Optimization of LTE scheduling using multi-objective evolutionary algorithms

The performance of cellular mobile telephone (LTE) networks is impaired by the amount of interference experienced within the network. In particular, LTE networks are susceptible to inter-cell interference which is experienced when two neighboring cells transmit on the same portion of the frequency spectrum. We propose an adaptive scheduler that can be adopted by LTE networks to minimize inter-cell interference; the scheduler can itself be optimized for the prevailing network environment by using evolutionary algorithms. We describe this scheduler along with a number of experiments that explore its optimization in a number of network environments. In particular we find that either single-objective or multiobjective evolutionary algorithms find parameterisations for the adaptive scheduler that lead to mean throughput for users and sectors that is comparable to established schedulers, but with significantly improved performance for the users at the 5th percentile of throughput.