ENERGY SAVINGS OF WIRELESS COMMUNICATION NETWORKS BASED ON MOBILE USER ENVIRONMENTAL PREDICTION

摘要

The massive deployment of wireless base stations has led to dramatic increase of wireless networks energy consumption. The prediction of mobile user environment can be used to manage network resources dynamically and save the network energy consumption, but prediction accuracy is still a challenge. This paper proposes an environmental prediction method based on mobile user classification. First, we determine the user classification criteria and the optimal machine learning classification method according to the user's daily behaviour patterns. In addition, we analyse the characteristics of each user cluster and propose a position prediction model based on weighted Markov model. Simulation results show that the prediction accuracy of the proposed method is about 8% higher than the traditional second-order Markov model. In addition, the proposed prediction scheme is applied to manage the transmit power of a selected wireless base station of China Mobile to test its performance. Test results show that about 12% of the electrical energy can be saved while ensuring the quality of service.