Predictability of Cellular Network Traffic Based on Conditional Entropy

Bo Sun, Lei Yao, Zida Liu, Junlan Feng, Xing Wang, Lin Zhu, Chao Deng, Yong Zhang

Abstract: The prediction of cellular services can help operators adjust the network status in advance to meet the QoS requirements of users. Evaluating the upper boundary of prediction is the key to designing prediction algorithms. The traffic of cells in cellular networks has strong temporal periodicity and is related to the traffic features of adjacent cells. Therefore, three types of entropy are proposed to measure its predictability: (1) Temporal entropy, (2) Spatial entropy, and (3) Spatiotemporal entropy. The Fano equation is used to calculate the predictability boundary, which provides a prediction boundary for the base station traffic prediction of cells in cellular networks, and the effectiveness of the proposed entropy has been verified through simulation experiments.