TY - JOUR
T1 - Machine Learning for 6G Enhanced Ultra-Reliable and Low-Latency Services
AU - Liu, Yan
AU - Deng, Yansha
AU - Nallanathan, Arumugam
AU - Yuan, Jinhong
N1 - Funding Information:
This work was supported in part by the Engineering and Physical Sciences Research Council (EPSRC), U.K., under Grant EP/W004348/1, in part by the Australian Research Council (ARC) Discovery Project under Grant DP220103596, in part by the ARC Linkage Project under Grant LP200301482, and in part by Shanghai Sailing Program, under Grant 23YF1449400.
Publisher Copyright:
© 2002-2012 IEEE.
PY - 2023/4/1
Y1 - 2023/4/1
N2 - Ultra-reliable and low-latency communications (URLLC), as one of the major communication services of the fifth-generation (5G) and the sixth-generation (6G) cellular networks, is critical to supporting a variety of emerging mission-critical applications. However, the modern mobile networks could not satisfy the latency and reliability requirements, as well as other Quality of Service (QoS) requirements, including spectrum efficiency, energy efficiency, capacity, jitter, round-trip delay, network coverage, etc. To fulfill diverse QoS requirements for various URLLC applications, machine learning (ML) solutions are promising for future 6G networks. In this article, we first categorize the 6G URLLC vision into three connectivity characteristics, including ubiquitous connectivity, deep connectivity, and holographic connectivity, with their corresponding unique QoS requirements. We then identify potential challenges in meeting these connectivity requirements, and investigate promising ML solutions to achieve the intelligent connectivity for the 6G URLLC service. We further discuss how to implement the ML algorithms to guarantee the QoS requirements for different URLLC scenarios, including mobility URLLC, massive URLLC, and broadband URLLC. Finally, we present a case study of downlink URLLC channel access problems, solved by centralized deep reinforcement learning (CDRL) and federated DRL (FDRL), respectively, which validates the effectiveness of machine learning for URLLC services.
AB - Ultra-reliable and low-latency communications (URLLC), as one of the major communication services of the fifth-generation (5G) and the sixth-generation (6G) cellular networks, is critical to supporting a variety of emerging mission-critical applications. However, the modern mobile networks could not satisfy the latency and reliability requirements, as well as other Quality of Service (QoS) requirements, including spectrum efficiency, energy efficiency, capacity, jitter, round-trip delay, network coverage, etc. To fulfill diverse QoS requirements for various URLLC applications, machine learning (ML) solutions are promising for future 6G networks. In this article, we first categorize the 6G URLLC vision into three connectivity characteristics, including ubiquitous connectivity, deep connectivity, and holographic connectivity, with their corresponding unique QoS requirements. We then identify potential challenges in meeting these connectivity requirements, and investigate promising ML solutions to achieve the intelligent connectivity for the 6G URLLC service. We further discuss how to implement the ML algorithms to guarantee the QoS requirements for different URLLC scenarios, including mobility URLLC, massive URLLC, and broadband URLLC. Finally, we present a case study of downlink URLLC channel access problems, solved by centralized deep reinforcement learning (CDRL) and federated DRL (FDRL), respectively, which validates the effectiveness of machine learning for URLLC services.
UR - http://www.scopus.com/inward/record.url?scp=85156089018&partnerID=8YFLogxK
U2 - 10.1109/MWC.006.2200407
DO - 10.1109/MWC.006.2200407
M3 - Article
AN - SCOPUS:85156089018
SN - 1536-1284
VL - 30
SP - 48
EP - 54
JO - IEEE WIRELESS COMMUNICATIONS
JF - IEEE WIRELESS COMMUNICATIONS
IS - 2
ER -