A Hybrid Machine Learning Based Model for Congestion Prediction in Mobile Networks

Congestion avoidance in radio access networks enhances considerably the end-user Quality of Service (QoS). Congestion should be predicted in advance to allow Self Organizing Networks (SON) algorithms to perform appropriate parameter adjustments (such as handover parameters for mobility load balancing). For this purpose, a novel hybrid model efficient congestion prediction mechanism is proposed in this paper. This hybrid learning model combines unsupervised and supervised learning algorithms. The unsupervised learning consists of a co-clustering algorithm based on Latent Block Model (LBM) that groups similar cells according to their KPIs behaviour over time. Following the co-clustering model, a logistic regression approach is applied on each cluster to predict congestion and alert operators to avoid congestion occurrence in mobile networks. The applicability of the hybrid model is validated for a real data represented by Key Performance Indicators (KPIs) collected periodically for 12 days in a live Long-Term Evolution (LTE) network. The hybrid proposed model has proven its efficiency in congestion prediction in terms of accuracy, precision, recall and F-measure.