Lionel M. Ni

Geocast in vehicular networks aims to deliver a message to a target geographical region, which is useful for many applications such as geographic advertising. This is a highly challenging task in vehicular network environments due to the rare encounter opportunities and uncertainty caused by vehicular mobility. As more vehicles are equipped with on-board navigation systems, vehicle trajectories are ready for exploitation. We observe that a vehicle has a higher capability of delivering a message to the target region if its own future trajectory or trajectories of those vehicles to be encountered overlap the target region. Motivated by this observation, we develop a message forwarding metric, called coverage capability, to characterize the capability of a vehicle to successfully geocast the message. When calculating the coverage capability, we are facing the major challenge raised by the absence of accurate vehicle arrival time. Through an empirical study using real vehicular GPS traces of 2,600 taxis, we verify that the travel time of a vehicle, which is modeled as a random variable, follows the Gamma distribution. The travel time modeling helps us to make accurate predictions for inter-vehicle encounters. We perform extensive trace-driven simulations and the results show that our approach achieves 37.4 percent higher delivery ratio and 43.1 percent lower transmission overhead comparing with GPSR which is a representative geographic routing protocol.

We propose a novel entity authentication approach for pervasive computing environments. A person uses a single device, the master key, which aggregates all his digital forms of access tokens for entity authentication. The master key discovers and selects proper tokens for its owner. With an emphasis on usability, the master key secures authentication, protects privacy information from outsiders and insiders, and supports various claimant-verifier relations. We analyze privacy and security properties of our approach and protocols, and we investigate the overhead. Performance measurements show that our protocols are efficient