Research On The Sybil Attack Detection In Mobile Wireless Sensor Networks

Wireless sensor network ?WSN? is a multi-hop self-organized network that canmonitor the environment by deploying wireless sensors in a certain region. Thecollected information from the monitored environment are passed to users through thewireless self-organized multi-hop network. Due to the simple structure, limited batterypower and limited storage, sensors are easily compromised by the attackers. As one ofthe common security threats, Sybil attack access the network information by forgingmultiple identities using one wireless device, the forged identities are called Sybil nodes.Network operations such as distributed storage, routing and topology management canbe greatly damaged by Sybil nodes. The existing Sybil attacks detection methods aredesigned for static WSNs rather than mobile WSNs. In this thesis, my works aboutSybil attacks detection in mobile sensor networks are summarized as follows:?1?According to the actual requirement, the features of Sybil attacks in mobilewireless sensor networks are analyzed. We proposed a Sybil attack detection approachbased on ID frequency in MWSNs. The simulation result of the proposed approachshows that we can detect all Sybil nodes with high accuracy. Compared to traditionaldetection approaches, our approach does not require additional communication andcomputation overhead;?2? The feature of the Sybil nodes' mobility is analyzed to build a centralizeddetection model. We proposed a Sybil attack detection approach based on ID affinity inMWSNs. The simulation result of the proposed approach shows that this approach canalso detect all Sybil nodes with high accuracy. Compared to traditional detectionapproaches, this approach does not require additional communication and computationoverhead;?3? In order to avoid the over reliance on the sink node, we proposed adistributed method combined with broadcast gossip algorithm to detect Sybil attacks inmobile wireless sensor networks. Then the validness of the distributed detection methodis verified by comparing the detection time with the converging time of broadcastgossip algorithm. The simulation results show that this distributed detection method candetect all Sybil nodes within a very short period which is less than the theoreticalconverging time.