A Reputation Based Routing Mechanism To Contrast Selfish Node In Delay Tolerance Network

Abstract

Delay Tolerant Networks, that is, DTNs are specialized networks designed for scenarios with significant physical distances and no direct medium between sender and receiver. These networks face high error rates and data loss, with security being a major concern. Nodes in DTNs often collaborate to transfer packets in a multi hop fashion, which is resource-intensive, consuming significant energy, CPU time, memory, and bandwidth, leading to reluctance in participation. newlineTo conserve energy, some nodes only send and receive packets relevant to them, ignoring others. This selfish behavior results in non cooperation in routing by expelling unwanted packets, rejecting route requests, and failing to forward messages, leading to communication breakdowns. newlineThis thesis proposes a novel approach, Deep Autoencoder-based Nonnegative Matrix Factorization, that is, DANMF. DANMF autonomously learns suitable functions for nonlinear mapping based on data attributes and utilizes the deep autoencoders nonlinear structure for high generalization. The model calculates a weighted cumulative social tie combined with residual energy to identify selfish nodes. newlineThe study also proposes an incentivized reputation strategy that groups nodes based on their social attributes and calculates a weighted social tie. Nodes with higher residual energy but lower social ties are identified as selfish and penalized. The new model demonstrates a higher detection rate and lower false positive rate compared to traditional methods. newlineAdditionally, the thesis introduces Fuzzy Selfish Node Detection using Arithmetic Optimization Algorithm and Cellular Automata, which uses fuzzy logic and the Arithmetic Optimization Algorithm to detect selfish behavior and assess node reputation. The model operates in two phases cluster formation with a cluster head, and reputation assessment through fuzzy logic. This approach outperforms conventional methods in terms of residual energy and detection accuracy. newline newline