Interference suppression capability of faster than symbol rate sampling and frequency domain oversampling


Thesis Type: Doctorate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Electrical and Electronics Engineering, Turkey

Approval Date: 2016

Student: EREN BALEVİ

Supervisor: ALİ ÖZGÜR YILMAZ

Abstract:

Detection of symbols in the presence of many interference sources is a difficult task in wireless channels. It is obligatory to reduce the interference for reliable communication. In this dissertation, minimum mean square error (MMSE) detection is investigated to suppress interference. Faster Than Symbol Rate (FTSR) sampling and Frequency Domain Oversampling (FDO) methods are proposed to enhance the interference suppression level of MMSE detection for both single user and multiuser communication. The aim of single user communication is to mitigate Intersymbol Interference (ISI) by MMSE equalization with FTSR sampling or FDO, whereas ISI and Multiple Access Interference (MAI) are reduced jointly in multiuser communication by MMSE detector with FTSR sampling. FTSR sampling is first investigated in single user communication for MMSE equalization when zeros are padded among transmission blocks. There is a performance degradation in equalization of ISI channels depending on practical conditions. In compatible with that, the performance of Zero Padding (ZP) based FTSR sampled MMSE equalization is analyzed accounting for practical channel conditions. Although FTSR sampling is generally considered as a remedy to time or phase errors, the results illustrate that the advantage of FTSR sampling on equalization is always present for a finite block length ISI channel even if perfect time information is available. Block length is the key parameter that affects the performance of FTSR sampling such that there is a high Signal to Noise Ratio (SNR) improvement with smaller block lengths in comparison to the conventional Symbol Rate (SR) sampled MMSE equalization. Other parameters such as channel characteristics and excess bandwidth have influence on the performance of FTSR sampled MMSE equalization. The concept of FTSR sampling is applied to the Cyclic Prefix (CP) based MMSE equalization as well. The performance of this equalizer is evaluated in regard to average mean square error (MSE) and bit error rate (BER). It is semi-analytically proven that FTSR sampling provides an improvement in average MSE for asynchronous communication. This motivates the fact that there is a potential gain that can be exploited by FTSR sampling for CP based MMSE equalization. In this thesis, the complexity increase due to FTSR sampling is compensated by proposing a low complexity CP based MMSE equalizer implementation. FTSR sampling exhibits superior performance for CP based MMSE equalization in the following two applications. The first one is to yield more tolerant MMSE equalization when CP is shorter than the maximum channel delay spread. The second one is to obtain a reasonable performance from $1$-bit quantized MMSE equalizer for a single tap Rayleigh fading channel. In multiuser communication, FTSR sampling is studied to exploit additional Degrees of Freedom (DoF) of excess bandwidth for the purpose of increasing simultaneous transmissions. The analysis reveals that FTSR sampling achieves extra rank proportional to any excess bandwidth being used provided that multipath channel taps are not equally spaced, which is a physical reality. The impact is further quantified by an FTSR sampled MMSE detector and a practical iterative receiver based on FTSR sampling. Spatial domain interpretation is employed as well to assess the effectiveness of FTSR sampling such that multiple receptions are considered as virtual antennas. The simulation results illustrate that interference can be greatly reduced by the FTSR sampled iterative receiver such that the number of users can be increased in a given network. A counterpart of FTSR sampling method in frequency domain is the FDO, which is investigated to improve the error performance of MMSE Single Carrier Frequency Domain Equalization (SC-FDE) for multipath channels. The impact of FDO is analyzed in regard to the outage probability. The results show that FDO can significantly enhance the performance of MMSE SC-FDE when the ratio of block length to channel memory length is small such as in underwater acoustic channels. Its advantage is observed for moderate block length to channel memory length ratio in case of larger constellation sizes.