Training signal design and training-data power allocation in MIMO two-way relay systems

Abstract

Many works on wireless communications have assumed that perfectchannel state information is available at terminals. In practice,the channels should be estimated and a popular approach is thetraining based channel estimation method, where a known trainingsignal is transmitted. The training signal should be transmittedin every symbol block, where the data and training signalscoexist. There are several training methods depending on thearrangement of the training and data signals such astime-multiplexed, embedded, and superimposed training methods.These methods can be selectively applied in accordance with therelay deployment scenarios. In this dissertation, we consider theMIMO two-way relay systems, where the relay system consists of twosources and one relay, and each terminal is equipped with multipleantennas. The training signals for various methods are designedbased on the linear minimum mean square error channel estimate andthe training-data power allocation algorithms are proposed.In the time-multiplexed training method, the dedicated trainingand data signals are separated by time. Given that a mean squareerror (MSE) of the channel estimate exists at each source, weminimize the sum of the two MSEs. In particular, we first derive alinear minimum MSE channel estimator with an arbitrary trainingsignal. Second, we formulate the training signal design problem tominimize the sum of the two MSEs. Finally, we present an optimaltraining signal and a suboptimal training signal is obtained in aclosed form. The simulation results show that the proposedtime-multiplexed training signal has better MSE performance thanthe conventional training signals and the MSE performance of theclosed-form training signal is almost the same as that of theoptimal training signal.In the embedded training method, the training signal isarithmetically added into the data signal and the added symbol istransmitted in each symbol block. Since the data symbols aretransmitted in every symbol time,...