Research On Trajectory-Based Download Task Scheduling For Drive-Thru Internet Via WiFi

A vehicular network uses cars as mobile nodes and uses Internet access points(APs) and roadside units(RSUs) as fixed nodes to create a wireless and mobile network, the direct communication between vehicles, between vehicles and RSUs using an Ad Hoc network, the direct communication between vehicles and APs using Internet via Wi Fi.Vehicle can access Internet to obtain various information services, such as querying shopping malls, restaurants, gas stations, traffic management, road conditions and the latest weather, news, etc. And these vehicles also obtain these services whose fee is more expensive through 3G or satellite, however, the vehicles generally can obtain free and high-performance Internet link through APs. For this reason, important research content on the vehicle network is how to take advantage of AP to provide Internet access services for the vehicular users and especially their download service.Through governments have invested heavily in the deployment of a variety of APs, but these efforts are still unable to meet the growing demand for users to access Internet. Currently, communication ranges of APs in the developed cities can not fully cover vehicles running on the streets.The uncovered areas among APs is called hole area. Vehicles running in hole area cannot access Internet through Aps, however, they can access Internet via multi-hop using Ad Hoc network.If APs can’t fully meet download requests of all vehicles, they should schedule the download requests received to improve service capacities of the entire vehicle networks. Obviously, if a node that can provide forwarding data for more adjacent nodes, then its download request should have a higher priority than the others and it preferentially be scheduled for execution.Nodes’ scheduling priorities depend on the download request categories of adjacent nodes and the vehicle network’s topology. The categories of download request are relatively easy to be distinguished, yet vehicle network topology is dynamic. Therefore, the key issue is to get network topology and determine a scheduling priority based on network topology.This paper studies AP’s downloading task scheduling problems based on dynamic network topology, network topology mainly relates to vehicles’ trajectories. Some vehicles send their GPS information to server periodically, such as taxis, their trajectories are certain; Others’ trajectories are uncertain. Therefore, this paper will study AP’s downloading task scheduling problems under the two circumstances which vehicle trajectories are certain and uncertain, as following.First, when vehicle trajectories are certain, this paper studies a single AP’s downloading task scheduling problem based on ad hoc links among vehicles.Owing to vehicles in hole area without Wi Fi, they can only download data from the other adjacent nodes which have downloaded data needed. The nodes that can forward more data for adjacent nodes should have higher priorities. Based on this, this paper studies a single AP’s downloading task scheduling problem maximizing the total quantity of data downloaded throughout the hole area. However, this problem is proven to be NP-complete. Therefore an approximate algorithm is presented to solve it in this paper. Experimental results show that it is more efficient than other existing algorithms.Second, when vehicle trajectories are uncertain, this paper studies a single AP’s downloading task scheduling problem based on opportunistic links among vehicles in hole area.According to current running states, link uncertainty caused by uncertain trajectories could be predicted. Therefore, based on opportunistic reachability graph that has been predicted, this paper studies a single AP’s downloading task scheduling problem to maximize the downloading success rate. However this paper has proved that this problem is NP-complete. For this reason, this paper proposes an APL approximate algorithm based on opportunistic reachability graph. Experimental results show that APL is more effective and can gain better performance.Third, under the circumstance which vehicles have never run through coverage range of access point(AP), this paper studies no AP’s task scheduling problem based on opportunistic links according to the historical statistic.Vehicles can obtain data based on that the other vehicles serve as mobile gateways(MG) to save-carry-forward data for them, even though they have never run through AP. Owing to buses have the following features: dense trajectories, regular routes, predicable mobility, etc, they are the most appropriate MGs, This paper studies no AP’s task scheduling problem based on bus MGs, and proposes a task scheduling method based MGs forwarding to minimize expected delay, extensive experiments are shown that it provides an expected minimum delay under the premise of guaranteeing Qo S.Last, to solve the problem of communication load imbalance, this paper studies multi-AP task scheduling problem based on the spatial-temporal distribution of download request. Aiming at the problem of communication load imbalance, this paper utilizes the prominent characteristic of APs’ communication-load with spatial-temporal change, and models the spatial-temporal distribution of download request as task scheduling queue graphics sequence. Based on this graphics sequence, this paper proposes multi-AP task scheduling algorithm. Experimental results show that it has balanced communication load of the whole network from time and space.