On maximizing the sum network MISO broadcast capacity

Abstract

Recently, it has been shown that dirty paper coding (DPC) achieves the sum rate capacity of the Gaussian multi-user multiple-input single-output (MU-MISO) broadcast channel of a single isolated cell. However, when considering a multi- cell scenario, i.e., a cellular network, the optimal strategy to maximize the sum rate capacity in each of the cells is still unknown. Nevertheless, based on a game-theoretic framework DPC can be applied at each cell as a decentralized strategy in a cellular network, in order to maximize the sum broadcast capacity of the network. By treating the cells in the network as players in a strategic cooperative game, simultaneous iterative waterfilling can be performed, i.e., every cell computes its optimal beamforming vectors according to DPC and by considering the intercell interference generated in the previous iteration. At each iteration the beamforming vectors for each user in each cell are updated with the gradient projection algorithm in order to maximize the sum network broadcast capacity. The algorithm is repeated until it converges, i.e., a local maximum is achieved. This theoretic result approaches the maximum rate that can be transmitted in the downlink of a network. Additionally, in order to introduce some fairness into the network, we consider in a similar way as the previous problem, the task of minimizing the sum of the mean square errors of all the users in the network.