DETECTION OF SELFISH NODES WITH Q LEARNING AND IMPROVING ROUTING USING ACO BASED WSN

Selfish nodes are a major source of problems for wireless sensor networks (WSNs) as they erode network trust and performance. Selfish nodes are ones that act in a selfish manner for a variety of reasons, which prevents them from sharing any of their neighbours' packets with the BS and causes packet loss. One of the main causes of the self-centered behaviour is a lack of energy to send packets at that specific moment, which will actually lower network performance and trust. Our goal in this implementation is to apply the Q Learning based selfish nodes detection, which will watch a node's behaviour over time before labelling it as selfish. In addition to eliminating selfish nodes during routing to the BS, this method will also eliminate selfish nodes. The problem is that every node in the network will act selfishly occasionally depending on how much energy they have left over. As a result, we cannot label a node as selfish the moment we witness it acting selfishly; instead, we must watch it for longer periods of time and determine the acceptable trust threshold beyond which it will be labelled as such. And finally, routing will be done using the Ant Colony Optimization (ACO) which will eliminates the detected selfish nodes using by Q learning and forms a new routing using only trusted nodes.

Keywords: Wireless Sensor Network, Trust in the Network, Selfish Node Detection, Q Learning, Ant Colony Optimization (ACO).