The manufacturing industry is on a never-ending quest to improve efficiency. In recent years, a cornerstone in this effort has been the move to Industry 4.0, which seeks to radically change the digital aspects of industrial automation and manufacturing to better meet changing market demands. This involves the introduction of Autonomous Mobile Robots (AMRs), extended data collection and processing, and flexible reconfigurable manufacturing. These additions all necessitate reliable and efficient network communication and computing capabilities. Furthermore, many of the use cases require some form of mobility, thereby making wireless networking a requirement. 5G has been touted as a solution to satisfy industries’ needs; however, it is also surrounded by significant hype, and limited empirical evidence regarding its performance and practical viability for industrial applications. Addition- ally, wireless technologies introduce unique challenges, including interference, propagation issues, and compatibility with established industrial protocols. There is currently an inherent lack of experience with wireless and 5G in the industrial domain. 

This thesis aims to bridge this knowledge gap by systematically evaluating the maturity of commercially available wireless systems for industrial use. The central hypothesis is that commercial 5G communication has reached a sufficient level of maturity to serve as a key enabler for Industry 4.0, capable of replacing wired communication in many industrial manufacturing scenarios and enable new industrial use cases. The primary objective is to produce a set of empirically grounded guidelines for system implementors. 

To achieve this, the research addresses several critical questions concerning the performance trade-offs of different 5G deployment models (public vs. private), the comparative performance of licensed 5G versus unlicensed technologies like Wi-Fi 6 and MulteFire, and the impact of co-channel interference on co-existing private 5G networks. Using experimental evaluation, we demonstrate that multi-connectivity implemented at the transport layer can provide a viable way of achieving high reliability communication in factories and even in outdoor rural areas with poor coverage. This is achieved using a mix of terrestrial and non-terrestrial networks. Further, we explore the intersection of industrial protocols and wireless communication, and how protocol design influences real-world per- formance and scalability. 

Finally, we demonstrate how 5G-enabled edge cloud can be used to enable new use cases such as offloading real-time perception and control tasks for mobile robotic systems. Our results show that private indoor 5G SA delivers 7.3 ms round trip latency when mobile at the 99.99 percentile, whereas private radio access network combined with a public core delivers 12.3 ms, both under mobility. 

This shows that industrial customers do not necessarily have to deploy full- scale private 5G networks and can instead obtain much of the same latency and throughput performance through a hybrid solution. We also show how 5G SA networks are susceptible to co-channel interference, and in the worst-case scenario, this interference can jam the uplink channel and reduce capacity by up to 94 %, highlighting the importance of proper planning when deploying even private cellular networks. Our multi-connectivity results show significant gains from mixing terrestrial and non-terrestrial communication and allow us to eliminate coverage outages and provide less than 100 ms round-trip latency with 99.99 % reliability. Through the investigation of OPC UA Safety, we also show that protocol design can have a significant impact on real-world performance. Here we identify a potential design issue in the protocol over multicast IP. This causes significant traffic overhead with quadratic scaling. 

Finally, we demonstrate the importance of edge computing in ensuring low latency compute and demonstrate that a mobile robot’s navigation software can be offloaded with little to no impact on its operational performance and precision.

After the defence there will be a small reception at Fredrik Bajers Vej 7A/4-108