Design of a LoRa transmission-loss test system for near-ground environments

Traditional transmission loss models for LoRa wireless communication systems operating in near-ground environments exhibit limited applicability. Another issue is the difficulty of effectively supporting the design of near-ground wireless transmission networks. Furthermore, most of the currently available test methods for LoRa communication systems are costly and exhibit limited flexibility. Based on the above, we propose a new wireless communication system for testing the transmission quality of LoRa wireless communication systems operating in near-ground environments.

The proposed test system consists of a communication and an analysis part. The communication part includes a chip integrating a wireless microcontroller and wireless communication functions, as well as a Micro SD storage unit. The communication functions and the storage unit are used for LoRa signal communication, as well as the storage of the communication signals and their parameters (i.e., signal-to-noise ratio and signal strength); the microcontroller is used to implement visualization data functions and a key menu function for adjusting the communication parameters at any time during testing. The analysis is performed using the upper computer software, which is designed to read the communication quality parameters of the LoRa signals stored in the memory card. The strength of the transmitted signal is determined according to the traditional, empirical, and improved empirical transmission loss models. Finally, the host computer outputs the results generated by the three models. These results are the values of the parameters to be fitted for different models, the sum of residuals, and the correlation coefficient.

The parameter fitting values are the environmental impact constant and the attenuation coefficient; these are used to represent the signal transmission characteristics in certain environments. The sum of residuals and the correlation coefficient represent the degree the fitted model represents the actual signal. When the sum of residuals is small and the correlation coefficient is large, the degree the model fits the actual signal increases. During testing, the strength values of the LoRa modulated signals near the ground (i.e., 1 m above the ground) are collected assuming three ground scenarios: snow-covered, farmland, and cement-covered ground; then, the signals are processed and analyzed using the proposed computer software. The value of the environmental impact constant obtained from the software using the empirical model with a one-step fitting was negative. This erroneous result indicates overfitting of the empirical model. Therefore, we employed an improved empirical model. In this model, the data are gradually fitted, starting from three pieces and ending with all nine pieces of data. Then, all fitting results are summed, and their average values are calculated. These values replace the corresponding parameter values in the empirical model to obtain a rational model. The rationality, validity, and convergence of the improved empirical model were verified using the proposed computer software.

The experimental results showed that the improved empirical model constructed using multistep fitting can reasonably and accurately describe the transmission loss characteristics of LoRa modulated signals in near-ground environments, providing a useful reference for the construction of future LoRa near-ground wireless transmission networks.