Lightweight Behavioral Trust Validation for Collaborative Edge-Device Contributions

Abstract

Distributed devices that continuously submit measurements, alerts, predictions, and compressed analytical summaries characterize edge collaborative environments. Even when it authenticates the device, it cannot guarantee integrity of further contribution. BeTrust-Edge is a lightweight behavioral trust framework for contribution-level validation across heterogeneous edge devices. The structure separates the trustworthiness of historical devices and trustworthiness of current contributions, meaning reputation cannot ovelap with anomaly submissions. Normalization that is aware of capabilities accounts for differences in latency, reporting frequency, and resource behavior of constrained and higher capability nodes. A dual-memory estimator draws on rapid evidence and stable long-term behaviour, while trust velocity captures progressive decay. The divergence of trust and credibility further identifies the contradiction between established reliability and the quality of current contribution. Following that, each donation is accepted, revised, observed, or denied through an interpretable four-level policy. The framework was built in Python and analyzed using behaviororiented event traces generated from CICIoT2023 traffic and then subjected to noisy, delayed replay gradual-drift manipulative on-off scenarios. When contrasted to fixed weighted trust, exponential reputation, as well as fixed threshold validation, the overall validation accuracy is 93.35%. The method reduces bad effects and has a low decision latency and produces constrained devices genuine contributions. The findings effectively validate behavior without deployment of deep learning, distributed ledger, custom hardware or computeintensive security infrastructure and under heterogeneous and dynamic edge operating conditions.