Abstract: Facing the more and more complicated mobile communication network, intelligence is a direction for the future network self- optimization techniques. Applying machine learning algorithms to self- optimization, the network can intelligently coordinate different kinds of optimization objectives according to the variation of environment and status, and achieve the optimum parameters settings. In this paper, based on analysis of thecommonused machine learning algorithms, by considering the characters of the data in the future network, a preliminary intelligent optimization framework and steps are proposed. Furthermore, the commonly used machine learning algorithms for different optimization functions are classified and summarized.

Key words: SON, 5G, machine learning, self-optimization, intelligent optimization