Multistage Blind Source Separation In Mimo Systems
No Thumbnail Available
Links to Files
Electrical and Computer Engineering
Doctor of Engineering
Citation of Original Publication
This item is made available by Morgan State University for personal, educational, and research purposes in accordance with Title 17 of the U.S. Copyright Law. Other uses may require permission from the copyright owner.
In this work, the problem of equalization and blind source separation (BSS) in multiple-input multiple-output (MIMO) communication systems is discussed. MIMO uses multiple antennas at the receiver and the transmitter. The MIMO systems have received significant attention in the communication society, due to its achievable capacity gain. The high capacity gain in MIMO systems arises from exploiting the spatial and temporal diversity in the received signal from different antennas. In MIMO systems, it is possible to transmit several signals on the same bandwidth, without allocating a specific sub-channel to each signal. When multiple signals are transmitted over a MIMO channel, signal processing techniques are used not only to equalize the signals, but also to separate the transmitted sources at the receiver. An equalizer is needed to remove the inter-symbol interference (ISI) and reverse the effect of the channel on the signal. Constant modulus algorithm (CMA) is a well investigated adaptive blind equalization technique and it is widely used to equalize signals with constant modulus. However in most higher order modulation schemes, the signals are not of constant modulus. For instance, in quadrature amplitude modulation (QAM) schemes such as 16-QAM, the signals have several amplitude levels. For non constant modulus signals such as 16-QAM, Constant Modulus Algorithm (CMA) is used in a linear combination with Alphabet Matched Algorithm (AMA). For signal equalization, the cost function of the CMA+AMA equalizer is adaptively minimized. For source separation, a multistage channel estimation/signal cancellation method has been developed and analyzed to separate the recovered data sources. In this method, individual signals are recovered and equalized by MISO CMA+AMA equalizers at each stage. After equalization, the corresponding channel vector to the captured source is estimated and a replica of the contribution of the captured signal is subtracted from the received signals. This gives way to the next signal to be captured and equalized at the next equalization stage. A triply selective channel model (i.e. one that utilizes frequency, time, and space) is implemented, and the proposed multistage equalization and source separation technique is evaluated over variations of this channel model. Experimental results illustrate the practicability and effectiveness of the proposed technique for the BSS problem.