Seer: Trend-Prediction-Based Geographic Message Forwarding in Sparse Vehicular Networks

Geographic message forwarding in vehicular ad hoc networks (VANET) has attracted much attention and become one of the most promising research areas recent years. In this paper, inspired with the intuition that drivers´ route are with high regularity, we propose a prediction-based message forwarding strategy named Seer. Seer trains a 2nd-order Markov model based on long-term historic trip GPS data. Then probabilistic predictions about driving trend is made by looking at the intersections the driver just passed by. Seer can work without special service such as the traffic navigation systems and it can avoid leaking the position privacy of the driver. With extensive simulation in ONE, we show that Seer can achieve higher packet delivery ratio and lower delay, comparing with random or position-based message forwarding strategies.