Customized Slicing for Industrial Applications

Wireless networks are required to provide mobile ultra-reliable and low-latency services for applications in vertical industries, promoting the implementation of smart manufacturing in Industry 4.0. Most of industrial applications require high reliability and low latency in data transmission, which brings challenge to integrate 5G technologies in industrial Internet. Time-sensitive networking (TSN) which provides deterministic and more reliable communications over standard Ethernet is preferred to interconnect with 5G system, overcoming the high cost of wired connections and lack of flexibility. In the scenarios of 5G TSN integration, satisfying the distinct requirements of TSN traffic flows and 5G traffic flows simultaneously is a challenging problem. This chapter first analyzes the characteristics of emerging industrial use cases and briefly reviews the standardization work on IEEE TSN and 5G Ultra-Low Latency (ULL). Then, three typical technical directions of 5G TSN integration and research studies in network slicing for wireless TSN are summarized. In order to achieve efficient resource partitioning while guaranteeing the service requirements, an artificial intelligence (AI) based time-sensitive radio access network (RAN) slicing framework is proposed in this chapter. Through AI-engine deployed in the fog nodes aside with the wireless TSN base station, scheduling the time-critical traffic and nontime-critical traffic cross the wired and wireless domains can be intelligent. In addition, the deep reinforcement learning based method for RAN slicing in 5G TSN integration provided by AI-engine is illustrated.