A fuzzy reputation system for Radio Access Network sharing

5G network slicing allows the coexistence of multiple virtualized networks on the same infrastructure. Leveraging this concept, it becomes possible to create marketplaces where Infrastructure Providers (InPs) lease network resources to different Mobile Virtual Network Operators (MVNOs) while accommodating their specific Quality of Service (QoS) requirements. In addition to the cost criteria, an MVNO choosing between several InP might be interested in evaluating the InPs actual capacity to deliver the expected Service Level Agreement (SLA). Existing 5G literature on trust evaluation is often based on blockchain. While this technology offers transparency and auditability, the inherent consensus mechanism often brings additional costs. In this paper, we propose a distributed reputation system based on fuzzy logic that can provide a robust and dynamic estimation of an InP behavior while respecting the subjective requirements of MVNOs. We evaluate this system in a Radio Access Network (RAN) sharing simulation and we show that it can redirect MVNO to InP that are capable of meeting their needs. We further test different trust decay strategy in order to find one that is able to both quickly react to a network outage and forgive InP once the punctual outage is solved.