TARP: Timing Analysis Resilient Protocol for Wireless Sensor Networks

In timing analysis attackers study the transmission pattern of different nodes in a network with the goal of extracting information about users, applications, or the structure of the network, even when the traffic is encrypted. Defeating timing analysis attacks requires expensive traffic mixing measures that equalize the transmission pattern at all nodes; such measures are especially expensive for battery operated wireless devices. In this paper, we first introduce TARP, a traffic mixing approach for defeating timing analysis tailored towards sensor networks. While TARP improves on traffic mixing approaches by combining multiple packets destined to different destinations in a single frame (amortizing packet overhead), traffic mixing remains expensive. To this end, we propose two techniques for improving the energy efficiency of TARP: (1) Using multi-path routing to exploit the available capacity engineered to defeat timing analysis; and (2) Adaptive transmission control to allow the transmission pattern to be adapted to the offered load without exposing the structure of the network. Furthermore, we define and explore the notion of relaxed timing analysis resilience where resilience is provided with a limited scope that is well defined in space and/or time. By controlling the scope to fit the application requirements, substantial savings in energy (or delay) can be achieved, while retaining desired levels of timing analysis resilience. Together, the proposed techniques significantly reduce the overhead of TARP, making timing analysis resilience more affordable for critical applications.