Multi-player Bandits for Distributed Cognitive Radar

With new applications for radar networks such as automotive control or indoor localization, the need for spectrum sharing and general interoperability is expected to rise. This paper describes the application of multi-player bandit algorithms for waveform selection to a distributed cognitive radar network that must coexist with a communications system. Specifically, we make the assumption that radar nodes in the network have no dedicated communication channel. As we will discuss later, nodes can communicate indirectly by taking actions which intentionally interfere with other nodes and observing the resulting collisions. The radar nodes attempt to optimize their own spectrum utilization while avoiding collisions, not only with each other, but with the communications system. The communications system is assumed to statically occupy some subset of the bands available to the radar network. First, we examine models that assume each node experiences equivalent channel conditions, and later examine a model that relaxes this assumption.