Ionogram analysis: a neural network approach

Knowledge of the state of the ionospheric radio channels is of great importance for both ionospere research and radio wave propagation predictions. Diagnostic of ionospheric radio channels is carried out by the analysis of ionograms. An efficient method of ionogram processing is proposed. It uses an artificial neural network (ANN) with the mean field theory updating scheme. Because of a complex character of ionospheric traces with quite a heavy background, a modified rotor model of Hopfield network is used. To speed up the convergence of the ANN evolution, a special initial ANN configuration is constructed in a vicinity of the global minimum of the ANN energy function. It is done by applying a special angular histograming within a sliding window, whose size is determined by the average local track curvature. Our model was tested on ionograms obtained on the chirp- ionosonde (ISTP, Irkutsk). Result analysis shows the efficiency of our approach and its prospects for the solution of the ionograms processing problems.© (1999) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.