Nov 24, 2014In October, I attended the inaugural meeting of RAIN RFID, an organization dedicated to promoting adoption of EPC ultrahigh-frequency RFID. Bill McBeath, chief research officer at ChainLink Research, reported that 60 percent of new RFID deployments are cloud-based. This is not surprising. Cloud computing is a natural fit for RFID applications, given their widely distributed nature and the massive amounts of data they collect.
But these applications are vulnerable to the well-known security risks of cloud computing—and more. That's because they are accessed by human users and RFID readers. A reader tracking goods on a retail shelf, for example, might autonomously connect to the cloud via the Internet to deliver its data.
Despite the ongoing reports of database hacking, there are effective, albeit not guaranteed, ways to secure data in the cloud (just Google "cloud security"). Companies developing cloud-based RFID applications must implement those countermeasures, and then take the following steps.
Secure the communication protocol between a reader and the cloud. Use a secure protocol like HTTPS and configure each reader with an access credential, similar to a password. That way, the cloud application can accept data only from an authorized reader. In addition, set each reader to authenticate the cloud—for example, by examining a digital certificate presented by the cloud application—so it doesn't send data to a "man-in-the-middle" attacker.
Protect the access credential. Unlike a password, a reader's credential must be stored in nonvolatile memory on the device for use each time it connects to the cloud. To prevent an attacker from breaking into the reader's network connection and stealing the credential, encrypt the credential. Examine every way a network connection could be made to or from the device, and make sure it is secured. In addition, ensure that only authorized personnel have physical access to the reader.
Give each reader a different access credential. This limits the damage if a reader is compromised. It also makes it easier to identify and isolate the compromised reader. Grant reader credentials limited abilities within the cloud application. If the reader's job is to send tag reads to the application, the reader's credential should allow only that operation, not other operations such as reading the data or generating reports, which may be needed by other system components connecting to the cloud. That way, if a reader credential is compromised, the attacker may be able to flood the system with bad data, but at least he or she will not be able to read any of the good data or do other harm.
Prepare a backup plan. If all these measures fail, you will need a workable procedure to change each reader's access credentials. Think about how that procedure will work before—not after—you deploy hundreds or thousands of readers!
Ken Traub is the founder of Ken Traub Consulting, a Mass.-based firm providing services to companies that rely on advanced software technology to run their businesses. Send your software questions to email@example.com.