| Abstract [eng] |
Smart electricity meters in low-voltage networks can be used not only to automate energy metering, but also to monitor the state of the network, for example, to detect anomalies, assess power quality, or perform diagnostics. However, in many systems, meter data are currently transmitted only every 10–15 minutes, while more detailed network analysis usually requires more frequent measurements. Increasing the measurement frequency also increases the amount of transmitted data, making data transmission one of the main limitations of such a system. This work analyzes how a distributed IoT communication system can be used to transmit one-second- resolution data from smart meters. The aim of the work is to propose methods for collecting electricity network monitoring data and to evaluate how such a system performs in different data collection scenarios. To achieve this aim, application requirements and transmission protocols are analyzed, a research methodology is developed, a system prototype is implemented, and experimental tests are conducted. The developed system consists of an ESP32-C6 controller that reads DSMR/P1 data, Wi- Fi communication infrastructure, and a data collection server with a database. The experiments evaluate application-layer protocols, data packetization strategies, transmission reliability under packet-loss conditions, an event-based transmission model, and changes in system load as the number of controllers increases. The study found that the HTTP protocol is not suitable for continuous telemetry data transmission due to its high protocol overhead. CoAP places less load on the controller, but provides lower reliability under unstable network conditions. MQTT provides the best compromise between transmission reliability and resource usage. Data aggregation reduces network load, but when more than 10 measurements are aggregated, memory fragmentation increases noticeably. Event-based transmission reduces the amount of transmitted data, but due to its limited accuracy, it cannot completely replace periodic monitoring. It was observed that as the number of controllers increases, data traffic increases linearly, and on the server side, the main limiting factor is CPU load rather than memory usage. Based on the obtained results, it is recommended to use the MQTT protocol, batch 5–10 measurements, select the QoS level based on the importance of the data, and apply event-based logic as an additional mechanism for reducing the amount of transmitted data. The data collection solution described in this work was used in the preparation of two articles published in international scientific journals and one article submitted to an international scientific journal, which is currently under review [1, 2]. |
