A linear programming solution to subcarrier, bit and power allocation for multicell OFDMA systems

We address the adaptive allocation of subcarrier, bit and power in the downlink of multicell OFDMA systems which a re operating at a frequency reuse factor of one. When a centralized scheduler is used, the presence of co-channel cell interference inter-relates the assignment of the subcarriers; and powers in all the cells, making the optimal allocation of radio resources a challenging problem. The lack of optimal solution in the literature for such a centralized approach is attributed to the difficulties in solving the mixed integer nonlinear programming problem. We show that through some manipulations, the solution to the original problem can be decoupled into solving two subproblems sequentially: first solve for power allocation which is a linear programming and then subcarrier-bit allocation which is a binary linear programming. The elimination of the nonlinear constraints in our approach results in the significant reduction in the computation complexity, making optimal solution becomes possible. Although optimal solution not always can be computed in real time when the number of subcarriers is large, it can still serve as a performance benchmark to future developed heuristic algorithms. Finally, results in our paper are more practical since discrete values are used for bit-loading. This is different to most of the reported work in the literature which adopted information theoretic approach.