Not so long ago, we adopted a family pet from a local animal shelter. I was surprised to learn that all animals adopted from the center have a radio frequency identification (RFID) chip injected under their skin, just behind the shoulders. The chip contains an ID number that can be used to identify the animal if it is lost. The RFID scanning devices have become inexpensive enough that most vets and animal control officers now have one. By entering the serial number stored on the chip, they can retrieve the animal's history and address from a Petfinder Web site. For me, this realization was an "aha" moment. You know a technology is starting to mature when animal shelters and the local dog catcher - who are not usually early adopters of technology - get into the game.

The emergence of auto-identification (auto-ID) technology is a major part of a larger trend that combines the virtual world of the Internet with the world of physical objects. Although RFID is the most prominent, there is a broader range of auto-ID technologies that includes real-time location systems (RTLSs) that report the location of objects and various sensing technologies such as ZigBee that allow networks of inexpensive sensors to communicate anything that can be measured by machines. In contrast to the high profile emergence of e-commerce, auto-ID applications have been called "silent commerce" or the "Internet of things." Technology watchers such as Gartner and Forrester have different names for this idea, but the trend is unmistakable. The ability to track the location and state of physical objects through embedded devices that can be read by computers will change our world.

It's Everywhere

While it is unlikely that we'll see RFID tags on soda cans or rolls of bathroom tissue for quite some time, the technology is popping up in larger and higher value items in everything from airplane luggage to pharmaceuticals. In the retail supply chain, RFID sensors placed on shipping pallets and cases are helping to track and trace the movement of goods. As the cost of RFID tagging gets lower and the reader technology becomes more reliable, the emphasis will shift from the warehouse to the store shelf. As shown in Figure 1, RFID can provide data at every link in the supply chain. At the factory, auto-identification allows manufacturers to capture specific information about a particular item including date of manufacture, location, shelf life and process used. Tags on palettes and cases enable tracing and tracking of items as they move to distribution centers and, ultimately, to the store shelf. RFID readers in store "Smart Shelves" will enable the store to keep track of shelf stock and avoid stock outages while relaying demand-planning data back to the factory.

Figure 1: RFID in the Supply Chain

The eyes of the RFID world were on Wal-Mart as it mandated that 100 of its top suppliers deliver pallets of product identified using RFID tags. While industry pundits debated the merits of the technology, suppliers complained that the technology lacked a positive return on investment, privacy advocates decried loss of consumer protections and things looked bleak for RFID proponents. While there continued to be setbacks in both the reliability of the technology and cost of RFID implementations, the movement to adopt RFID in the past year has been positive. According to Linda Dillman, the CIO of Wal-Mart, in a Forrester research interview:

  • The ability to reliably read tags at distribution points has climbed from 40 to 60 percent into the high 90s.
  • The process of generating picking lists for out-of-stock items in stores has been improved with RFID.
  • By reading RFID data on stocking carts, stores can tell if they have been cleared before opening the store.

Managing the Data Deluge

During the tech bubble of the '90s, the development of e-commerce was hampered by the fear of millions of Web users overloading a company's enterprise applications. The need to interface enterprise applications, networks and databases to the Internet gave rise to a distinct set of middleware technologies including JNDI, JDBC, JMS, XML and Web services, to name a few. Much of the work done to date on auto-ID technology has focused on the real-time capture of data. However, as in the world of e-commerce, the world of "silent commerce" will give rise to a deluge of data that will be both large in volume and high in throughput.

Figure 2: Electronic Product Codes Identify the Type and Instance of the Object

This is not obvious at first glance. After all, aren't we just replacing bar code data with another 96 bits of data identifying an item? There are three key differences:

  1. Figure 2 shows the makeup of an RFID tag. It identifies the manufacturer, the class of object and the serial number of the object. Barcodes only identify the product type and the manufacturer. For us ex-programmers, this means we can identify the instance of an object, not just its type.
  2. Once you can track an instance, there is a wealth of associated data you can and will capture. While an RFID chip may contain a 24-digit code, the information describing the item in detail is held on a separate server or servers. For a manufactured good, this detailed description might contain a product's name and broad category (laundry detergent, auto part, clothing and so on), detailed description of its use, date of manufacture, its expiration date, its current location or even its current temperature.
  3. Bar code readers require line of sight, meaning that using a bar code is mostly a manual process. A single RFID reader can perform hundreds of reads in a second. RFID data is usually a stream in which a sample is taken several times in a single minute.

Dealing with this deluge of data will require a new approach. Fortunately, work is being done to address this issue. The infrastructure for this real-time data collection system is being designed by a joint effort of EAN International and the Uniform Code Council (UCC) called EPCglobal. This organization has partnered with the industry to define a reference architecture for this infrastructure called the EPC Network. As shown in Figure 3, the EPC Network consists of the following five major components:

  1. EPC codes used in RFID tags and the RFID readers that follow a specific internationally recognized format specified by EPCglobal.
  2. ID system (EPC tags and readers). The specifications include transmission requirements for both the tags and tag readers, plus the interface protocols for communication among these components.
  3. Application-level event middleware (ALE) that provides a buffer between enterprise applications and the real-time data collection devices. A particular network may have a hierarchy of ALE servers, each performing a different data collection task.
  4. EPC information services (EPC IS) that provide higher-level functions by receiving updates from ALE software and matching reader data with data stored on EPC IS servers. Enterprise applications can query an ALE server or an EPC information services directly for the data they need.
  5. Discovery services that allow applications to map an EPC code to a specific server on the network in the same way an IP address get matched to a Web site address using domain-naming services on the Internet.

Figure 3: EPC Network Components

The standards for the EPC Network components are still evolving as the group of early adopters and technology vendors discover and solve many of the toughest problems. Not too surprisingly, the tag and reader standards are furthest along.

For more information, visit

Integration Strategies: Who's Connecting the Dots?

While this might seem fairly straightforward at first, it's clear that the same system that supports Wal-Mart's inventory management systems or the U.S. Post Office logistics is probably not going to resemble the system that helps me find my cat should it get lost. One of the biggest tasks facing IT organizations will be integrating RFID data with the enterprise applications that make use of the information. Software vendors offering EPC Network components include start-ups as well as existing players. ConnecTerra, OATSystems and GlobeRanger are relatively new companies that offer real-time data collection capabilities. Other companies such as WhereNet (RTLS) and GenuOne offer business solutions based on auto-ID technology.

Integration of RFID data is high on the list of many large-scale software vendors. Enterprise resource planning (ERP) leader SAP offers an RFID goods flow system which accepts data from data collection systems through SAP's Smart Item Infrastructure (SII) and channels that data to SAP's Portal and Business Information Warehouse modules. IBM offers comprehensive services and software packages such as the IBM WebSphere RFID Premises Server that is designed to integrate readers with enterprise applications. Progress Software's ObjectStore RFID Accelerator focuses on transforming raw RFID event streams into meaningful business data. Oracle's recently announced Sensor-Based Services claims to provide a transparent method of integrating RFID and sensor data into a business software infrastructure. Data integration specialists, such as Ascential Software (now part of IBM), have already developed RFID early adopters programs to help them to deal with the inevitable data quality issues. While reference architectures exist for components such as the ALE, there is substantial opportunity for differentiation among these vendors because the nature of the integration problems to be solved varies greatly.

The Outlook

RFID and other auto-sensing technologies have the potential to improve efficiency, deliver better asset utilization, reduce shrinkage and counterfeiting and increase sales by reducing out-of-stocks. These technologies can even help improve the safety of the things we eat and the drugs we buy. However, there are still many outstanding problems with the reliability of tag reads as well as the cost of deployment. Today the industry is focused on solving the data collection issues at the edge of the network. As these problems are solved, the data integration and data management issues will become the next big hurdle. Technologies that allow sensors to talk to machines will challenge IT organizations by introducing data into the IT infrastructure in ways that they were never designed to manage. For people who manage the information infrastructure for large organizations, the real challenges are just beginning.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access