Abstract: A continuous predictive model based on time sequence theory and the connatural characteristic of CDMA was presented. According to the unsteady time sequence access feature of CDMA cell, a predictive mode depending on mutually complementary historical data and sampled data was established by this model.In order to improve the convergence of the model the sampled data was smoothed according to the extremum made by the partial differential equation.Through the full consideration of the periodic,random and bursting properties in operation of CDMA network, the model had got good performance in self-adaption and could eliminated the influence of the burst data to a certain extent.The model has the ability of predicting the network load accurately for a certain period of time,so the QoS parameters can be adjusted in advance,and hence avoid the dithering of the sysem and enhance the successful access rate.Experiment results show that this model can increase effectively the QoS performance of the CDMA network with smaller calculating cost.

Key words: CDMA network, predictive model, QoS, radio access