Yoon, Daejung2012-01-272012-01-272011-12https://hdl.handle.net/11299/120070University of Minnesota Ph.D. dissertation. December 2011. Major: Electrical Engineering. Advisor: Professor Jaekyun Moon. 1 computer file (PDF); xi, 97 pages.Iterative decision-directed (DD) channel estimation (CE) and detection algorithms for multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) systems are investigated. A main strength of the MIMO-OFDM is a potential capability to support high data rates. However, interference between MIMO antennas has been a serious obstacle to high data rate communication. Accurate channel state information (CSI) is critical to reduce antenna interference and improve throughput performance of the MIMO-OFDM communication systems. First, we develop soft-decision-driven sequential CE algorithms specific to turbo equalization for the MIMO communication. Two kinds of channel estimators are proposed: an optimal Kalman-based channel estimator geared to the pipelined turbo equalizer and a low-complexity estimator design for practical implementation. An effective strategy is established for the channel estimators dealing with different qualities of feedback decisions from the turbo equalizer. The proposed CE algorithms employ puncturing on observation samples to effectively deal with the inherent correlated error input that cannot easily be removed by the traditional innovations approach. Performance of the optimal estimator is excellent at compensating loss due to imperfect CSI; however, computational complexity of the MIMO CE becomes a challenge as the number of MIMO antenna links increases in practical systems. The proposed lowcomplexity algorithm resolves the MIMO channel estimation problem into a single-input single-output CE form to avoid heavy computation load associated with matrix operations. Also, in order to reduce packet losses due to the inherent correlated error, a novel packet recovery scheme is introduced that reprocesses failed packets by innovating on the inaccurate CSI. The recovery scheme detects erroneous OFDM-symbol locations by comparing extrinsic (EXT) information from the turbo equalizer. For the error correction, it applies additional turbo iterations to the erroneous OFDM symbols with the innovated CSI. In demonstrating the viability of the proposed schemes, a MIMO-OFDM communication system is constructed to comply with the IEEE 802.11n WLAN standard.en-USChannel estimationIterative detectionMIMOOFDMTurbo equalizationElectrical EngineeringThesis or Dissertation