Resource Allocation for Edge Collaboration with Deep Deterministic Policy Gradient in Smart Railway

With the emergence of the streaming service and big data for smart railway, such as real-time ultrahigh definition video transmissions, the combination of Internet of things and edge computing is indeed one of the promising technologys to satisfy the high level requirement of the sixth generation wireless communication system, such as Gbit/s experience in high-speed railway (HSR) scene. However, due to the extreme complexity of joint optimization task offloading and resource allocation, and the demand for computing-constrained vehicular terminals of railway (VTR), it is urgent to introduce the mobile edge computing and resource allocation algorithms into intelligent railway systems. In this paper, we propose a deep deterministic policy gradient based edge collaboration resource allocation algorithm (DECRA). The proposed architecture can intelligently allocate edge computing resources and channel resources to achieve task offloading and computing migration. Numerical results demonstrate the effectiveness of the proposed DECRA algorithm, which achieves resource allocation smartly when the speed of the VTR is up to 240km/h, and makes DECRA practical for the HSR in the edge computing system.