Detecting wormhole attacks in 3D wireless ad hoc networks via 3D forbidden substructures

Abstract

A wormhole in wireless ad hoc networks is a physically low-latency link that connects two topologically distant nodes. Thereby, an adversary can launch the wormhole attack by tunnelling recorded packets from one node to the other and retransmitting them in the network. Since the wormhole attack is independent of MAC layer protocols and immune to cryptographic techniques, it has been one of the most dangerous security threats to wireless ad hoc networks. At present, efficient algorithms have been applied to 2D networks to detect wormhole attacks by seeking for forbidden substructures. However, when we employ their defined forbidden substructures to detect wormhole attacks in 3D networks (which are more pervasive in reality), we encounter severe obstacles. Through an in-depth examination, we discover the existence of efficient 3D forbidden substructures by introducing maximum independent sets (MaxIS) into the network. Essentially different from 2D forbidden substructures, 3D forbidden substructures can hardly be intuitively perceived. Driven by above understandings, we design a MaxIS-based wormhole detection algorithm for 3D networks using only connectivity information. Furthermore, we conduct thorough theoretical analyses and illustrate that our algorithm is able to detect almost 100% wormhole attacks even in a wireless network with very poor connectivity.