Factor graph-based distributed frequency allocation in wireless sensor networks

As wireless sensor networks (WSNs) become denser, simultaneous transmissions (on the same time slot and frequency channel of two or more terminals) may cause severe interference. Appropriate interference-aware allocation is a complex problem and distributed frequency allocation is even harder. This work studies the problem of assigning frequency channels for a given WSN routing tree, such that: a) time scheduling and frequency allocation are performed in a distributed way, i.e. information exchange is only performed among neighboring terminals, and b) detection of potential interfering terminals is simplified. The algorithm imprints space, time and frequency constraints, assuming half-duplex, single-antenna radios into a loopy factor graph (FG) and performs iterative message passing. Convergence to a valid solution is addressed based on appropriate modifications of the resulting message passing inference algorithm. The proposed algorithm is compared with two distributed frequency allocation algorithms, based on game-theory or min-max interference control. It is shown that the proposed distributed algorithm offers comparable performance with state-of-the-art, even though it utilizes simplified interfering terminals set detection.