TACIT: Trustworthy Arbitration Through Calibrated Inference Tracking for rApp Conflict Resolution in O-RAN

Open Radio Access Network (O-RAN) deployments face a coordination challenge when multiple radio applications (rApps) concurrently propose conflicting parameter changes.We present TACIT, a decentralized conflict resolution mechanism that advances the state of the art in three ways: (1) it resolves direct, indirect, and implicit conflicts without requiring explicit detection or classification, (2) it achieves O(N) computational complexity in the number of rApps, independent of parameter space size, and (3) it employs an asymmetric reputation mechanism that isolates unreliable rApps through continuous online learning. TACIT works by soliciting each rApp’s assessment of a proposed change’s impact on its own KPIs, then aggregating these self-interested inputs using reputation-weighted Nash Social Welfare optimization. No rApp needs to understand other rApps’ objectives, and no centralized function needs to replicate this understanding. Evaluation through ns-3 simulation with 19 cells and 7 concurrent rApps demonstrates that TACIT outperforms unconstrained operation, static priority ordering, and unweighted Nash Social Welfare across throughput, energy efficiency, and edge user metrics. When an rApp’s predictions degrade, the reputation mechanism isolates its influence within a single configuration window without operator intervention.