Design and Implementation of an Edge-Computing-Based Faulty Gateway Recovery Scheme for ZigBee Networks

Abstract

This paper focuses on the network-interruption problem caused by single-point-of-failure of the gateway in ZigBee smart-home networks and proposes an edge-computing-based self-healing scheme for faulty gateways. By establishing a three-level device hierarchy-Class-0 (core gateway), Class-1 (backup gateway) and Class-2 (terminal devices)-and exploiting Class-1 devices equipped with concurrent ZigBee + Wi-Fi radios as distributed redundant nodes, the scheme introduces an edge-side multi-dimensional device evaluation model (M-score). A machine-learning algorithm fuses signal quality, network-topology position and device stability to intelligently select the optimal backup node. When the master gateway fails, the system triggers a fast-reaction mechanism driven by routing-packet density (F-score); the top-ranked Class-1 device automatically performs Wi-Fi association, downloads its configuration from the cloud and rebuilds the ZigBee network, accomplishing gateway replacement without human intervention. Experimental results show that the scheme reduces the Mean Time To Repair (MTTR) from over an hour in traditional solutions to under 15 minutes with a $\mathbf{1 0 0} {\%}$ success rate, markedly improving smart-home network reliability and user experience. These gains stem from optimizing the key maintainability metric MTTR and from the effective implementation of the self-healing mechanism, all achieved without extra hardware cost and deployable at scale through firmware upgrades alone.