Experimental analysis of machine learning based node failure prediction using game theory strategy

Wireless Sensor Network (WSN) provides ability to communicate between one-end to other-end without any range restrictions. The successful communication largely dependent on node sustainability, bandwidth sufficiency and battery lifetime. The major problem raised over the Wireless Network medium is the node failures, which can happen due to intruder attacks, battery failures and so on. This research proposes a machine learning based node surveillance scheme to identify the node status and provide a feasible communication medium without any flaw. A node surveillance scheme called Learning Enabled Node Surveillance Model (LENSM) is introduced, which is derived based on the Game Theoretic strategy. This model inspects each node and maintains the traces into a trace filefor further monitoring. Traces are generated as a model for future validations. The node behavior is traced with respect to the trained model and the node can easily be monitored during communication. At every point of communication, the source node checks the route and fix it in a sequential manner for communications. The pathway is sequentially specified and is cross-validated based on the trained model as well as the efficiency of the respective neighbor is cross-validated by means of transmitting empty packets to it. The respective neighbors’ efficiency and energy level can easily be measured by analyzing the responses for the empty packet. This strategy and sequential path selection processare implemented based on the Game Theory Modelling. The efficiency of the proposed approach is assured by means of enhanced throughput, improved node surveillance, energy efficiency enhancements and quick response rate recorded in the performance observed in the simulation model.