Doctor of Philosophy (Ph.D.)
A cyber-physical system (CPS) comprises of a network of processing and communication capable sensors and actuators that are pervasively embedded in the physical world. These intelligent computing elements achieve the tight combination and coordination between the logic processing and physical resources. It is envisioned that CPS will have great economic and societal impact, and alter the qualify of life like what Internet has done. This dissertation focuses on the privacy issues in current and future CPS applications. as thousands of the intelligent devices are deeply embedded in human societies, the system operations may potentially disclose the sensitive information if no privacy preserving mechanism is designed. This dissertation identifies data privacy and location privacy as the representatives to investigate the privacy problems in CPS. The data content privacy infringement occurs if the adversary can determine or partially determine the meaning of the transmitted data or the data stored in the storage. The location privacy, on the other hand, is the secrecy that a certain sensed object is associated to a specific location, the disclosure of which may endanger the sensed object. The location privacy may be compromised by the adversary through hop-by-hop traceback along the reverse direction of the message routing path. This dissertation proposes a public key based access control scheme to protect the data content privacy. Recent advances in efficient public key schemes, such as ECC, have already shown the feasibility to use public key schemes on low power devices including sensor motes. In this dissertation, an efficient public key security primitives, WM-ECC, has been implemented for TelosB and MICAz, the two major hardware platform in current sensor networks. WM-ECC achieves the best performance among the academic implementations. Based on WM-ECC, this dissertation has designed various security schemes, including pairwise key establishment, user access control and false data filtering mechanism, to protect the data content privacy. The experiments presented in this dissertation have shown that the proposed schemes are practical for real world applications. to protect the location privacy, this dissertation has considered two adversary models. For the first model in which an adversary has limited radio detection capability, the privacy-aware routing schemes are designed to slow down the adversary's traceback progress. Through theoretical analysis, this dissertation shows how to maximize the adversary's traceback time given a power consumption budget for message routing. Based on the theoretical results, this dissertation also proposes a simple and practical weighted random stride (WRS) routing scheme. The second model assumes a more powerful adversary that is able to monitor all radio communications in the network. This dissertation proposes a random schedule scheme in which each node transmits at a certain time slot in a period so that the adversary would not be able to profile the difference in communication patterns among all the nodes. Finally, this dissertation integrates the proposed privacy preserving framework into Snoogle, a sensor nodes based search engine for the physical world. Snoogle allows people to search for the physical objects in their vicinity. The previously proposed privacy preserving schemes are applied in the application to achieve the flexible and resilient privacy preserving capabilities. In addition to security and privacy, Snoogle also incorporates a number of energy saving and communication compression techniques that are carefully designed for systems composed of low-cost, low-power embedded devices. The evaluation study comprises of the real world experiments on a prototype Snoogle system and the scalability simulations.
© The Author
Wang, Haodong, "A privacy preserving framework for cyber-physical systems and its integration in real world applications" (2009). Dissertations, Theses, and Masters Projects. Paper 1539623552.