A Multi-layered Outlier Detection Model for Resource Constraint Hierarchical MANET

For sharing resources using ad hoc communication MANET are quite effective and scalable medium. MANET is a distributed, decentralized, dynamic network with no fixed infrastructure, which are self-organized and self-managed. Achieving high security level is a major challenge in case of MANET. Layered architecture is one of the ways for handling security challenges, which enables collection and analysis of data from different security dimensions. This work proposes a novel multi-layered outlier detection algorithm using hierarchical similarity metric with hierarchical categorized data. Network performance with and without the presence of outlier is evaluated for different quality-of-service parameters like percentage of APDR and AT for small (100 to 200 nodes), medium (200 to 1000 nodes) and large (1000 to 3000 nodes) scale networks. For a network with and without outliers minimum improvements observed are 9.1% and 0.61% for APDR and AT respectively while the maximum improvements of 22.1% and 104.1%.