Towards Model-Based Policy Design Methods for More Effective Spectrum Allocation

Modern embedded systems are increasingly likely to be distributed across multiple devices and platforms that must interact with high precision across wireless networks. Traditional ways of managing the wireless radio spectrum suffer from two fundamental limitations, which our research addresses: (1) spectrum is divided a priori into static coarse-grained partitions, and (2) partitions are non-overlapping. To overcome these limitations, we propose an approach to spectrum allocation based on dynamic allocation of diverse portions of the overall spectrum and overlapping allocations to increase utilization. Our previous work has established the potential of a new generation of allocation strategies based on both memory-like and spectrum-specific features. Here, we describe new directions for that work based on Markov decision theory.