The ongoing effort for an efficient, sustainable, and automated interaction between humans, machines, and our environment will make cyber-physical systems (CPS) an integral part of the industry and our daily lives. At their core, CPS integrate computing elements, communication networks, and physical processes that are monitored and controlled through sensors and actuators. New and innovative applications become possible by extending or replacing static and expensive cable-based communication infrastructures with wireless technology. The flexibility of wireless CPS is a key enabler for many envisioned scenarios, such as intelligent factories, smart farming, personalized healthcare systems, autonomous search and rescue, and smart cities.
High dependability, efficiency, and adaptivity requirements complement the demand for wireless and low-cost solutions in such applications. For instance, industrial and medical systems should work reliably and predictably with performance guarantees, even if parts of the system fail. Because emerging CPS will feature mobile and battery-driven devices that can execute various tasks, the systems must also quickly adapt to frequently changing conditions. Moreover, as applications become ever more sophisticated, featuring compact embedded devices that are deployed densely and at scale, efficient designs are indispensable to achieve desired operational lifetimes and satisfy high bandwidth demands.
Meeting these partly conflicting requirements, however, is challenging due to imperfections of wireless communication and resource constraints along several dimensions, for example, computing, memory, and power constraints of the devices. More precisely, frequent and correlated message losses paired with very limited bandwidth and varying delays for the message exchange significantly complicate the control design. In addition, since communication ranges are limited, messages must be relayed over multiple hops to cover larger distances, such as an entire factory. Although the resulting mesh networks are more robust against interference, efficient communication is a major challenge as wireless imperfections get amplified, and significant coordination effort is needed, especially if the networks are dynamic.
CPS combine various research disciplines, which are often investigated in isolation, ignoring their complex interaction. However, to address this interaction and build trust in the proposed solutions, evaluating CPS using real physical systems and wireless networks paired with formal guarantees of a system’s end-to-end behavior is necessary. Existing works that take this step can only satisfy a few of the abovementioned requirements. Most notably, multi-hop communication has only been used to control slow physical processes while providing no guarantees. One of the reasons is that the current communication protocols are not suited for dynamic multi-hop networks.
This thesis closes the gap between existing works and the diverse needs of emerging wireless CPS. The contributions address different research directions and are split into two parts. In the first part, we specifically address the shortcomings of existing communication protocols and make the following contributions to provide a solid networking foundation:
• We present Mixer, a communication primitive for the reliable many-to-all message exchange in dynamic wireless multi-hop networks. Mixer runs on resource-constrained low-power embedded devices and combines synchronous transmissions and network coding for a highly scalable and topology-agnostic message exchange. As a result, it supports mobile nodes and can serve any possible traffic patterns, for example, to efficiently realize distributed control, as required by emerging CPS applications.
• We present Butler, a lightweight and distributed synchronization mechanism with formally guaranteed correctness properties to improve the dependability of synchronous transmissions-based protocols. These protocols require precise time synchronization provided by a specific node. Upon failure of this node, the entire network cannot communicate. Butler removes this single point of failure by quickly synchronizing all nodes in the network without affecting the protocols’ performance.
In the second part, we focus on the challenges of integrating communication and various control concepts using classical time-triggered and modern event-based approaches. Based on the design, implementation, and evaluation of the proposed solutions using real systems and networks, we make the following contributions, which in many ways push the boundaries of previous approaches:
• We are the first to demonstrate and evaluate fast feedback control over low-power wireless multi-hop networks. Essential for this achievement is a novel co-design and integration of communication and control. Our wireless embedded platform tames the imperfections impairing control, for example, message loss and varying delays, and considers the resulting key properties in the control design. Furthermore, the careful orchestration of control and communication tasks enables real-time operation and makes our system amenable to an end-to-end analysis. Due to this, we can provably guarantee closed-loop stability for physical processes with linear time-invariant dynamics.
• We propose control-guided communication, a novel co-design for distributed self-triggered control over wireless multi-hop networks. Self-triggered control can save energy by transmitting data only when needed. However, there are no solutions that bring those savings to multi-hop networks and that can reallocate freed-up resources, for example, to other agents. Our control system informs the communication system of its transmission demands ahead of time so that communication resources can be allocated accordingly. Thus, we can transfer the energy savings from the control to the communication side and achieve an end-to-end benefit.
• We present a novel co-design of distributed control and wireless communication that resolves overload situations in which the communication demand exceeds the available bandwidth. As systems scale up, featuring more agents and higher bandwidth demands, the available bandwidth will be quickly exceeded, resulting in overload. While event-triggered control and self-triggered control approaches reduce the communication demand on average, they cannot prevent that potentially all agents want to communicate simultaneously. We address this limitation by dynamically allocating the available bandwidth to the agents with the highest need. Thus, we can formally prove that our co-design guarantees closed-loop stability for physical systems with stochastic linear time-invariant dynamics.:Abstract
Acknowledgements
List of Abbreviations
List of Figures
List of Tables
1 Introduction
1.1 Motivation
1.2 Application Requirements
1.3 Challenges
1.4 State of the Art
1.5 Contributions and Road Map
2 Mixer: Efficient Many-to-All Broadcast in Dynamic Wireless Mesh Networks
2.1 Introduction
2.2 Overview
2.3 Design
2.4 Implementation
2.5 Evaluation
2.6 Discussion
2.7 Related Work
3 Butler: Increasing the Availability of Low-Power Wireless Communication Protocols
3.1 Introduction
3.2 Motivation and Background
3.3 Design
3.4 Analysis
3.5 Implementation
3.6 Evaluation
3.7 Related Work
4 Feedback Control Goes Wireless: Guaranteed Stability over Low-Power Multi-Hop Networks
4.1 Introduction
4.2 Related Work
4.3 Problem Setting and Approach
4.4 Wireless Embedded System Design
4.5 Control Design and Analysis
4.6 Experimental Evaluation
4.A Control Details
5 Control-Guided Communication: Efficient Resource Arbitration and Allocation in Multi-Hop Wireless Control Systems
5.1 Introduction
5.2 Problem Setting
5.3 Co-Design Approach
5.4 Wireless Communication System Design
5.5 Self-Triggered Control Design
5.6 Experimental Evaluation
6 Scaling Beyond Bandwidth Limitations: Wireless Control With Stability Guarantees Under Overload
6.1 Introduction
6.2 Problem and Related Work
6.3 Overview of Co-Design Approach
6.4 Predictive Triggering and Control System
6.5 Adaptive Communication System
6.6 Integration and Stability Analysis
6.7 Testbed Experiments
6.A Proof of Theorem 4
6.B Usage of the Network Bandwidth for Control
7 Conclusion and Outlook
7.1 Contributions
7.2 Future Directions
Bibliography
List of Publications
| Identifer | oai:union.ndltd.org:DRESDEN/oai:qucosa:de:qucosa:86764 |
| Date | 09 August 2023 |
| Creators | Mager, Fabian |
| Contributors | Zimmerling, Marco, Wollschläger, Martin, Picco, Gian Pietro, Technische Universität Dresden |
| Source Sets | Hochschulschriftenserver (HSSS) der SLUB Dresden |
| Language | English |
| Detected Language | English |
| Type | info:eu-repo/semantics/publishedVersion, doc-type:doctoralThesis, info:eu-repo/semantics/doctoralThesis, doc-type:Text |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | info:eu-repo/grantAgreement/German Research Foundation/SPP 1914/ZI 1635/1-1//coCPS |
Page generated in 0.0022 seconds