Abstract
The advanced services provided by vehicle ad hoc networks (VANETs) often require vehicles to process complex computing tasks that may not be completed by individual vehicles within a required delay limit. Offloading tasks to road side units (RSUs) is a typical approach to enhancing service performance in VANET; however, RSUs may not always have sufficient resources for handling all task offloading requests. With the increasing amount of computing capacities available on vehicles, offloading tasks to other vehicles offers a promising alternative to RSU-based task offloading. However, vehicle-to-vehicle task offloading in VANET faces some new challenges that have not been fully addressed, among which is the degraded delay performance caused by vehicle mobility. In order to solve this problem, we propose an Online Pre-filtering Task Offloading System (OPTOS) that is able to mitigate the impact of vehicle mobility on task offloading performance. OPTOS comprises a process that selects candidate vehicles for hosting offloaded tasks and an HGSA algorithm that assigns tasks to vehicles for minimizing task completion delay while balancing utilization of computing capacities on different vehicles. We have conducted extensive experiments using a real-world dataset for evaluating the performance of the proposed OPTOS. Obtained results indicate that OPTOS is effective for reducing task completion delay and increasing task success rate in various VANET scenarios with different levels of vehicle mobility.
Highlights
In recent years, mobile vehicles are expected to provide drivers with more and more advanced services, including traffic accident analysis, traffic flow forecasting, route planning and traffic light management, etc [1]–[3]
We develop a half-greedy simulated annealing (HGSA) algorithm that is employed in Online Pre-filtering Task Offloading System (OPTOS) for assigning tasks to candidate vehicles and minimizing the total task completion delay
We focus our research on highway scenarios because vehicles moving on highways are more likely to form stable vehicular clouds, in which task offloading to vehicles becomes a reasonable alternative to the conventional road side units (RSUs)-based task offloading
Summary
Mobile vehicles are expected to provide drivers with more and more advanced services, including traffic accident analysis, traffic flow forecasting, route planning and traffic light management, etc [1]–[3]. The specific problem we want to solve in this paper is to find a task offloading strategy in a vehicular cloud that minimizes task completion delay with the negative impact of vehicle mobility eliminated. We first formulate vehicle-to-vehicle task offloading as an optimization problem based on our analysis on communication and computing delay in vehicular clouds. 1) We proposed OPTOS, an online pre-filtering task offloading system that fully considers the impact of vehicle mobility to obtain a reliable vehicle-to-vehicle offloading strategy. A key element of OPTOS is a step for filtering out vehicles that may leave the vehicular cloud before completing the tasks offloaded to them This step ensures that tasks are only assigned to reliable vehicles that can successfully complete their tasks without being interrupted by mobility avoiding the extra delay in task completion caused by task reassignment.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
