Asian Journal of Advances in Research

The convergence of the Internet of Things and fog computing has opened substantive new avenues for transforming how refrigeration systems are monitored, diagnosed, and maintained. Traditional reactive and scheduled maintenance approaches impose considerable operational costs, energy penalties, and risks of unplanned failures across refrigeration infrastructure that spans domestic appliances, commercial installations, pharmaceutical cold storage, and large-scale food supply chains. This article critically reviews the state of knowledge concerning IoT-enabled smart refrigeration maintenance facilitated through fog computing frameworks, examining the architectural foundations, sensor integration strategies, communication protocols, data processing pipelines, predictive maintenance algorithms, energy optimisation potential, and security considerations relevant to this rapidly developing field. A narrative review methodology was employed, drawing on peer-reviewed literature published primarily from 2013 onwards across multiple scientific and engineering databases. The review finds that fog computing architectures offer compelling advantages over purely cloud-dependent systems with respect to latency reduction, bandwidth conservation, and real-time responsiveness — qualities that matter a great deal in refrigeration contexts, where a temperature excursion, a compressor anomaly, or a refrigerant leak can escalate within hours into a food safety incident or an equipment failure. Machine learning approaches, particularly anomaly detection algorithms and predictive models deployed at fog nodes, have produced encouraging results in identifying incipient faults before they become disruptive. Even so, persistent challenges remain in achieving interoperability across heterogeneous device ecosystems, ensuring robust cybersecurity, and scaling frameworks economically across diverse deployment contexts. The review concludes by identifying priority research directions, including digital twin integration, federated learning, and standardised open architectures, while acknowledging key gaps in empirical validation and cross-sector implementation studies.