Iq Imbalance Of Channel Estimation And Compensation Algorithm Of Ofdm System Research

Transceiver based on direct conversion has three main advantages as follows:simple structure, multiple standards, and multi-band transmitting/receiving. But, there will be serious IQ imbalances; IQ imbalances will destroy orthogonality between sub-carriers of OFDM system, resulting in imaging interference and degrading BER performance of OFDM system. Therefore, how to realize a high-precision estimation of channel parameters in OFDM systems with IQ imbalances and how to effectively reduce and compensate IQ imbalances have become two very important research topics. This paper focuses on the research of channel estimation and compensation algorithm for OFDM systems with IQ imbalances, the main contents are organized as follows:(A) For OFDM systems with IQ imbalances at both transmitter and receiver, the frequency-domain (FD) based least square channel estimation algorithm (LS, Least square) requires many pilot symbols, the waste of resources, so the time-domain (TD) LS is proposed to improve the channel parameter estimation by using at least two pilot OFDM symbols (POSs). From simulation and theoretical analysis, it follows that due to the mining of the time domain characteristics of the radio channel, the time-domain LS requires only two OFDM training symbols to achieve the same error performance as the traditional FD-LS with16POSs.(B) Since the actual various wireless channels such as medium and short wave channels, hilly channel, and underwater acoustic channel are sparse, in order to fully exploit the sparse characteristics of these channels, and further improve the performance of channel estimation, we focus on the study of various channel estimation methods based on compressed sensing. We first establish the TD channel estimation model with IQ imbalances, and propose an TD-LS channel estimator requiring only one POS. Based on these, we introduce iterative shrinkage algorithm (IS) and parallel coordinate descent algorithm (PCD), and finally two estimators TD-LS-IS and TD-LS-PCD are formed TD-LS-IS and TD-LS-PCD are almost the same BER performance, their performance are better than those of LS and matching pursuit, but the complexity of TD-LS-PCD is lower than that of TD-LS-IS.(C) Considering those traditional linear compensation algorithms for OFDM systems with the IQ imbalances such as LS behave are much worse than the optimal ML (Maximum likelihood) detection, we mainly focus on constructing some suboptimal compensation schemes being close to ML on BER performance, two quasi-ML compensation methods are proposed based on the study of the classic ML compensation algorithm. The simulation results show that, compared with conventional LS and minimum mean square error (MMSE), the quasi-ML compensation methods perform better than the existing the linear compensation methods such as LS and MMSE, and have approximately the same error performance as the classic ML algorithm, but theirs computational complexities are significantly lower than the latter.