110 likes | 258 Views
Temporal Privacy in Wireless Sensor Networks. Pandurang Kamat, Wenyuan Xu, Wade Trappe, Yanyong Zhang Wireless Information Network Lab Rutgers University. Outline. Introduction Application Scenario Assumptions Formulation Simulation Challenges. Introduction.
E N D
Temporal Privacy in Wireless Sensor Networks Pandurang Kamat, Wenyuan Xu, Wade Trappe, Yanyong Zhang Wireless Information Network Lab Rutgers University
Outline • Introduction • Application Scenario • Assumptions • Formulation • Simulation • Challenges
Introduction • Sptaio-temporal information in wireless sensor networks are important. • Must ensure that • location of source nodes (Extensively discussed) • time when source nodes observed the target (received little attention) • Temporal privacy is defined as the mutual information between the received time sequences and the creation time sequences.
Introduction • In order to minimize the mutual information, buffer each packet at intermediate nodes along the routing path between a source and the sink. • According to information-theoretic study, using random delays that follow an exponential distribution will better protect temporal privacy than other distributions.
Application Scenario • Animal habitat • Animals move through the environment with sensor network ready • One sink and monitor the locations of animal
Assumptions • The assumptions for the adversay are • Able to eavesdrop (read packet header) • Unable to decrypt the packet • Able to estimate packet creation time based on network knowledge • Unable to interfere with the proper function of the wireless sensor network • Unable to inject or modify packets, alter the routing path, or destroy sensor devices. • Always located in the node of the sink and only observes the packets into the sink.
Formulation h τ time z x x’ h: hop count τ: transmission delay x: true creation time x’ : estimated creation time (x’ –x)2 : mean square error Higher estimation error better preserves the temporal privacy of the source.
Formulation • Use the concept of differential entropy in Information Theory to generate a formula and show that choosing a delay distribution will make adversary learn as little as possible about x from z.
Adaptive Buffering • Packets will buffer in an intermediate node with a random time along the routing path from a source to the sink. • Adaptive buffering strategy that preempts buffered packets to accommodate newly arriving packets if the buffer is full. They call it as RCAD (Rate-Controlled Adaptive Delaying)
Challenges • High cost of a sensor with large enough buffer size • High traffic v.s. very limited buffer size