January 23, 2006
Finding Ways Around RFID's Errors
By Evan Schuman, Ziff Davis Internet
For all its powerful potential, retail IT users know all too well the error problems with RFID, especially when the more powerful UHF chips get close to liquids or metal.
But as RFID enters mainstream status in 2006, retail users are going beyond traditional engineering means to address those shortcomings and falling back on the time-honored shared-risk, shared-reward approach with suppliers.
Consider the case of Rachel Bolt, assistant director of information systems for the $700 million Piggly Wiggly grocery chain, with 114 stores in South Carolina and Georgia.
She's a big fan of RFID's potential, but she absolutely considers the read rates she's seeing today as "not acceptable," she said.
"With a 97 percent read rate at the case level, I'm still going to have to count it. What's the point?"
Bolt said that she wants 100 percent read accuracy, but when asked what she would realistically accept, she paused and said, "99.99999 percent."
This means that we're not actually doing RFID today. We're simply enabling ourselves to play in that space. There are foundational pieces that a company must have in place. If and when the time is right for us to start receiving the products in our DC [distribution center], we'll have what we need to implement RFID, but I'm not ready to embrace it yet."
To deal with less-than-fully-reliable pallet readings, Bolt is asking that her suppliers provide more accurate advance ship notices. If that happens, she is considering leveraging RFID to allow for the Holy Grail of supply-chain: good faith receivings.
That would involve allowing a consumer goods manufacturer—such as Nabisco—to bring deliveries to the chain and "drop it, leave it and nobody has to check it in."
Such a potential process with the chain's more than 900 suppliers "puts the fear of God" into their operations people, Bolt said.
That's where shared-risk, shared-reward comes into play, in the form of a negotiated error rate. Piggy Wiggly has been in discussions with Nabisco for just such a negotiated rate.
Bolt's plan is to conduct periodic audits. If the audit is covering two months of shipments and if, let's say, products come up short by one percent, the supplier would be charged that amount retroactively for the prior two months.
"Wouldn't that put the impetus of accuracy into their court rather than ours," Bolt asked. "True good faith receiving, it's incredible what that could do to streamline our process."
But it's all based on a foundation of accurate RFID readings, which Bolt can't get.
Part of the debate involves where retailers should use UHF (Ultra-High-Frequency) as opposed to HF (High-Frequency) versions of RFID.
The newer UHF versions have two key differences: read distance and read speed.
UHF can read from a much greater distance, roughly five meters on a typical passive UHF tag, compared with about a meter for the same tag's HF cousin, said Raghu Das, a managing director for IDTechEx, a technology consulting house in the UK.
UHF also has the edge when it comes to faster data transfer rates, allowing many more tags to be read simultaneously.
But UHF suffers from two distinct disadvantages, and that is the core of the error issue. Much more so than HF, UHF readers get flummoxed by liquids and metals.
"Water has a tremendous effect on UHF" and that includes "moisture in the human body," Das said. "That's one of the reasons people are struggling to get it working."
Although the liquid issue can be dealt with, liquids can crop up in sly ways. The liquid inside the bloom of warehouse workers have caused problems and there have been reports of tracked cows confusing the readers, too.
Dog food maker Purina is reported to have experienced UHF RFID problems when tracking bags of dry dog food. Why? The dry food apparently absorbed enough moisture from the air to confuse the readers.
The initial compromise has most retailers experimenting with UHF for backroom warehouse supply-chain tracking only, with HF being used for the initial item-level shelf experiments.
HF's shorter read distance is preferable for item-level anyway because it can reduce readers "seeing" an unintended product.
IDTechEx sees the UHF error rates—coupled with the likely increase in item—level experimentation in the next few years—causing a seesaw effect with UHF versus HF purchases.
"Today, the cumulative sales of RFID, the majority aren't high-frequency," Das said. By the end of, "2006, UHF will become the dominant frequency by tag volume sales. High-frequency will still have dominance for dollar value. By 2007, UHF will dominate in both. By 2010, HF will be the dominant frequency again."
Das sees another potential downside for UHF use in item-level tracking. "Wal-Mart initially said that it wanted all drugs to be tagged with UHF," which might not work with a liquid drug, Das said. But another possibility is that the UHF tags could generate sufficient hurt to impact the product.
"An RFID tag is like any electronic circuit. If it's on, it will heat up," Das said. "At quite a high frequency, there is a chance that it could knock off some of the weak parts of a long molecule drug."
Others in the RFID space doubted that the small RFID tags could generate sufficient heat to impact products.
What there is almost universal agreement on among retail and consumer goods IT managers is that RFID is problematic, expensive, will require extensive investment and is absolutely worth it.
"This is not an IT problem. This is an engineering problem," said a veteran IT manager at cigarette giant Phillip Morris, who asked that his name not be used.
"It's not the time to shy away from it because everyone has a lot to learn. We're in for five years of incredibly painful learning. If you look at the Wal-Mart experience, it is certainly very concentrated on the learning side. It's a disruptive technology and it will change the nature of B2B."
By an engineering problem, the Phillip Morris manager is referring to tweaks with positioning the chips and the pallets they are typically on, as well as reader positioning.
"If you are transmitting and receiving these waves from one antennae, you're doomed," he said. "That's how we have to organize, to make the network intelligent with diversity antennae systems, collaboratively scanning with multiple receivers."
Asked how difficult it is to make such a process work reliably, the Phillip Morris manager said, "Damn tough, but the EPC reader spec allows for this, allows for collaborative readers."
Yankee Group analyst John Fontanella agreed. "This absolutely is an engineering problem, but I believe most CIOs have a very accurate idea of the limitations of RFID," he said. "You need to have that pilot. You need to have that experience."
Gerd Wolfram is the chief technology officer of the Metro Group, Germany's largest retail chain with about $68 billion in annual revenue, 2,400 stores in 30 countries and about 250,000 employees.
Wolfram is quite familiar with the challenges surrounding RFID, but sees things getting better. "UHF on the pallet is not a problem. With Gen2, it's getting better and better. Compared with Gen1, you read quicker, you read more and you get a better read rate. In the end, you can read more in a shorter time."
But he wants more action from global standards groups. "What's still missing is the standardization" from UPCGlobal, he said.
He doesn't see the UHF vs. HF debate settled, but he is leaning toward the conventional approach. "There still is discussion whether we should use UHF for item-level tagging. The pro is that if you use UHF, then you only have one reader generation in the store," Wolfram said.
"On the other hand, there are limitations [liquid and metal conflicts] that are not solved today. We're going to start with HF. I personally think it will be HF for the item level. It's the right technology."
From the perspective of Stan Drobac, the VP of RFID strategy for RFID tag manufacturer Avery Dennison, Wolfram is ahead of the pack by simply having focused on it at such a specific level.
"The whole question of UHF versus HF is a level of detail that many CIOs haven't gotten into at all," he said.