| dc.description.abstract | Cyber-physical systems -- physical processes that are controlled by computational frameworks across networks -- have seen a large scale digitalization and democratization due to the improvements in edge computing, the internet-of-things, and machine learning. Edge computing enables the use of computing resources that are close to the physical processes; the internet-of-things has enabled improved and connected sensing across domains; and artificial intelligence that utilizes machine-learning techniques has enabled improved decision making at the edge for these systems. The effects of these systems can be seen in our cities, even down to components of the critical infrastructure. These systems present many novel opportunities which can improve our communities, including that of transactive energy, where home-owners and businesses who own and operate energy resources, and who (along with the bulk grid) are able to exchange energy to their mutual benefit. Another example is in the area of outsourcing computations to devices at the edge, which would allow us to take advantage of under-utilized compute power. While rapid progress has been made in building such cyber-physical applications, the new system architectures have also led to increasing concerns regarding issues such as trust between system participants in the context of competing interests and reliability. These concerns become especially apparent when we consider that the subsystems of modern cyber-physical systems are owned, operated, and maintained by independent stakeholders; that is, they are managed by different organizations or individuals, and must coordinate if they are to take coordinated actions on the physical world.
Blockchain-based distributed ledgers have arisen as a mechanism to address some of the concerns that arise in multi-stakeholder cyber-physical systems (MSCPS). Distributed ledgers allow for the construction of trusted systems wherein that trust can be decentralized among the participants. The disadvantage of distributed ledgers is that each participant must duplicate the useful work performed and so the computational power is limited to the lowest common denominator. Additionally, the process for determining which participant has permission to write to the data structure is expensive. However, blockchain-based ledgers provide a valuable tool for cases where trust cannot be centralized.
This dissertation discusses challenges inherent to MSCPS, when and where distributed ledgers can be applied to address these challenges, and the complications that can arise due to their use. Specifically, this work contains contributions in three areas to the state of the art and to efforts to mitigate these challenges, specifically in the contexts of transactive energy and outsourced computation. First, this dissertation addresses the conflicting requirements of privacy and safety in a transactive energy market. Second, this dissertation presents market protocols designed to enable trust in decentralized systems without relying exclusively on cryptographic protocols. Third, this dissertation presents a hybrid solver design pattern for blockchain-based markets, which allows stakeholders to coordinate without a central authority while overcoming some of the inefficiencies of a purely blockchain-based market, allowing for reduction of costs and preservation of trust. Through integration with model-based middleware, resilience in MSCPS markets is also enabled. The work described herein has been codified as TRANSAX, which is a platform which allows homeowners to safely and privately trade energy resources; MODiCuM, which allows mistrusted customers to outsource computation to mistrusted suppliers by following a protocol which enables mutually beneficial transactions; and SolidWorx, which is a domain-neutral architecture for the trading of resources in smart and connected communities. | |
