The aim of this thesis is to advance the knowledge on the efficient design and evaluation of distributed marketplaces with an emphasis on trust and privacy. Distributed systems are an integral part of today's computing infrastructures, enabling multiple nodes to work towards a common goal. Although distributed, most of today's computational systems are still built with a centralized architecture, which assumes complete governance by a single organization. In the case of centralized marketplaces, the correct trade execution guarantees, \ie digital trust, and data privacy are provided centrally, containing all processes and operations within a single organization's boundaries. This puts the marketplace operator in a prime position to govern trade settlement conditions. However, trust issues are raised if more than one organization has to govern the marketplace. In such a case, trust and privacy are decentralized, and control is distributed among all organizations which are part of the marketplace system. Thus, a decentralized marketplace requires a robust and secure consensus mechanism, which enables digital trust while allowing organizations to process and store private data for further usage in trade settlements. 

This thesis investigates both centralized and decentralized marketplace architectures applied to use cases of AI artifacts and renewable energy trading. It begins with a study of a marketplace for Artificial Intelligence (AI) artifacts where multiple organizations collaborate on AI pipeline execution. The study defines a Secure Virtual Premise, which enables AI pipeline execution in a centralized marketplace governed by a trusted third party. The thesis continues with a survey of the telecommunication services marketplaces, where both centralized and decentralized architectures are discussed. In addition, the survey provides an in-depth investigation of blockchain technology as a main trust-enabling platform, providing distributed storage and data assurance to all processes in a decentralized marketplace. Having mapped the state-of-the-art, the research shifts towards an in-depth investigation of blockchain-based decentralized renewable energy marketplaces. The main aim of such a marketplace is to incentivize the widespread adoption of renewable energy sources, resulting in the decarbonization of electricity distribution systems. The designed marketplace enables automation and trusted execution of peer-to-peer (P2P) energy trade settlements in decentralized systems while preserving users' data privacy. Furthermore, the marketplace is aligned with the data and P2P energy trade regulations. The studies provide an in-depth requirements definition, system architecture, implementation, and performance evaluation of marketplaces based on two major blockchain platforms. The final study of this thesis provides the improvements towards the renewable energy marketplace model aiming at an enhancement of trust, privacy, and scalability.

In this paper, we investigate how to design a security architecture of a Platform-as-a-Service (PaaS) solution, denoted as Secure Virtual Premise (SVP), for collaborative and distributed AI engineering using AI artifacts and Machine Learning (ML) pipelines. Artifacts are re-usable software objects which are a) tradeable in marketplaces, b) implemented by containers, c) offer AI functions as microservices, and, d) can form service chains, denoted as AI pipelines. Collaborative engineering is facilitated by the trading and (re-)using artifacts and, thus, accelerating the AI application design.

The security architecture of the SVP is built around the security needs of collaborative AI engineering and uses a proxy concept for microservices. The proxy shields the AI artifact and pipelines from outside adversaries as well as from misbehaving users, thus building trust among the collaborating parties. We identify the security needs of collaborative AI engineering, derive the security challenges, outline the SVP’s architecture, and describe its security capabilities and its implementation, which is currently in use with several AI developer communities. Furthermore, we evaluate the SVP’s Technology Readiness Level (TRL) with regard to collaborative AI engineering and data security.

The field of artificial intelligence (AI) has yet to fully capitalise on the potential to develop AI systems through collaboration between system developers and data scientists, especially when they belong to different organisations. We studied how using a marketplace creates value in such value chains by allowing organisations to advertise and share AI assets like data and models and enable multi-party development of AI systems while protecting these assets. The paper describes an embedded multi-case study of a marketplace under development, the Bonseyes marketplace. The cases were a collaboration between universities, a large company outsourcing data science, and a small business offering AI models as products. Value creation was linked to reduced development time, streamlined product communication, and protection of shared assets. However, participants required data protection concerns to be addressed, as well as marketplace maturity. The marketplace created value with clear goals for developing an AI system and supporting tools, templates, and examples. Our study benefits researchers wanting to advance AI systems development by offering rich examples of how a marketplace can enable multi-party collaborations.