The Impact of Spectrum Sensing Time on the Performance of the GADIA Algorithm

In this paper, we analyze the performance of the GADIA (Greedy Asynchronous Distributed Interference Avoidance) algorithm ([1], [2]) given the spectral sensing time of the various active agents. Under the assumption that the sensing time is small compared to the update intervals, it is shown that the algorithm converges to an equilibrium point. We also derive the steady state variance of the error due to spectrum sensing collisions by different clusters. Moreover, we propose a heuristic adaptive rate scheme that reduces the number of frequency band switchings required to reach the equilibrium. Simulation studies show that the algorithm achieves more than 96% of the aggregate Shannon capacity corresponding to optimal/near- optimal strategies. Additional simulations are provided, verify our analytic results on the performance of the algorithm.