Trusted computing for fault-prone wireless networks

Abstract

We consider a fault-prone wireless network in which communication may be subject to wireless interference. There are many possible causes for such interference: other applications may be sharing the same bandwidth; malfunctioning devices may be creating spurious noise; or malicious devices may be actively jamming communication. In all such cases, communication may be rendered impossible. In other areas of networking, the paradigm of "trusted computing" has proved an effective tool for reducing the power of unexpected attacks. In this paper, we ask the question: can some form of trusted computing enable devices to communicate reliably? In answering this question, we propose a simple "wireless trusted platform module" that limits the manner in which a process can access the airwaves by enabling and disabling the radio according to a pre-determined schedule. Unlike prior attempts to limit disruption via scheduling, the proposed "wireless trusted platform module" is general-purpose: it is independent of the application being executed and the topology of the network. In the context of such a "wireless trusted platform module," we develop a communication protocol that will allow any subset of devices in a region to communicate, despite the presence of other disruptive (possibly malicious) devices: up to k processes can exchange information in the presence of t malicious attackers in O( max (t3, k2)log2 n) time. We also show a lower bound: when t < k, any such protocol requires Ω(min)k 2, n) logk n) rounds; in general, at least Ω(min(t3, n2)) rounds are needed, when k ≥ 2. © 2010 Springer-Verlag.