Multilayer systems are predestinated for high-rate wireless data transmission, where the source simultaneously radiates several data streams via multiple transmit antennas. The main computational effort of transmission schemes of this type is required at the receiver for separating the superimposed information signals.We distinguish between two types of detection schemes, where the first type requires
channel knowledge for the receiver and the second type performs a totally blind data estimation.
Recently, several detection algorithms have been investigated under the assumption of known channel state information at the receiver. This channel knowledge may be obtained by pilot-aided transmission and channel estimation. For these reference-aided transmission the well-known linear detection schemes are generally outperformed by successive detection schemes, namely, the V-BLAST algorithm.
In order to decrease the required computational complexity of this nonlinear scheme, we will restate it by applying the QR decomposition of the channel matrix. This yields the same successive interference cancelation (SIC) detection algorithm with complexity comparable to linear detection.
A detection approach avoiding pilot symbols is given by the concept of blind source separation (BSS). BBS approaches normally lead to linear spatial filters. In order to achieve the performance of reference-based detection, an iterative combination of BSS and nonlinear SIC is proposed. Furthermore, the application of BSS is favorable in many practical cases, as the sophisticated MIMO estimation
problems can be transformed to well-known solutions for SISO communication. The performances of the discussed schemes are demonstrated by simulations as well as by measurement results.