Exact SER analysis of ZF-DFE for M-QAM on multipath fast fading channels

Abstract

Zero-forcing decision feedback equalization (ZF-DFE) has been widely investigated in an effort to alleviate intersymbol interference (ISI) problem caused by delay spread on both wired and wireless communication channels. Among nonlinear equalizer, the ZF-DFE is most common because it is fairly simple to implement and performs well. However, on channels with low signal-to-noise ratio (SNR), the ZF-DFE have error propagation problem by decision error. Furthermore, the analysis of the error propagation in ZF-DFE is difficult by the fact that little is known about the distribution of the errors. One of the previous works uses Markov model to analyze ZF-DFE performance with error propagation for pulse amplitude modulation (PAM) signals. Another approach uses an approximate Fourier series to evaluate the effect of residual ISI from future symbols, and derive approximate BER expressions for quadrature phase-shift keying (QPSK) on static Rayleigh fading channels. Even though theses approaches provide upper bound considering the effect of error propagation, they are restricted only to PAM or QPSK signals. Also, upper bound of ZF-DFE performance over M-ary quadrature amplitude modulation (M-QAM) is derived by using minimum mean square error (MMSE) concept on static Rayleigh fading channels. However, the effect of error propagation is not considered by assuming correct decision of previous symbol. Most of previous works consider static multipath channels. However, mobile radio communication channels are time-varying multipath fading channels. Therefore, practical analysis on wireless radio channel characterized by fast time-variations is necessary. In this thesis, we analyze the symbol error rate (SER) of ZF-DFE on wireless radio channels. We bring the concept of “error event” and “state” related to error propagation, which is introduced for zero-forcing successive interference cancellation (ZF-SIC) detector. However, we need a new approach in order to find exact...