Adaptive DSR to mitigate packet dropping attacks in WMNs using cross layer metrics

Abstract

Wireless Mesh network (WMN) is a form of Wireless Ad hoc network, where mesh topology is used to connect the radio nodes together and it is prone to vulnerabilities due to its open architecture. Packet loss is the major issue due to its open wireless physical media, frequent topological changes, scalability and power constraints. The most common blackhole and grayhole attacks besieged on network layer degrade the performance of WMN by dropping packets during transmission. In general, packet drops in WMN during transmission may happen due to mobility or buffer overflow or power depletion or malicious behavior of the intermediate nodes. This paper proposes an Adaptive Dynamic Source Routing protocol (ADSR) to detect the actual cause for packet drops, identify and isolate the misbehaving nodes using the cross layer metrics. It employs proactive, reactive mode of detection of packet dropping nodes and the packets are rerouted through alternate paths using packet salvaging technique. Bait-RREQ with virtual IP address is used to detect the blackhole nodes proactively and node behavior analysis is used to detect the packet dropping nodes reactively. The performance of ADSR is analyzed using a mathematical model by considering the characteristics of a node’s behavior. Using extensive simulations the performances are thoroughly analyzed, evaluated and compared with traditional DSR, WatchDog (WD), Bait-DSR (BDSR) protocols. Simulation results show that ADSR is better than other mentioned protocols with negligible increase in network overhead (less than 2%) which is due to the use of Bait-RREQ and packet salvaging technique. The proper identification of actual cause for packet drops significantly reduces the false positive rate of detection of malicious nodes with an increased packet delivery ratio (not lower than 95%) and throughput.