Customizable Fuzzy-Neuro Inference System Attack Detection Based on Trust for Mobile Wireless Sensor Networks

Wireless Sensor Networks (WSN) when simulated for implementations use Mobile Sinks (MS) which are vulnerable to unauthorized attacks. Existing works do not ensure attacker node identifications making prevention of attacks during initial iterations would a difficult task. Neighbour node selections based on their node information alone can increase the possibility of intrusions. The detection of an attacker node in the initial stages reduces information loss and increases authentication. In continuation to this statement a TNF (Trust based Node Filtering) technique which assigns trust values to nodes is introduced in this work. This proposal is an authentication scheme against replication attacks. Neighbour nodes which cannot be trusted are not considered for authentication processes thus ensuring guaranteed data transmissions. Malicious nodes are detected using a FIS (Fuzzy Inference Systems) with the proposed CFNIS (Customizable Fuzzy- Neuro Inference System). Nodes are also classified as being in normal or attacking node. Secure Key generations are performed using RSA and Elgamal algorithm and a honey pot server controls authentications. CFNIS simulation showed higher security against replication attacks.