Prediction in mobile ad hoc network based on fuzzy time series

Several parameters like routing protocol, mobility pattern, average speed of mobile nodes, path length from source to destination, previous delay, etc., affect the end-to-end packet delay in mobile ad hoc network. But the nature of relationship between end-to-end delay and those parameters is still unclear. The end-to-end delay can be represented as a fuzzy time series. In this paper, a new method to forecast the end-to-end delay is presented. The method fully capitalises on the two key technologies, automatic clustering and automatically generated weights, to handle the forecasting problems. First, the automatic clustering algorithm is utilised to generate clustering-based intervals. Then, the variation magnitudes of adjacent historical data are used to generate fuzzy variation groups. Third, the final forecasted variation can be obtained by the weights of the fuzzy variation. Finally, the phase of forecasting is performed. Based on performance evaluation criterion, we found that the predicted value of the proposed method gives satisfactory packed delay prediction in ad hoc network.