Differential Semi–Blind Spatial-Temporal Beamforming for DS–WCDMA Systems in Frequency Selective Time -Varying Channels

In this paper , we propose a new differential semi-blind spatial-temporal beamforming receiver for asynchronous direct sequence wideband code division multiple access (DS-WCDMA) systems. The new receiver presents improved robustness to phase var iations and reduction of the effects of nonlinear distor tion. I . INTRODUCTION Although the main problem arising from high data rate DSWCDMA systems is the severe inter-symbol interference (ISI), in addition to the multiple access interference (MAI), the presence of a time-varying fading channel, even with a relative small Doppler spread (compared to the symbol duration), may also reduce overall system performance. In this situation, it may be convenient to employ differential encoding/decoding schemes. Using this simple approach, phase fluctuations in the received signal do not affect system performance as long as they can be considered negligible within a symbol interval. We note, however, that this robustness to channel variability is achieved at the expense of a higher signal-to-noise ratio (SNR). As we address the problem of differential detection, the method of data modulation considered is the π/4 shifted DQPSK, which eliminates the needs for accurate carrier recovery and greatly simplifies the implementation of the receiver. The choice is motivated by the efficiency of the method with respect to traditional QPSK or OQPSK, which require coherent detection [1]. Another advantage is that transitions in the signal constellation do not pass through the origin that reduces the effects of nonlinear distortion caused by the R.F. amplifier. This is obtained by the superposition of two QPSK signal constellations offset by 45 degrees relative to each other, resulting eight phases. The phases are alternatively selected from one of the two QPSK constellations [2]. In [3], a semi-blind spatial-temporal beamforming receiver (BST-SBCMACI) based on the SBCMACI (semi-blind constant modulus algorithm with channel identification) [4] was initially presented. The algorithm uses the constant modulus property of the transmitted signal and performs semi-blind subspace channel identification as a precursor to semi-blind equalization. The resulting receiver allows for coherent combination of the desired signal multipath, cancellation of the interfering users, removal of phase ambiguities present in blind algorithms and significant reduction of the required number of training symbols. In spite of all the advantages of this receiver, stationary conditions frame by frame are necessary. In order to relieve this requirement, we propose in this paper a new differential BST-SBCMACI receiver (BST-SBCMACI-DIFF) that presents improved robustness to phase variations and reduction of the effects of nonlinear distortion by employing π/4 shifted DQPSK data modulation and differential detection [2]. The paper is organized as follows: The system model is presented in section II; LS optimization and semi-blind subspace channel identification in section III; SBCMACI in section IV; simulation results in section V and finally, concluding remarks in section VI. I I . SYSTEM MODEL We consider the reverse link of an asynchronous DS-CDMA system employing complex spreading [5] and π/4-DQPSK data modulation to reduce peak-average ratio and achieve better robustness to phase variations. There are M users in the system and each user may transmit Nb data symbols per packet over assumed stationary conditions. It is also assumed that the receiver employs an antenna array consisting of A identical elements equally-spaced by 2 ant λ , where ant λ is the carrier frequency wavelength. Assuming that the inverse signal bandwidth is large compared to the propagation time across the array, the complex envelopes of the signals received by different antenna elements from a given path are identical except for phase and amplitude differences that depend on the path angle of arrival (AOA), array geometry and the element pattern [6]. The AOA of the lth multipath signal from the mth user is l m θ and ( ) l m θ a is the array response vector (spatial signature vector) to the multipath signal arriving from the direction l m θ , with ( ) ( ) ( ) [ ] l m A l m l m a a θ θ θ , , 1 = a . We can represent the reverse link baseband complex signal in the following vector form: