Distributed Power Allocation in Cognitive Radio Networks under Network Power Constraint

We consider utility maximization in a multi-cell network under a total transmit power constraint, e.g. given by a cognitive radio geo-location database. The network utility in the downlink is maximized by allocating transmit powers in the network, while meeting the network-wide transmit power constraint. Distributed algorithms for allocating downlink transmit power are discussed, which involve exchange of prices that reflect interference between cells. Using primal decomposition, we present an online algorithm which guarantees that the network power constraint is met at all times. To this end, each cell adjusts its power level while taking into account the interference prices received from neighboring cells. Depending on the pricing information, a transmitter may reduce its power so that it can be used by some other transmitter. Distributed optimization enabled by the exchange of interference prices among cells results in an efficient distribution of total power among the transmitters. Simulation results illustrate that exchange of prices can yield a significant gain over non-cooperative and partially cooperative power allocation approaches in indoor small multi-cell networks.