计算机工程

• •    

车联网中视频内容理解任务的计算卸载决策研究

  

  • 发布日期:2021-01-14

Research on Computing Offloading Decision of Video Content Understanding Task in Internet of Vehicles

  • Published:2021-01-14

摘要: 视频信息为车辆的智能网联化提供了丰富的信息,提高视频内容理解的精度成为推进车辆网联化的重大挑战。本文 提出了内容驱动的计算卸载指导方式和一种基于改进蒙特卡洛树搜索的计算卸载决策算法。通过对视频内容的预处理,有效 分析视频内容理解任务的重要性,进而采用基于强化学习的启发式搜索算法进行计算卸载决策,并采用了 DNN 进行了优化。 仿真对比表明,所提出的算法能够在时延约束下有效地降低能耗并提升视频内容理解任务的精度,并且收敛速度快、复杂度 低。

Abstract: Video information provides a wealth of information for the intelligent network of vehicles. Improving the accuracy of video content understanding has become a major challenge to promote the network of vehicles. Therefore, how to make more effective calculation and unloading decisions to improve the accuracy of video content understanding has become an important issue. In this paper, a content - driven instruction method and a decision - making algorithm based on improved Monte Carlo tree search (MCTS) are presented. Through the preprocessing of video content, the importance of video content analysis task was effectively analyzed, and then the heuristic search algorithm based on reinforcement learning was adopted to calculate the unloading decision, and DNN was adopted for optimization. Simulation results show that the proposed algorithm can effectively reduce energy consumption and improve the accuracy of video content understanding tasks under the delay constraint, which has a fast convergence speed and low complexity.