Given the advent of Intelligent Transportation Systems (ITSs), drivers and passengers could now spend more of their time enjoying entertainment applications, e.g., watching TV or streaming movies. However, such services can drastically increase the traffic load on the existing network infrastructure (i.e. Roadside Units (RSU) and Cellular Base Stations (CBS)). Recently, Unmanned Aerial Vehicles (UAVs) have been playing a remarkable role in offloading terrestrial networks and providing cellular services thanks to their agility and flexibility. Hence, this paper explores a cooperative approach for content caching and delivery in the context of internet of connected vehicles, where a RSU, having access to a library of contents but with limited communication coverage, collaborates with a UAV to deliver contents to vehicles on a road segment. In this context, the connected RSU is responsible for delivering contents to the UAV cache unit by leveraging passing by vehicles. The RSU loads the contents on these vehicles that in turn upload them to the UAV cache unit. We model this cooperation problem mathematically as a mixed integer non-linear programming (MINLP) problem with the objective to maximize the number of served vehicles. Owing to the complexity of solving this problem, it is alternatively cast as an MDP whose solution is obtained through a Dual-Task Reinforcement Learning method (DTDRL). Simulation results show the superiority of our proposed collaborative solution over non-collaborative methods.

鉴于智能交通系统 (ITS) 的出现，驾驶员和乘客现在可以将更多时间用于娱乐应用，例如看电视或流媒体电影。然而，此类服务会极大地增加现有网络基础设施（即路边单元 (RSU) 和蜂窝基站 (CBS)）的流量负载。最近，由于其敏捷性和灵活性，无人驾驶飞行器 (UAV) 在卸载地面网络和提供蜂窝服务方面发挥了显着的作用。因此，本文探讨了一种在联网车辆互联网环境下内容缓存和交付的合作方法，其中 RSU 可以访问内容库但通信覆盖范围有限，与无人机合作，将内容交付给车辆。路段。在这种情况下，连接的 RSU 负责利用车辆经过，将内容传送到 UAV 缓存单元。RSU 将内容加载到这些车辆上，然后将它们上传到无人机缓存单元。我们将此合作问题在数学上建模为混合整数非线性规划 (MINLP) 问题，其目标是最大化服务车辆的数量。由于解决这个问题的复杂性，它或者被转换为一个 MDP，其解决方案是通过双任务强化学习方法 (DTDRL) 获得的。仿真结果表明我们提出的协作解决方案优于非协作方法。RSU 将内容加载到这些车辆上，然后将它们上传到无人机缓存单元。我们将此合作问题在数学上建模为混合整数非线性规划 (MINLP) 问题，其目标是最大化服务车辆的数量。由于解决这个问题的复杂性，它或者被转换为一个 MDP，其解决方案是通过双任务强化学习方法 (DTDRL) 获得的。仿真结果表明我们提出的协作解决方案优于非协作方法。RSU 将内容加载到这些车辆上，然后将它们上传到无人机缓存单元。我们将此合作问题在数学上建模为混合整数非线性规划 (MINLP) 问题，其目标是最大化服务车辆的数量。由于解决这个问题的复杂性，它或者被转换为一个 MDP，其解决方案是通过双任务强化学习方法 (DTDRL) 获得的。仿真结果表明我们提出的协作解决方案优于非协作方法。它或者被转换为 MDP，其解决方案是通过双任务强化学习方法 (DTDRL) 获得的。仿真结果表明我们提出的协作解决方案优于非协作方法。它或者被转换为 MDP，其解决方案是通过双任务强化学习方法 (DTDRL) 获得的。仿真结果表明我们提出的协作解决方案优于非协作方法。