Waiting time prediction: an enhancement for call admission control of mobile system using neural networks

Admission control plays an important role for quality-of-service (QoS) in modern mobile telecommunication systems, and the waiting time prediction is expected in case of failure of making a connection due to the bandwidth reason. In this paper, we provided a waiting time prediction method for handling failed admission request in the mobile telecommunication system by neural networks (NN-WTP). Very simple input parameters, such as the day of a week, time, number of active connections that occupy the voice channel, and number of active connections that occupy the data channel are required for the backpropagation neural networks to simulate a hyperplane. After a short training period, the waiting time prediction can be performed on this hyperplane. Finally, we did simulation with a whole day network request usage history to demonstrate the feasibility and high performance of this method