Discord model for detecting unexpected demands in mobile networks

摘要

Predicting the number of calls in a cell in average must be an important issue in mobile computing. It is quite useful in performance modeling and usually assumed as a random Poisson value. For effective management of cellular networks, the average number of calls should be carefully traced and the unexpected changes in the numbers need to be promptly detected. In this paper, we propose an efficient algorithm detecting those changes in the behavior of users based on the technique proposed for point-of-change problem just in terms of the number of call arrivals. Computer simulation reveals the proposed method effectively detect the discord, and the developed model is very accurate that the difference is as small as mostly less than 1%.