Blockchain might help major apparel brands from Nike to Macy’s better share product data across the retail supply chain, according to a white paper Auburn University’s RFID Lab published Wednesday.
The study, named the “Chain Integration Project” (CHIP), saw those retailers and others run Hyperledger Fabric nodes on a slice of their mammoth supply chains. The study found blockchain to be a promising way to share serialized data after following tens of thousands of products including Nike Kids’ Air Force 1 shoes and Michael Kors parkas as they moved between distribution centers.
RFID Lab is one of the most prominent outposts for U.S. retailers’ experiments with emerging supply chain tech, but Blockchain Fellow Allan Gulley said it’s a relative newcomer to distributed ledger technology. And so CHIP, which began in 2018, became a blockchain trial by fire for the Auburn research institute.
Many retailers have already been keeping internal tabs on product movements via radio frequency identification (RFID) tags installed in every unit. As one example, Gulley said every box of Nike shoes comes with an RFID tag that helps the athletics giant track its sprawling inventory. However, different retailers tags store data differently and there’s little to no data interoperability.
“Everyone was speaking a different language.” Gulley said. “The way they sent data to us was wildly different from one company to another. There wasn’t a good common language in place and there wasn’t a common platform for them to share that data.”
That gave the RFID Lab two opportunities: Build retailers a common language, and build them a platform.
The language aspect took up about 70 percent of the researchers’ time, Gulley said. With the help of the many students who operate the lab, Gulley’s team built a “translator tool” that reworks different data streams into the EPCIS standard developed by Belgian non-profit GS1.
Implementing Hyperledger Fabric was simpler but still came with its own set of dilemmas, Gulley said. Rookie-level bugs throttled the system’s initial transaction throughput – the team was measuring in seconds per transaction instead of transactions per second – but optimization increased throughput by over 6,500 percent.