Shared Spectrum Allocation via Pathloss Estimation in Crowdsensed Shared Spectrum Systems

The RF spectrum is a natural resource in great demand. The research community has addressed this unabated increase in demand via development of shared spectrum paradigms, wherein the spectrum is made available to unlicensed users (secondaries) as long as they do not interfere with the transmission of licensed incumbents (primaries). In this paradigm, typically a centralized entity (spectrum manager) is responsible for allocation of spectrum bands to a requesting SU, based on known parameters of the primaries and channel state. Optimal power allocation is important for efficient management of spectrum.In this work, we consider a crowdsourced architecture of the shared spectrum paradigm, wherein the spectrum is monitored by a large number of inexpensive spectrum sensors deployed in the area of interest. In this context, we propose a spectrum allocation algorithm that is based on estimation of path loss from the sensing reports of these spectrum sensors. Such an architecture obviates the need to assume a propagation model, but requires design of accurate estimation techniques. We consider various possible estimation schemes involving splitting of aggregate power received at each sensor and multiple interpolation schemes. We evaluate the performance of our proposed schemes with respect to the optimal allocation, and observe that allocation using the best of our methods is close to the optimal.