RECEIVED IQ IMBALANCE COMPENSATION FOR MASSIVE MIMO SYSTEMS

Abstract

Massive-MIMO has got special attention as one of the latest-most important tech- nologies used for a dramatically increasing needs of the spectral efficiency for the com- munication systems by exploiting the spatial domain professionally, especially with the tremendous progress in VLSI technology. Linear channel estimation techniques like Least-Squares estimation (LS), Minimum Mean-Square Error estimation (MMSE) are commonly used in massive-mimo systems. Also, the linear data detection techniques like Matched Filter detection (MF or known also as Maximum Ratio Combiner (MRC)), Zero Forcing (ZF) detection, or MMSE detection are commonly used in the massive-mimo systems. Radio Frequency (RF) impairments represent a major problem challenging the systems design for general, and the large scale antennas systems as particular. It can limit the performance of any system due to the sensitivity to the accuracy of the RF components, which leads to different distortion for the signal path within the system. These impairments shall be manipulated either in the analog or the digital domain in order to retrieve the required signal with minimum errors as well as minimizing the loss of system capacity due to these impairments. This thesis presents different ways to repair the effect of In-phase (I) and Quadrature- phase (Q) Imbalance (IQI) of the RF components, using a linear channel estimation and a linear data detection process for massive-mimo systems suffer a fading channel. Regarding the proposed IQI compensation algorithms, there are two pilot aided algorithms, one of them does not need an IQI estimation so it is resistant to the high IQI values and has low complex hardware implementation, however, the other, depends on a combined IQI and channel estimation, with higher complexity than the previous one. A blind algorithm is also proposed, with a good performance, but with a relatively high complexity than the others. The system practical capacity is shown and then an expression for the asymptotic capacity is achieved toward some of the proposed algorithms exploiting the massive properties, also the complexity is analysed for the different algorithms. The Bit Error Rate (BER) is obtained for different QAM modulation orders as uncoded performance, also the coded performance was obtained by integrating the proposed algorithms with a massive-mimo system with the LDPC decoder as a channel coding technique.