Performance Optimization in Heterogeneous Wireless Access Networks Based on User Heat Maps

In this paper an approach for dynamic and proactive performance optimization in dense and dynamic heterogeneous wireless access networks taking into consideration the user distribution, mobility and activity is proposed. The approach is based on building up User Heat Maps (UHM) in consecutive time slots for a given area and predicting the map state in the next time slot. To avoid storage of big volumes of data and computational complexity and to ensure real-time operation the prediction is based on a Neural Network (NN) architecture utilizing the data from UHM. The approach is demonstrated with a scenario for optimizing the overall cell throughput based on controlling the electrical tilt of the antenna at the serving access node. The simulation results show that such an approach could lead to performance improvement in dense and dynamic heterogeneous access networks characterized by frequent changes in user activity and mobility.