| Abstract |
TETRA communication system, a flexible system, effectively and optimally transfers voice and data; however, path loss in the tetra band was affected by landscape contours, environmental factors, and propagation medium due to an inadequate distance between the two ends of the cable and the height and position of the transmitter and receiver. Hence, a novel, Characterization of Path Loss in the Tetra Band for Smart City Project is developed. It uses a Sparse ConvAutoNet-driven Regression Random Machines approach to improve antenna quality by mitigating signals coming from certain low elevation directions and moving the antenna away from reflection objects. Traditional techniques have shown that higher frequencies do not provide a longer range, affecting the geographic coverage. Furthermore, previous methods suffered from slow cellular network and data transfer speeds due to classification and clustering issues in pattern recognition. To overcome these limitations, Support Vector Regression (SVR) and Regression Random Machines (RRMs) are employed, utilising a free kernel choice function to enhance the stability of the cellular network and data transfer, eliminating classification and clustering issues. As a result, the suggested model results in a lower RMSE of 1.05, lower MSE of 0.9, higher R-squared of 0.97, lower MAPE of 0.037. © 2025 Engineers Australia. |
