Autonomous vehicles will rely heavily on vehicle-to-everything (V2X) communications to obtain a large amount of information required for navigation and road safety purposes. This can be achieved through: (i) leveraging millimeter-wave (mmWave) frequencies to achieve multi- Gbps data rates, and (ii) exploiting the temporal and spatial correlation of vehicular contents to offload a portion of the traffic from the infrastructure via caching. Characterizing such a system under mmWave directional beamforming, high vehicular mobility, channel fluctuations, and different caching strategies is a complex task. In this article, we propose the first stochastic geometry framework for caching in mmWave V2X networks, which is validated via rigorous Monte Carlo simulation. In addition to common parameters considered in stochastic geometry models, our derivations account for caching as well as the speed and the trajectory of the vehicles. Furthermore, our evaluations provide interesting design insights: (i) higher base station/vehicle densities does not necessarily improve caching performance; (ii) although using a narrower beam leads to a higher SINR, it also reduces the connectivity probability; and (iii) V2X caching can be an inexpensive way of compensating some of the unwanted mmWave channel characteristics.

中文翻译：

---

支持毫米波的 V2X 缓存的分析框架

自动驾驶汽车将严重依赖车对一切 (V2X) 通信来获取导航和道路安全所需的大量信息。这可以通过以下方式实现：(i) 利用毫米波 (mmWave) 频率来实现数 Gbps 的数据速率，以及 (ii) 利用车辆内容的时间和空间相关性，通过缓存从基础设施中卸载一部分流量. 在毫米波定向波束成形、高车辆移动性、信道波动和不同缓存策略下表征此类系统是一项复杂的任务。在本文中，我们提出了第一个用于在毫米波 V2X 网络中缓存的随机几何框架，该框架通过严格的蒙特卡罗模拟进行了验证。除了随机几何模型中考虑的常用参数外，我们的推导考虑了缓存以及车辆的速度和轨迹。此外，我们的评估提供了有趣的设计见解：(i) 更高的基站/车辆密度不一定会提高缓存性能；(ii) 虽然使用更窄的波束会导致更高的 SINR，但它也降低了连接概率；(iii) V2X 缓存可以作为一种廉价的方式来补偿一些不需要的毫米波信道特性。