Power- and Memory-Efficient Low-Cost IoT Sensing System for Biodiversity Surveillance Developed on Resource-Constrained Platform

Imperative demand is arising for biodiversity surveillance due to the decline in certain insect species. Existing biodiversity surveillance systems depend on high-end processors and heavy operating systems (OSs) that require numerous packages, resulting in increasing complexity and power consumption. In contrast, resource-constrained Internet of Things (IoT) platforms are cheap and require minimal code footprints for operation, resulting in reduced complexity. However, due to the resource scarcity of such platforms, it is required to integrate efficient management strategies for handling a large volume of data generated by the surveillance system. To address this, a low-cost sensing system on a resource-constrained platform is developed that offers power-efficient biodiversity monitoring through a proposed task-scheduling architecture. The system employs the message queuing telemetry transport (MQTT) protocol for the publish-subscribe paradigm. However, the existing JavaScript object notation (JSON)-based MQTT payload format limits the payload size for transmission and increases random access memory (RAM) utilization. To overcome this issue, a raw binary MQTT transmission pipeline is proposed which maximizes the payload size while reducing RAM usage and transmission time, leading to lower energy consumption compared to JSON-based transmission. Furthermore, the proposed over-the-air (OTA) update framework enables remote updates to system configurations. Graphical and tabular comparisons of current consumption, battery lifetime, and cost demonstrate that the designed sensing system outperforms the existing DarkNet53, PICT, and Moth traps systems. Moreover, the presented cloud pipeline architecture provides insect species inferences with an approximate precision of 97%.