TARgeting DIstributed Sensors for Situational Surveillance (TARDIS3)

The TARgeting Distributed Sensors for Situational Surveillance (TARDIS3) program will study the feasibility of sensor-leakage techniques to enable explosive event detection without relying on friendly human or technological assets. Specifically, we propose placing targeted advertisements through public marketplaces that request and surreptitiously exfiltrate the sensor data needed to detect and localize explosive events. Advertisements allow third parties limited access to consumers’ devices for a nominal fee, including accelerometer data and other commonly used sensors. Furthermore, the required data can be collected using only JavaScript and HTML5 code without exploiting security weaknesses – it is available by design. Furthermore, the Android security model allows sensor monitoring to continue for some time after the source app is closed or backgrounded, considerably extending the potential duration of sensor availability. We believe this approach to data collection, coupled with rigorous statistical approaches for event detection and localization, will lead to a robust and long-lived distributed sensing capability. MTRI Inc., in conjunction with its academic colleagues at Michigan Technological University, is uniquely positioned to carry out this kind of research. We possess deep experience in sensor analysis, the software development expertise needed to develop the capability for mobile platforms, and the signal processing background necessary to exploit the collected data. Our team has already developed a proof-of-concept demonstration of the proposed approach that collected accelerometer data from an up-to-date Android phone without user interaction. The proposed seven-month $166K Phase I effort will produce 1. a report on the capability of mobile sensors to detect explosive events at long range, 2. a proof-of-concept demonstration of data collection and exfiltration from an ad-hoc network of mobile devices, and 3. algorithms for processing that data into the detection and localization of representative events. To mature this technology further, MTRI Inc. and its partners will also study the persistence of WebView execution and seek ways to obtain a sufficient degree of sensor access without the continued purchasing of ads. An essential element of the TARDIS3 approach is the absence of persistent code (no app installation), which facilitates distribution and covert monitoring in geographically distant areas. Depending on our findings on operating constraints (e.g., process focus and device power state), this approach may also apply to other situations where passive monitoring is beneficial: the search for survivors in inaccessible locations by disaster response teams, hostage situations, or urban warfare/terror scenarios. Finally, this is a dual-use technology, and the ability to provide advertisers with characterizations of consumer behaviors that are complementary to existing marketing analytics provides a clear path to commercialization.