Pareto-metaheuristic multi-objective network optimization for OFDMA-based systems

In this paper, a new planning method is proposed for the next generation wireless networks that are based on Orthogonal Frequency Division Multiple Access (OFDMA). As a consequence of the wide variety service demands in terms of data rate and Quality of Service (QoS), the traffic pattern is considered to be heterogeneous. Therefore, the complexity of obtaining Base Stations (BSs) positions increases with the randomness of the traffic distribution. In addition to this challenge, the capacity of each BS is limited due to power and bandwidth constraints, propagation losses, Gaussian antenna pattern, and the Co-Channel Interference (CCI), which in turn increase the complexity measures for an efficient network design. According to Nash Equilibrium, combined efficient systems must perform equally to achieve certain performance. This implies that the traffic of a cellular system should be equally distributed over all the BSs to achieve the highest network performance. Hence, we formulate the planning problem as a non-linear multi-objective optimization problem. The optimum solution should not dominate the throughput of one BS over the others and this is referred to as Pareto optimal. However, loading all cells equally may not be possible in certain traffic distributions. Therefore, the proposed method tends to approach the optimal solution by tackling the problems of BS positioning and resource allocation simultaneously. We adopt a hybrid approach, i.e. Pareto-Metaheuristic (PMH) that achieves a balanced throughput over all cells as well as minimizing the number of the installed BSs targeting a certain service outage probability. Simulation results show that, in addition to maximizing the individual cell throughput, the network throughput variation decreases as the number of iteration increases.