A Moral Hazard Detection Framework: Reinforcing Trust in ORAN

With the emergence of the Open Radio Access Network (ORAN) concept and related standardization efforts, future radio access networks are anticipated to feature elements from diverse vendors. Although the ORAN elements can authenticate as legitimate, the system may fail to meet service requirements if some network components do not adhere to their respective agreements, i.e., moral hazard. This issue raises concerns about the network's end-to-end performance, complicating fault attribution and conflict resolution. Therefore, there is a need for an automated zero-trust framework capable of continuously detecting instances of moral hazard. The complexity is exacerbated by the dynamic nature of network elements or artificial intelligence (AI) model performance, which may degrade over time intentionally (e.g., malicious tampering) or unintentionally (e.g., model obsolescence or device performance decline), limiting the effectiveness of offline testing. To address this, we develop a mechanism based on subjective logic principles, incorporating a logic-based argumentation framework that explicitly accommodates argument schemes, argument accrual, and burden of proof. Building upon this framework, we apply contract theory to incentivize compliant devices to participate truthfully in the ORAN ecosystem, thereby enhancing system performance. The simulation results show improved system efficiency and reduced operational costs.