Real-time Wormhole Detection And Location In Wireless Sensor Networks(WSNS)

In the current decade, the research on wireless sensor networks(WSNs) is an attractive field in the computer networks. Comparing with wired networks, WSNs are resource-constrained and network dynamics are strong such that security technologies based on cryptography in wired networks thereby cannot be applied into WSNs directly. Wormhole attack is one of the most serious security issues in wireless sensor networks in which attackers tunnel the packets between distant locations in the network through a high-speed link. Thus, wormhole nodes can attract a large amount of network traffic and, as a result, collect and manipulate the network traffic to launch a variety of attacks such as eavesdropping, dropping, corrupting and modifying the packets. Existing works on wormhole detection, such as time-bounding, geometric localization and topology recovery by neighbor matrixes collecting are all subject to course time-synchronization, additional hardware, huge communication overheads and failing to detect active adversary who can change its attack strategies adaptively, respectively. To this end, we propose a topology-based, passive and real-time wormhole detection scheme, which based upon the key observations that a large amount of network traffic will be attracted and the paths will be shortened by the wormholes. We adopt light-weight and secure topology detection algorithm to harvest topology information with the price of very little communication and computation overhead. Moreover, our scheme can track real-time topology changes and then deduct whether wormholes exist in the network or not accordingly. Once the existing of wormholes is confirmed, the scheme can locate the wormholes. As the active adversary can hide its location, our scheme is aimed at locating wormholes as themselves, or nodes within one-hop neighbors of real wormholes. In a word, our approach relies solely on network routing information and does not necessitate specialized hardware or poses rigorous assumptions on network features. We evaluate our system performance through extensive simulations of 100 to 500 nodes for various network scales in a 25*25 grids in aspect of false positive, false negative and time delay, and show that our scheme is light-weighted and well suited for false alarms, scalability and time delay.