SMART METER DATA COMPRESSION: A SOLUTION FOR REDUCING AND OPTIMISING ENERGY METER RESOURCES IN NBIOT NETWORKS

This study proposes a strategy to optimize data compression in embedded devices based on the STM32WBA microcontroller. The main objective was to reduce the volume of data transmitted in NB-IoT networks, with the aim of reducing the volume of this data without compromising the integrity of the information, while at the same time while respecting the memory and computing capacity constraints of these systems. The research used the HEATSHRINK library, based on the LZSS algorithm, adjusting parameters such as the size of the compression window (window_size) and the size of the lookahead (lookahead_sz2) integrated with the BZ2 algorithm based on the Burrows-Wheeler Transform (BWT) technique, to achieve a balance between compression rate, processing time and memory consumption. The experiments revealed that intermediate configurations, such as HSWB06_HSLB04 and HSWB06_HSLB05, achieved compression rates of up to 46.93% for sets of 96 samples. Combined with a 21.15% reduction in the DLMS header, these configurations resulted in a total compression gain of 68.08%. These parameters also had moderate processing times, in the range of 525 ms, and a memory consumption of 200 bytes, making them suitable for devices with limited resources. On the other hand, more extreme configurations, such as HSWB10_HSLB09, showed inferior performance, with compression rates below 40% and processing times of over 7,900 ms, making them unfeasible for systems that require low latency. A strategic approach to allocating microcontroller resources was also implemented. Original data was stored in FLASH memory, compressed data in RAM and temporary buffers were managed by STACK, ensuring operational stability even under high load conditions. The viability of these configurations was confirmed through memory analyses carried out with the Build Analyzer tool integrated into STM32CubeIDE, which showed minimal impact on the device's operations. The results obtained highlight the potential of the HEATSHRINK library as an efficient solution for data compression in embedded devices, provided it is properly adjusted to the specifics of the application. In addition, the study opens up avenues for future advances, such as the integration of hybrid compression techniques with machine learning, validations in real environments and adaptations for intermittent transmission systems, such as NB-IoT and LoRaWAN networks. These improvements have the potential to significantly expand compression applications in diverse sectors, including energy, health and transport, promoting greater energy efficiency, robustness and reliability.