Improved Demodulation of Phase Shift Keyed Signals Using Wavelet Thresholding

In this study the use of wavelet transform-based techniques to de-noise phase shift keyed (PSK) signals is investigated. Specifically, binary phase shift keyed (BPSK), quadrature phase shift keyed (QPSK) and 8-PSK signals are considered in the presence of additive white Gaussian noise (AWGN). Wavelet-domain decomposition of the received signals is carried out via the discrete wavelet transform (DWT). The Haar, Daubechies5, Coiflet5 and Symmlet4 wavelets are used, along with both hard- and soft-thresholding techniques. The improvements realized by matched filter-based demodulation in conjunction with the use of wavelet thresholding techniques are shown by bit error rate (BER) curves for each of the three modulation schemes. It has been found that by using appropriate wavelets and wavelet thresholding methods improvements in BER of over three orders of magnitude are achievable within the range of bit energy-to-noise spectral density ratios (E_b/N₀) from -5 dB to 8 dB.