During the digital transformation process of the society, emerging industry applications with critical data streams present great challenges to current networks. Traditional Ethernet provides best-effort frame delivery service without any performance guarantee such as latency and frame loss. Industrial Ethernet provides real-time and reliable communication services for industrial automation systems. However, the standards of industrial Ethernet are not compatible with each other, and the networking of different industrial systems becomes a great challenge. Time-Sensitive Networking (TSN) based on Ethernet provides high-quality multi-service flow transmissions on the same network through enhanced functions of time synchronization, deterministic flow scheduling and seamless reliability. It enables high-reliable and bounded low-latency flow transmission services. It is one of the key networking technologies for vertical applications such as industrial automation systems, in-vehicle networking, smart grid, and so on. Firstly, this survey introduces the standardization status of TSN from the following five aspects: time synchronization, low-latency flow control, reliability, network control and management, and use cases. The problems and key technologies of these standards are analyzed. For example, Time synchronization protocols such as NTP (Network Time Protocol), PTP (Precise Time Protocol), and gPTP (generalized PTP) are introduced and compared with a focus on the synchronization algorithms. Different flow scheduling mechanisms covered in the standards are described, analyzed, and compared qualitatively, based on their characteristics and performances. Network redundancy scheme based on the frame replication and elimination for reliability (FRER) is introduced in detail. Three models of TSN control and management are presented. They are centralized model, distributed model, and hybrid model. Vertical applications of TSN are briefly outlined including: professional audio and video, industrial automation, mobile fronthaul, service provider network, in-vehicle networking, and smart grid. Secondly, we survey the research state-of-the-art of TSN with a focus on the advantages and disadvantages of current solutions. Three research areas covered in this survey are time synchronization, flow control, and reliability. The main challenges to time synchronizations are algorithms enabling fast and efficient synchronization process, and clock precision. Current research works on the low-latency flow control mainly focus on traffic shaping, traffic scheduling, and traffic preemption. Traffic shaping problems include delay analysis, shaping mechanism and shaping overhead. Research topics of traffic scheduling include optimization algorithms, joint scheduling of routing and slot allocation, and software defined TSN. Flow preemption research works try to solve the problems of jitter and the adversary effect to the preempted traffic flow. Research works on TSN reliability include redundancy mechanisms, fault detection and restoration, and reliability of time synchronizations. Finally, research challenges and potential research directions are discussed. Future research works include TSN networking, Wireless TSN, and vertical applications of TSN. How to integrate TSN with other network layers remains to be a research problem. Due to the inherent characteristics of wireless channels, design of wireless networks with deterministic properties similar to TSN is a great challenge. Vertical applications of TSN such as 5G fronthaul, industrial automation systems, and in-vehicle networking are envisioned. In a word, TSN is a promising network technology in the fourth industrial revolution of the society. © 2021, Science Press. All right reserved.
