We address the problem of a group of n nodes, connected to the same broadcast channel (e.g., a wireless network), wanting to establish a common secret in the presence of a passive eavesdropper. If these nodes have no access to an out-of-band channel, the only practical solution today is to use public-key cryptography,which relies on the assumption that an adversary has limited computational capabilities. In this paper we ask the question: can this problem be solved without using any formof cryptography, and thus not relying on the adversary's computational limitation? We propose a secret-agreement protocol,where the n nodes of the group keep exchanging bits until they have all agreed on a bit sequence that the eavesdropper cannot reconstruct. In this task, the n nodes are assisted by a small number of interferers, whose role is to create channel noise in a way that bounds the amount of information the eavesdropper can overhear. Our protocol has polynomial-time complexity and requires no changes to the physical or MAC layer of network devices. We formally prove its optimality properties, show that it scales well to an arbitrary number of nodes, and experimentally demonstrate that it can generate 150 secret kilobits per second in a small testbed of 8 wireless nodes.
View on arXiv