PAT and Px Code Sidelobe Reduction Using Wavelet Neural Network

Abstract

Pulse compression is a significant aspect for improving the radar detection and range resolution. To improve the range detection, the pulse width is increased to overcome the transmitter maximum peak power limitations. However, pulse compression is accompanied with time sidelobes that can mask the small targets. The Wavelet Neural Network (WNN) is a new technique used for pulse compression sidelobe reduction. In this paper, Morlet function is applied as an activation function for WNN and the backpropagation (BP) is implemented for training the networks. The WNN is applied based on PAT and Px polyphase codes. The performance of WNN is evaluated in terms of Signal to Noise Ratio (SNR) and the computational complexity. The simulation results indicate that the WNN has higher Peak Sidelobe Level (PSL) than the Auto Correlation Function (ACF) with more than 100 dB and higher PSL than the Neural Network (NN) with more than 100 dB.

Classification of Prostate Cancer using Wavelet Neural Network

“…Where ℎ and denotes the neurons number in the hidden and the input layer respectively. Morlet function is deployed in the hidden layer as an activation function, and sigmoid function is deployed in output layer [11]. Equation 17 below indicates the Morlet function, as displayed in Figure 2.…”

“…Their results indicated that WNN has a powerful approximation capability and suitability in prediction and modeling. Therefore, WNN is a significant tool in classification and signals processing [11]- [12]. In this article, a new approach for classification of prostate cancer's risk is proposed using WNN.…”