Performance Evaluation of a Collaborative IoT Framework for Energy-Efficient Communities.

Smart communities of consumers have the chance to cooperate and reach coordinated behaviour within the Internet-of-Things (IoT) towards the achievement of common goals. A pilot testbed of a collaborative IoT-based framework for energy-efficient and sustainable communities is deployed and evaluated. The framework is centralised on a community aggregator, which implements a fair scheduler in response to both, the consumers' demand time preference and the utility's day-ahead supply from renewables. In this paper, we present the architecture design and analyse the performance cost as to validate the feasibility of the proposed collaborative architecture. Home Controllers and the community Aggregator are implemented on Raspberry Pi 4 boards throwing feasible performance results: overheads at the Aggregator lie within suitable ranges (19 seconds for 52 appliances) even in scenarios of rigid consumers' demand. Scenarios are configured from Living Labs' datasets containing appliances' consumption and operation time frames. Communication overhead imposed by the tested network protocols (sockets-UDP, MQTT, MQTT/TLS, and CoAP/DTLS), and links (WiFi, ZigBee and Z-Wave) result in CoAP/DTLS behaving more efficiently (90 ms) than MQTT/TLS (160 ms).