Abstract

Wireless closed-loop control is of major significance for future industrial manufacturing. However, control applications pose stringent quality of service requirements for reliable operation. Contrary to traditional ultra-reliable low-latency communications design goals such as low packet loss rates and low latency, research results in the domain of networked control systems (NCS) state that depending on the sampling period, control applications inherently tolerate a few consecutive packet losses. This translates into a better-suited metric to capture control application requirements and therefore a more conclusive design goal for wireless networks: ensuring a maximum age of information (AoI). With a Markov modeling approach, we propose to exploit the tolerance through a novel dynamic multi-connectivity scheme that we term state-aware resource allocation (SARA), which temporally negatively correlates packet losses, thus avoiding long packet loss sequences. Through statistical multiplexing, SARA enables a mean time to failure (MTTF) in the order of years while keeping the per-agent average channel usage close to one, also in a multi-agent setting with competition for resources. Compared with static dual-connectivity, the MTTF can be increased 100-fold whereas the number of required channels reduces by 40%. Our approach also statistically guarantees system-wide AoI distributions, which aid to ensure control performance.

Highlights

  • T HE fifth generation of cellular networks (5G) is a major enabler for digitalization across a wide spectrum of industries, including healthcare, manufacturing, automotive and energy

  • Building on the knowledge that most control applications are able to tolerate a certain number of consecutive packet losses as long as a given maximum bound on the age of information (AoI) is not exceeded, this article introduces the dynamic multiconnectivity concept state-aware resource allocation (SARA) that is able to adaptively change the number of parallel channels that are used for transmission depending on the number of packets that were lost previously

  • The system-wide MTTFsys is considered in addition to the single-agent mean time to failure (MTTF), which leads to much stronger statements, as each agent is considered as a single point of failure

Read more

Summary

INTRODUCTION

T HE fifth generation of cellular networks (5G) is a major enabler for digitalization across a wide spectrum of industries, including healthcare, manufacturing, automotive and energy. The control engineering research community has made major progress investigating so-called networked control systems (NCSs), dealing with the question of how to cope with communications imperfections (unreliable, latency-prone, data-rate-limited, ...) and how to reduce their impact on control performance through developing robust control algorithms These NCS research results stand in stark contrast to previously assumed QoS requirements for closed-loop control [8], which eventually sparked URLLC development. They demonstrate that while real-time applications profit from network capabilities beyond enhanced mobile broadband (eMBB), ultra-reliable communication (in terms of low packet loss rate (PLR)) is usually not required for ultra-reliable applications. We present two admission control schemes termed random and cliff to decide which agents will be underserved

RELATED WORK
VoI and AoI
SYSTEM MODEL
SINGLE-AGENT ANALYSIS
Failure Model
Performance Metrics
State-Aware Resource Allocation
MULTI-AGENT ANALYSIS
Results
Findings
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call