SVM Hotspot Identification for Cellular Networks

Cellular operators need to monitor network performance fluctuation to effectively manage and maintain large-scale cellular networks to bear enormous and ever increasing data traffic. Identifying dense traffic areas in real time is crucial to serve them responsively. In this paper, following a pre-processing step based on principal component analysis (PCA), support vector machine (SVM) algorithm with different kernel functions are employed to identify traffic hotspot using real key performance indicators (KPIs) data. The optimal parameter setting is found by the grid search technique. Numerical results indicate that the proposed approach is capable of identifying hotspot efficiently and accurately.