Optimal Power Allocation for Full-Duplex Communications Over OFDMA Cellular Networks

We consider the optimal power allocation problem for an OFDMA small-cells network where the BSs operate in full-duplex mode. Although the problem formulation is not convex, we show that, leveraging on the intimate relationship between the minimum mean square error and the interference to noise ratio, it is possible to derive an iterative allocation scheme that provably converges to a local maximum of the sum rate. Moreover, owing to the convexity of the mean square error formulation, the proposed approach copes with unlimited number of interferers, and as such it is amenable for implementation on a generic multicell scenario. Despite we do not explicitly specify any sub-channel assignment constraint, the solution at convergence allows to fulfill the exclusivity assignment constraint in almost all cases, i.e., the constraint that each sub-channel can be assigned to at most two user pairs in the cell. The proposed allocation scheme allows to achieve fast convergence to a solution that approaches the performance of a power allocation scheme that includes an exhaustive search over all possible sub-channel assignments.