A study on computationally efficient receiver design for MIMO and OFDM systems

Abstract

It is generally known that multiple-input and multiple-output (MIMO) and orthogonal frequency division multiplexing (OFDM) systems provide enormous advantages such as outstandingly high spectral efficiencies in rich scattering environments, and immunity to frequency-selective fading channel and the capability of high spectral efficiency, respectively. Due to these advantages, MIMO and OFDM systems have attracted a lot of attention as promising candidate schemes for next-generation wireless systems. For both two systems, efficient signal detector at receiver is needed because the accuracy and complexity of the detector determine the system performance and receiver efforts, respectively. For MIMO systems, especially double space-time transmit diversity (D-STTD) which achieves both diversity gain and multiplexing gain, many kinds of detectors, such as interference cancellation and maximum-likelihood (ICML), maximum-likelihood (ML) detector, and sphere detector (SD), are proposed. SD provides exactly same performance with that of optimal detector (ML) with tractable complexity, however, further reduction of the detection complexity is still crucial issue considering size and power limitation of the mobile station. For OFDM systems over time-frequency selective channel, many kinds of equalizers were introduced to reduce the effect of ICI due to mobility. The representative signal detectors are serial/block zero-forcing (ZF) and minimum mean squared error (MMSE) approach equalizer, banded structure based equalizer, and parallel interference cancellation (PIC) approach equalization. Due to the large size of the OFDM symbol block, equalization techniques generally provides high operational complexity maintaining good performance. Moreover, some equalizers achieve the computational efficiency at the cost of the degradation of the performance. In this thesis, we propose computationally efficient receiver design for MIMO and OFDM systems that are well suited to bot...