Abstract:

A method to predict the interference effect on unmanned aerial vehicle (UAV) data link in complex environment containing terrain, weather and electromagnetic conditions using the support vector machine (SVM) and the power rule is proposed. First, bit error rate (BER) of UAV data link and the propagation model are simulated, then obtain the simulation data in terms of the two single-tone interferences case， rate simulation results into five different degrees， and get importing learning samples. Compare the learning accuracy of two types of input parameters including feature normalization and feature dimension reduction and four different kernels by grid search and cross validation to obtain the best prediction model. Finally, the feasibility and validity of this method is demonstrated by comparing the theoretical calculation and the prediction results of an actual scenario by using this model. The prediction results can be also used in path planning and path analyzing of UAV, so this method has high practicability.