Segmentation, Clustering and Timing Relationship Analysis of MANET Traffic Flow

Abstract

Users in mobile Ad Hoc networks (MANET) usually encrypt their data packets to resist the evasdroppers, which makes the network management and Intrution detection difficult. However, user behavior, ultimately displayed as traffic flow, shows regularity along time. This paper aims to study the regularity through studding the timing relationship between traffic flows, whose results provide the technical support for user behavior analysis. First, segment the end-to-end flows based on the information of time intervals and packet lengths. Second, cluster the segments by an improved maximum-distance method. Third, analyze the time relationship between the clusters, i.e., traffic flow types, based on the clustering results. Simulation results verify the effectiveness of the method. DOI: http://dx.doi.org/10.11591/telkomnika.v11i8.3145