When Wal-Mart's CIO announced, in June 2003, that she was going to require its top 100 suppliers to track shipments to them by using RFID technology—instead of bar codes—by January 1, 2005, it sent a shock wave through the retailing and logistics industries. Just as a bar-code scanner reads a label's universal product code (UPC), radio-frequency identification (RFID) readers get information from an electronic tag that transmits signals containing data about the product. But because the new system requires only proximity, not the line-of-sight accuracy of a bar-code scanner, palettes of products can cycle through warehouses and onto trucks faster.
The resulting shock waves also hit Philips Semiconductor, whose Identification Systems division develops RFID chips, and its longtime partner Escort Memory Systems, which develops bar code scanners and RFID readers. For them, the Wal-Mart announcement meant that the RFID market was going to break wide open (the retailer's top 100 suppliers represent about 1 billion cases of merchandise each year). When the Department of Defense followed Wal-Mart's lead, requiring all its material to have RFID tags, it quickly became apparent that the opportunities were not limited just to retail. ABI Research expects the RFID market to top $2 billion by 2006 and $3 billion by 2008.
The time-to-market challenge, then, was to not only collaborate on a system—Philips bringing the chip expertise, and Escort bringing the reader expertise—but also to create one that would not be just a me-too product when it hit the market. Because the retailers driving RFID were insisting on speed and accuracy, the combined development team decided to set the bar high. It set out to create tags and readers that would deliver 99 percent accuracy (the industry usually accepts 85 percent), and because so many consumer packaged goods contain liquid, to do so even when the radio signals had to pass through it.
Working Together
Experts say that RFID may be as revolutionary as the UPC technology was when it was introduced: Replacing the UPC code is an electronic product code (EPC), which can register not only a manufacturer's name and a product code but also a unique serial number.
As a result, with RFID labels or tags, retailers such as Wal-Mart and Target (which has also expressed interest in RFID) will be able to track merchandise at a more granular level—that is, by box or item, rather than just by pallet.
"It potentially gives people real-time visibility into the supply chain," says Dennis Gaughan, an analyst covering enabling technologies for AMR Research. "That kind of visibility helps them plan better and reduce the amount of inventory they have to carry."
Even so, Luciano Mattoli, CEO of Escort, has been in the RFID industry long enough to understand its shortcomings. He recalls the days back in the 1980s when everyone designed their own tags and their own readers. "Everything was proprietary, and there were no de facto standards, even for which frequencies to use," he remembers. It wasn't until Philips and Texas Instruments first helped promulgate the use of an international standard frequency in 2001 and started designing RFID chips around which companies such as Escort could build systems—a.k.a. "readers," which understand the signal the RFID tag is sending—that the market showed potential.
In Mattoli's eyes, TI stumbled, because it decided to design both the chips and the RFID readers. "By doing that, TI prevented companies such as us from working with it." But the need for collaboration was immediately apparent, because neither Philips nor Escort wanted to become expert in the other's capabilities.
"Our play in RFID is silicon," says Christophe Duverne, vice president of Philips' Identification Systems division, "but people are asking for turnkey solutions. Although we had a strong knowledge of the whole system, we were not going to be able to drive the industry forward on our own. That's where the need for partnerships comes up."
The challenge lies in the complexity of implementing RFID systems. It's still a new enough technology that you can't just drop a chip into a reader. "When you design a chip, you have to make sure the reader can pick up the signal" no matter what the surrounding environment is, says Duverne. "That's why we have to take a real systems approach to the design. Escort is very strong at understanding how the environment is going to interfere."
Although Duverne's division also partners with IBM, Sony, and Nokia on other projects relating to Bluetooth and smart cards, he wanted to work with Escort, even though, at 70 employees, it's much smaller than most of its partners. A privately held division of Bologna, Italy-based Datalogic, a manufacturer of scanning and data-collection systems, Escort represents about 10 percent of the company's 2003 revenues of €131.5 million (approximately $156 million), according to executives there. Duverne believes that its size serves Philips well: "It's very inventive, agile, and practical."
Suresh Palliparambil, marketing manager at Escort, says, "The folks at Philips saw that we were a small, hungry company, so it challenged us with the most difficult technical problems. But whatever we had to do to engineer the solution, the end result still revolves around the chip." As a result, says Palliparambil, there is regular interaction between the two companies—"it taps into us, and we tap into it. Philips wants to hear the questions we get from customers and the reactions of those customers." Even though Escort is Philips' direct customer for chips, he adds, both companies are trying to serve the end customer.
Between the two companies, engineers for this project were spread across half the globe, from San Jose and Scotts Valley, Calif., and Foxborough, Mass., in the U.S. to Paris and Graz, Austria. When face-to-face collaboration was necessary, the engineers met in the most logical location. "In the early discussions, when we would talk about what the chip would look like, we'd do that here, because our engineers are here," says Dirk Morgenroth, marketing manager for Philips' identification division in Graz. "When the engineers needed to work on the hardware, they'd go to California."
With this close collaboration, Escort always knew what Philips was doing. Queries from potential Philips customers were routed via e-mail to Escort, which could then demonstrate which readers the chips would appear in. Escort got sample chips and could modify prototypes in order to demonstrate the system for customers. By doing this, says Morgenroth, "both of us were gaining experience in learning what the complete solution needed to have. We were exchanging our expertise in a noncompetitive way, which helped both of us."
The two companies were so far ahead of the competition in applying RFID to supply-chain and logistics issues that they were able to do these demonstrations months ahead of other companies. "We were getting a real 'wow!' effect, because people could see that what we were doing wasn't just on paper or PowerPoint slides," says Morgenroth.
"The result is a shorter time to market," adds Palliparambil. "Before anybody reads about the chip in the press, we've already tested it and written software for it." According to Morgenroth, the two companies saved six to nine months in getting the resulting system, the LRP2000-26 Pass Thru System, to market in the third quarter of 2003; earlier this year, Escort delivered a shorter version—4 feet instead of 6 feet high—of the system. He estimates that the system represents about 10 percent of the RFID market, "and that's a healthy percentage for an immature market."
Conquering the Technical Challenges
The technical challenges involved in such a system are not trivial. Traditional bar code solutions used in industrial situations work about 85 percent of the time, because the conditions can be harsh or dirty in food processing or manufacturing facilities. Labels that are unreadable have to be reprinted, which slows the process and affects the integrity of the data being collected.
But RFID has data-collection issues as well. When a forklift rolls past the system, every single tag on each box is transmitting information about the contents, which can be confusing for a reader. "All the signals can interfere with each other, so you have to write digital-signal processing (DSP) software to deal with these collisions," explains Palliparambil. "But there's only so much you can do in software, so some of it is done in the chip."
With the Escort system, even if the forklift is as far as 7 feet away (depending on the size of the label) and moving at 600 feet per minute, its antenna can read the tags on the boxes. They transmit at 13.56 MHz, the only frequency that's agreed upon internationally. The other frequency used in supply chain applications in the U.S. runs in the UHF range, 860 to 950 MHz (in some heavy industrial applications, a third frequency, 125 kHz, is used).
The advantage of the lower frequency, notes Escort CEO Mattoli, is that the signal can pass through any liquid. Being able to read all the tags on a palette of boxes, even though there may be liquid in the containers, is a big advantage in the consumer packaged goods arena. The drawback is that the range of the antenna on the reader is shorter (the RFID signal has to be stronger to get through liquid and hence gets weaker over longer distances).
Dealing with all the variables of length and frequency required some juggling. So prior to the development of either the chip or the reader, adds Morgenroth, the engineers on both teams sat down together to develop an optimized solution, taking into consideration the characteristics of the readers and the chips. "That offered us the best performance in the marketplace for that frequency, optimized for both our chips and their readers," he says. That optimization is important, because there's no exclusivity built into the Philips/Escort deal. The success of UPC codes is based on the existence of an agreed standard, and the partners knew that the EPC used in RFID scanning had to follow the same track. "We've always intended that our solution will be the basis for an open platform, so that we can get acceptance in the market not only for our own products but also for products we do with other partners," says Morgenroth. "We had to work within limits to get a standardized product."
