Voice Quality-Based Handover Strategies for VoLTE Networks

Abstract

In this paper, we propose a new handover scheme for enhancing Voice over LTE (VoLTE) user experience via voice quality prediction. The proposed scheme addresses the demand to sustain smoother progression and fulfilment of the VoLTE calls service utilizing the Perceptual Objective Listening Quality Assessment (POLQA) scores prediction. A novel neural network (NN)-based prediction algorithm is presented for improving the decision-making process of the target handover (HO) scheme. Specifically, the proposed algorithm utilizes a prediction system based on a neural network that reads multiple measurements from each neighbouring cell such as Reference Signal Receive Power (RSRP), Reference Signal Received Quality (RSRQ), Received Signal Strength Indicator (RSSI), Channel quality index (CQI) and Signal-to-Interference Noise Ratio (SINR). It then predicts the POLQA score of the neighbouring cells and the HO decision will be based on the highest POLQA score. The obtained results for the trained model demonstrate a 26 % improvement in terms of the VoLTE quality-of-service (QoS) compared to exiting HO mechanisms.