Internet of Vehicles (IoV) enables numerous in-vehicle applications for smart cities, driving increasing service demands for processing various contents (e.g., videos). Generally, for efficient service delivery, the contents from the service providers are processed on the edge servers (ESs), as edge computing offers vehicular applications low-latency services. However, due to the reusability of the same contents required by different distributed vehicular users, processing the copies of the same contents repeatedly in an edge server leads to a waste of resources (e.g., storage, computation, and bandwidth) in ESs. Therefore, it is a challenge to provide high-quality services while guaranteeing the resource efficiency with edge content caching. To address the challenge, an edge content caching method for smart cities with service requirement prediction, named E-Cache, is proposed. First, the future service requirements from the vehicles are predicted based on the deep spatiotemporal residual network (ST-ResNet). Then, preliminary content caching schemes are elaborated based on the predicted service requirements, which are further adjusted by a many-objective optimization aiming at minimizing the execution time and the energy consumption of the vehicular services. Eventually, experimental evaluations prove the efficiency and effectiveness of E-Cache with spatiotemporal traffic trajectory big data.

中文翻译：

---

面向智能城市车联网的深度时空残差网络边缘内容缓存

车联网 (IoV) 为智慧城市提供了众多车载应用，推动了处理各种内容（例如视频）的服务需求不断增加。通常，为了高效的服务交付，来自服务提供商的内容在边缘服务器 (ES) 上进行处理，因为边缘计算为车辆应用程序提供低延迟服务。然而，由于不同分布式车辆用户所需的相同内容的可重用性，在边缘服务器中重复处理相同内容的副本会导致ES中的资源（例如，存储、计算和带宽）浪费。因此，在通过边缘内容缓存保证资源效率的同时提供高质量的服务是一个挑战。为了应对挑战，提出了一种服务需求预测的智慧城市边缘内容缓存方法E-Cache。首先，基于深度时空残差网络（ST-ResNet）预测来自车辆的未来服务需求。然后，根据预测的服务需求制定初步的内容缓存方案，并通过多目标优化进一步调整，以最小化车辆服务的执行时间和能耗。最终，实验评估证明了 E-Cache 与时空交通轨迹大数据的效率和有效性。基于预测的服务需求制定初步的内容缓存方案，并通过多目标优化进一步调整，旨在最大限度地减少车辆服务的执行时间和能耗。最终，实验评估证明了 E-Cache 与时空交通轨迹大数据的效率和有效性。基于预测的服务需求制定初步的内容缓存方案，并通过多目标优化进一步调整，旨在最大限度地减少车辆服务的执行时间和能耗。最终，实验评估证明了 E-Cache 与时空交通轨迹大数据的效率和有效性。