Performance Analysis of Drive-thru Internet Access

Drive-thru Internet is considered to be an important solution to provide Internet access for vehicles. By deploying cost-effective and high bandwidth roadside WiFi networks, a vehicle can upload/download considerable data when drive through the coverage area, whereby a myriad of automotive applications can be employed, such as intelligent transportation system, infotainment applications like video/audio streaming, webpage browsing, etc. However, the high mobility of vehicles leads to the intermittent connection between a vehicle and roadside Access Points (APs), which would cause the Internet access delay and throughput degradation. In this thesis, we propose comprehensive modeling and analysis for the drive-thru Internet access performance considering the overhead of the access procedure, which includes the steps of network detection, user authentication and network parameters assignment. We also consider the situation that a vehicle drives through multiple roadside APs' coverage areas and evaluate the performance of traffic offloading from cellular networks to roadside WiFi networks. In specific, firstly, we develop an analytical model to study the dependency of the drive-thru Internet access delay with different factors, i.e., the wireless channel conditions, the number of co-associated WiFi clients, and the employed authentication mechanism, such as the WiFi Protected Access II (WPA2)-Pre-Shared Key (PSK) and the WPA2-802.1X modes. The access procedure is modeled as a discrete Markov chain to calculate the time to exchange all management frames and to evaluate the Internet access delay. The accuracy of the analytical model is studied via computer simulations, as well as experimental testing using Commercial Off-The-Shelf (COTS) WiFi products, together with a channel emulator that emulates the wireless channel conditions in a vehicular environment. Simulation and experiment results validate the accuracy of the proposed analytical model which provides useful guidelines for future selection/development of suitable WiFi network access schemes in a vehicular environment. Secondly, we take a further step to analyze the throughput performance of the drive-thru Internet access. The mobility of the vehicle is modeled as the transition of a series of zones in the coverage area, which is defined based on the relationship between the WiFi link rate and the distance of the AP and the vehicle. A three dimensional (3D) Markov model is proposed to combine the zone transition process and the transmission of the management frames and calculate the average throughput under conditions of different numbers of co-associated WiFi clients, channel qualities and different access protocols. Thirdly, we consider that when the vehicle drives through multiple roadside WiFi networks, and employ the Vehicle-to-Vehicle (V2V) assisted WiFi offloading mechanism, where nearby vehicles that associated to different APs can use their idle WiFi resource to offload part of peer's data traffic. The offloading performance is calculated by modeling the intermittent WiFi transmission as an M/G/1/K queueing process, and the performance gain of the V2V assistance is also analyzed. In summary, the research works in this thesis should provide guidelines for future research and development of drive-thru Internet