Using a Kalman filter to improve schedulers performance in mobile networks

Abstract

In this work we propose an optimization of schedulers used for resource allocation in Long Term Evolution (LTE) networks by implementing a Kalman Filter (KF) to the average data rate instead of computing it with the simple expression from the schedulers (20% the actual rate, 80% the previous rate). In order to verify if the predictions using the KF improves the schedulers performance, we have integrated it into LTE-Sim [1] and performed single-cell simulations with five different schedulers. By changing the number of users in the cell we show that, with the introduction of the filter to compute the data rate, there is an increase of throughput and a reduction of packet loss.