Enforcing data integrity in very large ad hoc networks

Ad hoc networks rely on nodes forwarding each other's packets, making trust and cooperation key issues for ensuring network performance. As long as all nodes in the network belong to the same organization and share the same goal (in military scenarios, for example), it can generally be expected that all nodes can be trusted. However, as wireless technology becomes more commonplace, we can foresee the appearance of very large, heterogeneous networks where the intentions of neighboring nodes are unknown. Without any security measures in place, any node is capable of compromising the integrity of the data it forwards. Our goal in this paper is to ensure the integrity of the data being disseminated without resorting to complex and expensive solutions. We achive this by discouraging malicious behavior in two ways: a) enforcing integrity checks close to the source and b) refusing to communicate with obviously malicious nodes. We find that by having nodes sample their traffic for corrupted messages, malicious nodes can be identified with high accuracy, in effect transforming our collection of nodes into a self-policing network.