Iterative Channel Estimation for UMTS Long Term Evolution

In this thesis, different iterative channel estimation algorithms for Long Term Evolution (LTE) downlink are investigated. LTE uses coherent detection, which requires channel state information. In order to achieve high data rate transmission over mobile radio channels, it is essential to have accurate channel state information at the receiver side. For channel estimation purpose LTE provides training data known as pilot symbols. The matter of discussion is whether the accuracy of channel estimate based on pilot symbols is satisfactorily sufficient to achieve high data rate transmission. Channel estimate can be further enhanced, if after pilot based channel estimation, additional information such as the hard or soft estimated data symbols from the decoder is utilized by the channel estimator. Using this additional information, different channel estimation algorithms are derived. Their performance is discussed and compared with each other. The impact of processing either extrinsic, a-posteriori or hard feedback information in the channel estimator is investigated. To assess the performance of channel estimators, we exploit the LTE Link Level Simulator, developed at the Institute of Telecommunications (TC), Vienna University of Technology. Channel estimators are compared in terms of Mean Square Error (MSE) and throughput for slowly changing channels. For the SISO (4 × 4 MIMO) transmission mode using soft feedback information, iterative Least Squares (LS) channel estimator improves about 0.8 dB (0.45 dB), and iterative Linear Minimum Mean Square Error (LMMSE) channel estimator about 0.4 dB (0.7 dB) with respect to the initial channel estimators. The iterative LMMSE estimator loses approximately 0.05 dB with respect to the system with perfect channel knowledge. Although, the performance of iterative LMMSE estimator is superb, its complexity is too high for a real-time implementation. In order to reduce the complexity, meanwhile preserve the performance of the iterative LMMSE estimator, iterative approximate LMMSE (ALMMSE) estimator is investigated. The iterative ALMMSE estimator uses the correlation between the L closest subcarriers. Variation of L allows us to adjust the performance and complexity of the estimator to attain a good trade-off. Accordingly, iterative ALMMSE estimator, for a chosen L = 12, gains about 0.75 dB (0.9 dB) with respect to initial ALMMSE estimator, and loses about 0.2 dB (0.3 dB) compared to iterative LMMSE estimator.