A Method for Link Quality Estimation in LoRa Network based on Support Vector Machine

Van Dai Pham, Hao Do Phuc, Tran Duc Le, Ruslan Kirichek
In this paper, we present the use of machine learning techniques to estimate the link quality of a fragment of the LoRa network. We performed experiments with a testbed based on Rostelecom-ITU-based Laboratory using several LoRa nodes (CubeCell and TTGO LoRa32). The data such as received signal strength indicator (RSSI), signal-to-noise ratio (SNR) of received packets, and packet reception rate (PRR) were collected and trained to classify the link quality levels based on support vector machine (SVM) model with Gaussian kernel. The results show the high accuracy (mean = 95\%) while using 10\% of the dataset for training.