Link Quality Analysis of Wireless Sensor Networks Based on CART Decision Tree

Aiming at the problem of link quality reliability in wireless sensor networks, a method for evaluating link quality based on the CART decision tree algorithm is proposed. Firstly, the link quality datasets is divided into three levels according to the link quality evaluation metric, and all kinds of data are balanced by random oversampling. Then, the received signal strength indication is used to generate the synthetic feature as the feature inputs. Finally, the evaluation model is constructed based on the CART decision tree algorithm. Experiments show that the number of decision tree layers affects the accuracy of the overall model. By analyzing the influence of the number of tree layers on the accuracy of link quality evaluation, it is determined that when the number of tree layers is 6, the accuracy of link quality evaluation reaches the highest 94.8 %, and the accuracy, precision, recall and F1 value are used to verify the reliability of the conclusion.