Abstract

Abstract. We present a novel local optimum based power allocation approach for spectrum sharing in unlicensed frequency bands. The proposed technique is based on the idea of dividing the network in a number of smaller sub-networks or clusters. Sum capacity of each cluster is maximized subject to constraint on total power of each user in a cluster. On its turn each user in a cluster maximizes the sum capacity by calculating power allocations that correspond to a local optimum. Total power constraint of each user and effect of interference from other users in the network is taken into account for finding local optimum solution. Comparison of achieved network sum capacity is done with the well known iterative water filling method. Numerical results show that the proposed cluster based local optimum method achieves higher capacity than selfish iterative water filling and is therefore suitable for geographically distributed networks.