Imagine a world where every single manufactured item is tagged with a tiny RFID chip that holds various bits of information about that item. And as each item moves along a supply chain—pharmaceutical, aerospace, consumer goods, automotive, livestock or even human identification—information is gathered, stored and accessed at various points in a mixture of systems.
The goal is to be able to know where everything—or anything—is at any given time. The challenge, however, is determining how such a scenario would play out: what the actual network would look like once companies up and down the supply chain collaboratively start inputting and exchanging information among trading partners and their partners partners.
In a joint project, the Massachusetts Institute of Technology and SAP are looking to see if proposed network architectures can fully sustain radio-frequency identification tags on items being used in a collaborative fashion by players up and down a supply chain to help track products from manufacturing to the consumer.
EPCglobal, the RFID standard-setting body, is going about the task of creating an Electronic Product Code to number the worlds goods. It has created standards to enable RFID data exchange and has proposed a network architecture, called the EPC Network, for enabling all the objects in the world to be linked via the Internet. At the same time, other groups—academic institutions and software companies, for instance—have floated potential, and competing, network architectures that could accommodate the global exchange of RFID data.
But before a network architecture can be agreed upon, some basic questions need to be answered: How much network traffic would be created in such a scenario where every physical item transmits data? What type of architecture is necessary to support a scalable network? What type of security would be necessary to validate queries? And what kinds of resources will be needed to handle all this traffic?
These are just some of the questions being tackled by MITs Auto-ID Labs—the successor to the Auto-ID Center, which was split in 2003 into Auto-ID Labs and EPCglobal—and SAP Labs in a joint project called Utilizing Distributed EPC Data with Enterprise Applications. The project, now in its early stages, looks to simulate a supply chain involving manufacturers, distributors, wholesalers and retailers moving millions of products a day. The aim: to determine whether several proposed architectures can handle predicted network traffic and, perhaps, to influence the outcome of evolving standards themselves.
On MITs side, heading the EPC data project is John Williams, a specialist in simulation and large-scale data modeling who has been asked by the Department of Homeland Security to simulate different networks, such as the Internet, telephone networks and water-supply systems, where communication has broken down at one time or another.
Williams has some clear ideas about how RFID data can be used to facilitate communication among companies and about the challenges that lie ahead.
"The idea would be to know where everything is at any time. One of the things—the main challenge—is that in a year, maybe a trillion products are manufactured, so to try and keep track of trillions of things, we havent built systems of that scale yet," said Williams, Auto-ID Labs director and principal investigator for the project, in Cambridge, Mass. "The largest [network] is the Internet, then the telephone network, but this [RFID-enabled network] is more demanding than that. One thing I am concerned with is, How do you build a system so you can go to a central place and make inquiries about [an item]? There are several issues once we make this infrastructure: We need to be able to make things secure and scalable."