ABSTRACT

A sensor consists of a sensing board (that can sense magnetism, sound, heat, etc.), a battery-operated small computer, and an antenna. Sensors in a network can communicate in wireless fashion by broadcasting messages over radio frequency; and due to the limited range of their radio transmission, the network is usually multihop. One of the challenging problems in designing sensor networks is to secure the network against adversarial attacks. This problem has received considerable attention in recent years.4,8,11,12

Two common examples of adversarial attacks are called message insertion attacks and message replay attacks. In a message insertion attack, the adversary inserts arbitrary messages into the message stream from a process p executing on one sensor to a process q executing on a second sensor. Such an attack can be thwarted by attaching a digest10 to each legitimate message in the message stream from p to q. The digest attached to a message is computed using the different fields of the message and a secret key that is shared only between processes p and q. The adversary does not know the shared key between p and q and thus cannot compute the digest of any arbitrary message that the adversary wishes to insert into the message stream from p to q.