Exploring Wi-Fi Network Diversity for Vehicle-To-Infrastructure Communication

Due to their ubiquity and low use-cost, the opportunistic use of Wi-Fi networks to offload data from moving vehicles is enticing. However, due to their limited coverage and variable performance, choosing what Access Points (APs) to use in order to maximize the amount of data that can be offloaded is challenging. This difficulty is exacerbated by the heterogeneity created by the introduction of new Wi-Fi standards such as 802.11ad, which renders heuristics designed for homogeneous environments, e.g., signal quality, ineffective. In this work we test the hypothesis that historical network performance, indexed by vehicular mobility information, can be used to effectively forecast future network performance, and consequently help select APs for data offloading in a heterogeneous Wi-Fi environment. Our approach was to perform a trace-based analysis on experimental data collected in a realistic vehicular environment. Our results show that a practical algorithm based on data rate forecasting from mobility information was able to transfer at least 80% of the optimal amount of data, under the tested scenarios.