Configuration and execution of a scalable IoT architecture for smart home automation

This paper presents the design, implementation, and evaluation of a scalable, cloud-connected smart refrigerator system that leverages Narrowband IoT (NB-IoT), Message Queuing Telemetry Transport (MQTT), and serverless cloud functions to enable intelligent control, real-time monitoring, and voice interaction. Addressing the growing demand for low-power, always-connected smart appliances, the proposed architecture combines lightweight communication protocols with a serverless backend hosted on Amazon Web Services (AWS), integrating Lambda functions, IoT Core, and DynamoDB for flexible, event-driven operation. The system supports both mobile and voice-based interfaces through MQTT-enabled applications and a custom Amazon Alexa skill, allowing users to interact seamlessly with the refrigerator. Security is ensured via a cryptographic authentication chip, and over-the-air firmware updates are supported using Amazon S3. A key contribution of this work is the use of NB-IoT for reliable long-range messaging in indoor environments, evaluated through experimental tests that analyze latency, CPU load, message loss, and acknowledgment behavior. Quantitative analysis confirms that the architecture performs reliably under frequent message exchanges, achieving median latency below one second in the majority of cases, while exhibiting acceptable performance variability under network stress. These results demonstrate that while latency occasionally fluctuates due to NB-IoT scheduling, overall responsiveness remains suitable for near-real-time smart appliance control. The system demonstrates robustness, energy efficiency, and scalability, making it an effective solution for next-generation smart refrigerator appliances. Unlike prior implementations that simply combine NB-IoT, MQTT, and serverless functions, this work implements an application-layer adaptation of MQTT QoS-1 acknowledgment handling tailored to NB-IoT latency characteristics, evaluated through empirical experiments. for the latency characteristics of NB-IoT networks. It further presents a empirical analysis of latency–CPU correlation based on experimental measurements to evaluate performance under real-world network stress, demonstrating measurable gains in reliability and responsiveness. These contributions establish a reproducible foundation for energy-efficient, low-latency IoT deployments beyond traditional smart home integrations.