Titolo della tesi: A Requirement-Driven Methodology for Compatibility Assessment in Storage Technology Selection for Agri-Food Traceability
The increasing demand for transparency, food safety, product authenticity, and regulatory compliance has fostered the adoption of traceability systems in the agri-food sector. Several storage technologies can support these systems, including conventional databases (relational and NoSQL) and DLT-based platforms such as permissionless and permissioned blockchains and DAG-based architectures. Although they support similar functional objectives, they differ substantially in non-functional aspects such as performance and governance, making storage technology selection a non-trivial task. Existing approaches can be broadly divided into rule-based approaches, which recommend technologies through predefined decision rules derived from architectural characteristics, and Multi-Criteria Decision-Making (MCDM) approaches, which typically assume a common evaluation model that is difficult to establish when heterogeneous technologies are involved.
This thesis proposes a requirement-driven methodology for storage technology selection in agri-food traceability systems. The methodology defines a common evaluation model in which technology attribute profiles and scenario requirements are represented through the same set of non-functional attributes. Technologies are represented through Technology Profiles, while scenario requirements are formalized through a Scenario Non-Functional Requirements Record. Compatibility is assessed attribute by attribute through dedicated compatibility functions and aggregated into a Compatibility Score that ranks candidate technologies according to the degree to which their attribute profiles cover scenario-specific non-functional requirements.
The methodology is instantiated and evaluated using Technology Profiles for conventional databases and DLT-based platforms. Validation relies on ten synthetic scenarios and five agri-food traceability case studies covering different governance models, operational requirements, and trust assumptions. In this setting, the requirement-driven compatibility assessment complements both rule-based classification and MCDM approaches: rather than optimizing over predefined criteria within a fixed solution space, it evaluates the compatibility of each candidate technology with the non-functional requirements of a given scenario.
The experiments produce differentiated Compatibility Score rankings across the evaluated scenarios. Rankings vary from one scenario to another, and no single technology dominates all configurations, although specific technological families emerge more frequently as top-ranked under particular requirement profiles. Across all synthetic scenarios and agri-food case studies, different configurations of technology attribute profiles and scenario-specific non-functional requirements lead to different Compatibility Score rankings, showing that the methodology can discriminate between heterogeneous technologies under varying requirement profiles. The profile-based representation also allows new technologies to be incorporated and existing ones to be updated by adding or revising Technology Profiles without modifying the compatibility functions or the overall evaluation process.