to a technology that is not yet technically sound.

From our own work, we know that a reliable operational read/write technology for RFID is a huge problem for which there is no viable solution at present. The EPC have developed a specification without having working systems on large scale as benchmarks. The indications are that Gen 2 will not survive and I can for see a time where Generation 6 will be reached before there is wide spread acceptance by the serious manufacturing industry who are crucial to take this from the introduction phase to the reality phase.

Transponder News:Are organisations such as EPC and ISO going to be able to define practical standards that balance the need of the users and the abilities of the current technology of the producers to make a practical system?
Mike Marsh:Organisations like EPC have a role to play in facilitating the development of RFID in retail. Their power to control the destiny of the technology is most probably much less than they realise. I do not think that there is another organisation that would necessarily do a better job, except that it has to be realised that this is a global project, that it is a technical project rather than a market pull project, and that they have to persuade serious players who can address the manufacturing shortfall to become involved for the project to succeed. It is unfortunate from the RFID aspect, that the largest retail groups in the world were the ones that were providing the market pull for the fledgling RFID technology. Had the technology had a chance to succeed with an application that was within the capacities of current suppliers, and grown from there, it would have provided a challenge that was achievable and could of expanded to meet the needs of retailers of the size of Walmart.