Malicious node detection in ad-hoc wireless networks

Abstract

Advances in wireless communications and the proliferation of mobile computing devices has led to the rise of a new type of computer network: the ad-hoc wireless network. Ad-hoc networks are characterized by a lack of fixed infrastructure, which give ad-hoc networks a great deal of flexibility, but also increases the risk of security problems. In wired networks, key pieces of network infrastructure are secured to prevent unauthorized physical access and tampering. Network administrators ensure that everything is properly configured and are on-hand to fix problems and deal with intrusions. In contrast, the nodes in an ad-hoc network are responsible for routing and forwarding data in the network, and there are no network administrators to handle potential problems. This makes an ad-hoc network more vulnerable to a misconfigured, faulty, or compromised node. We propose a means for a node in an ad-hoc network to detect and handle these malicious nodes by comparing data available to the routing protocol, such as cached routes in Dynamic Source Routing, ICMP messages, and transport layer information, such as TCP timeouts. This data can then be used along with network probes to isolate the malicious node.