Aircraft Type Recognition Based On Shortwave Communication

In shortwave communication of military and other use, the quality of acoustic signals received by radio shortwave about aircrafts are complex and very poor, especially according to background sound signals aircrafts are sometimes difficult to be identified. At present the main way to identify the acoustic signals information is artificial. There is an important military significance about aircraft type recognition based on shortwave radio communication.At present no open relevant literature is reported about aircraft type recognition of shortwave radio communication based on air-ground short voice acoustic signals. Wavelet packet decomposition(WPD) is used to extract acoustic signal characteristics of aircrafts. Wavelet packet transform proposes a better performance at time-frequency analysis performance of the signals. After wavelet packet decomposition different frequency bands of acoustic signals has the different energy characteristics of signals. The effective aircraft characteristic vectors are extracted.According to the characteristics of aircraft acoustic signal, support vector machine (SVM) and back propagation (BP) artificial neural network (ANN) as classifiers are adopted to identify five kinds of aircrafts respectively. Multiclass pattern classification algorithm based on SVM is more efficient recognition technique than BP ANN to solve problems of small samples. SVM has the excellent statistical learning ability. Its computation cost is small and operating speed is quick. So it can satisfy aircraft type recognition in real time. Compared with BP ANN, SVM exhibits more excellent performances such as no local optimum problem, no over-fit or under-fit problem, better convergence property, less training samples, higher correct recognition rate and higher reliability.In simulation experiments we present two aircrafts recognition algorithms: WPD-BP ANN and WPD-SVM. These algorithms have been tested and compared with recognition rate. The results show that WPD-SVM achieves the better performance than WPD-BP ANN under tested environmental conditions. It is capable to extract the effectual feature of aircraft cabin noise of the acoustic signals blending serious other noise and it is effective to recognize five kinds of aircrafts.