CoMeT-Net: Consensus Memory Template Network for Real-time Traffic Anomaly Detection

Abstract

Real-time anomaly detection in Open Radio Access Networks (O-RAN) demands high accuracy, low false alarms, and computational efficiency for resource-constrained edge deployment. Traditional methods struggle with computational overhead, inconsistent cross-domain performance, and suboptimal feature representations that miss subtle attacks on O-RAN's open interfaces. We present CoMeT-Net (Consensus Memory Template Network), a framework achieving state-of-the-art detection through three innovations: (1) structured memory banks enabling template-based consensus voting with $O(N \cdot C)$ complexity; (2) adaptive gating that downweights ambiguous features as a learned noise filter; (3) contrastive alignment unifying feature learning and classification. Deployed in O-RAN infrastructure via edge servers and Near-RT RIC xApp, CoMeT-Net enables dynamic threat mitigation through PRB throttling and RRC connection release. On network traffic datasets, CoMeT-Net achieves 99.35% F1 score with 10$\times$ lower false alarm rates than baselines while maintaining 0.3-3ms inference across hardware tiers from servers to Raspberry Pi 4. O-RAN testbed validation demonstrates effective isolation, degrading attacker latency to >1400ms while preserving 15-20ms for legitimate users.