Information theory is concerned with the reliable communication of data over an unreliable channel. Networked communication systems such as sensor networks have several sources of unreliability and uncertainty. For example: imprececision in node placement, fading, and interference adversely impact the performance. In theoretical studies, interference is typically treated as noise, and a more robust worst-case analysis of communication in the presence of interference may be useful.

Arbitrarily varying channels are an information-theoretic model for communication subject to adversarial interference. This is a worst-case model in which the interference is assumed to be controlled by a malicious adversary called a jammer. In some cases a positive capacity can only be realized if randomized codes are used. These are codes for which the encoder and decoder must share a secret key that the jammer does not know.

Our work has been to bound the key size needed to permit reliable communication for the Gaussian instance of this problem. A small key turns out to be sufficient, which has led us to look at applications of this result to networks such as sensor networks and large ad-hoc wireless networks.