Negative Effects of Selfish Nodes in Mobile Ad Hoc Networks (Manets): A Comprehensive Review and Analytical Framework

Mobile ad hoc networks depend upon the multi-hop collaboration of the nodes; that is, the nodes not just act as end systems but also behave as routers to forward the packets of other users. However, in the real world, the nodes might act selfishly to conserve scarce resources such as battery power, bandwidth, CPU time, and buffer space. Unlike other attacks that are intended to destroy the network, selfish behavior might be economically rational, intermittent, and context dependent. Hence, selfish behavior might not always be easy to detect and, more importantly, might cause performance degradation that manifests because of mobility and wireless effects. This paper integrates the impact of selfish nodes on MANETs and shows how local non-cooperative behavior can cause network-wide decline. We present a taxonomy of selfish behavior at different layers of the network such as routing misbehavior, selective forwarding, fake route report generation, resource hoarding, and MAC layer misbehavior. We then show how local selfish behavior can cause network-wide decline about packet delivery ratio, delay, jitter, routing overhead, energy dissipation, fairness, and network partitioning. To facilitate easy understanding of the effects of selfish behavior, we present simple analytical models that relate the level of selfishness and path lengths to the probability of end-to-end packet delivery. We then present a reproducibility template for the evaluation of selfish behavior about the choice of protocols used, the selfish behavior modeling used, the metrics used for evaluation, and statistical reporting of the results. The integration of the effects of selfish behavior highlights that a significant effect of selfish behavior might be the formation of a feedback loop that causes packet loss due to rerouting, which in turn causes more routing overhead that consumes scarce resources such as energy and bandwidth, which might cause other nodes to behave selfishly as well. Hence, the cost of selfish behavior might not just be packeting loss but the possible shortening of the network lifetime and the compromising of mission- critical applications due to the selfish behavior of a small number of nodes.