A NEW FRAMEWORK FOR PREDICTING THE IMPACT OF TRAFFIC ON THE PERFORMANCE OF MOBILE AD-HOC NETWORK (MANET): USING REGRESSION AS DATA MINING APPROACH
Kamal Moh’d Alhendawi
Al-Quds Open University, Gaza, Palestine
With the rapid technological advances in wireless communication and the increasing of usage of portable computing devices, it is expected that mobile ad hoc networks are increasingly developed towards enhancing the flexibility, scalability and efficiency of communication technology. The wireless ad-hoc network is a collection of mobile nodes in which these nodes have the ability to connect each other without backbone infrastructure (i.e. infrastructure less). Although many studies have been done on the performance assessment of MANET routing protocols, there is a need for investigating the impact of traffic load on the performance of MANET in order to justify the use of some routing protocol in MANET. This study is one of the fewest that aims at proposing a new framework towards predicting and validating the results of future scenario using data mining techniques. The regression analysis is used as data mining method in the prediction of the future scenarios. Practically, two experiments with eight scenarios are conducted. The findings indicate that the network size and traffic loads are proportionally related to the throughput. However, the findings show that the network size is inversely related to the delay in case of medium FTP traffic, and proportionally related in case of high FTP traffic. The results also indicate that data mining approach specially regression is an effective approach towards the prediction of the future network behavior.
Keywords: MANET; Mobility; Ad hoc network; Prediction model; Regression analysis; Traffic; Delay; Performance metrics.