Optimization of non-convex multiband cooperative sensing

This paper proposes a pattern search (PS) based algorithm to optimize non-convex multi-band cooperative sensing problem. We present a two-step PS-based solution: The first step uses uniformly distributed random sets of input points to find a solution. The set of points that gives the maximum throughput is chosen as input to the second step, which consists of solving the problem using PS algorithm. We give numerical examples that show the improvement of our method over existing algorithms.