基于Hawkes过程的车联网协同缓存及资源分配

Hawkes-based Collaborative Caching and Resource Allocation in Internet of Vehicles

随着网络流量呈指数级增长,能够访问多媒体内容的智能汽车也面临巨大的流量压力,为此提出了一种基于Hawkes过程更新内容流行度的车联网协同缓存及资源分配框架.研究了在路边单元和智能车辆中的协同缓存及资源分配策略,同时,考虑到内容缓存的更新周期远大于信道条件的变化周期,提出了双时间尺度模型.首先,使用基于Hawkes过程的方法,考虑内容请求的新鲜度和时效性,根据历史内容请求记录更新流行度;然后,对路边单元和车辆协作缓存策略的数据传输吞吐量和缓存能耗进行建模,以最大化边缘设备的缓存效益为目标,并利用深度强化学习求解优化问题.仿真结果表明,所提出策略相比其他策略可以得到更高的效益.

With the exponential growth of network traffic,intelligent vehicles that can access multimedia content are also faced with huge pressure of traffic.Therefore,a collaborative caching and resource allocation framework with updating content popularity based on Hawkes processes of Internet of vehicles was proposed.A collaborative caching and resource allocation strategy in roadside units and intelligent vehicles was studied.Considering that the update period of content caching was much larger than the change period of channel conditions,a double timescale model was proposed.First,while considering the freshness and timeliness of the content request,the popularity based on the historical content request record was updated by using the method of Hawkes process.Then,the data transmission throughput,caching energy consumption of roadside units and vehicles collaborative strategy was modeled with the goal to maximize the caching benefit of edge devices,and deep reinforcement learning was used to solve the optimization problem.Simulation results show that the proposed strategy can get higher benefit than other strategies.