author = {D. W\"{u}bben},
  year = {2006},
  month = {Feb},
  title = {Effiziente Detektionsverfahren f\"{u}r Multilayer-MIMO-Systeme},
  type = {Ph.D.-Thesis},
  school = {University of Bremen, Germany},
  publisher = {Shaker Verlag},
  ISBN = {978-3-8322-4803-1},
  address={Aachen, Germany},
The wish to obtain higher data rates in wireless data communication requires the consistent exploitation of the available physical resources. Special interest will be donated to not yet exploited resource space, which is available by the application of several antennas at the transmitter and the receiver. This thesis focusses on the multilayer MIMO system, where the data is divided at the transmitter into several data streams and emitted simultaneously. Due to the superposition of the data at the receiver simple detection of the data is no longer possible and thus requires new adapted methods from signal processing. 

This thesis prioritizes the analysis of these concepts and the development of new approaches for efficient detection. Therefore, several receiver architectures for non-frequency selective and for frequency selective transmission channels are presented. The successive detection will prove to yield a good compromise between complexity and performance. With the developed approach on basis of a sorted QR decomposition the required complexity is clearly reduced with only a marginal difference to the optimal successive detection. This fact is shown with respect to Monte-Carlo simulations and a detailed analysis of complexity. 

A further promising approach is given by the combination of successive detection with lattice reduction, a well known tool from lattice theory. The developed enhancement with respect to the MMSE criterion realizes almost the performance of the optimal detection. In comparison to the schemes known from literature this results in an obvious improvement of the detection quality with simultaneously decreasing the complexity. 

Within frequency selective scenarios spatial and temporal superposition of the signals occurs and results in a two dimensional equalization problem. To solve this task, the impulse shortening of MIMO systems and the extension of the successive detection method to frequency selective channels are investigated. In order to apply the detection concepts presented for nonfrequency selective channels it is also possible to transform the system into a multiplicity of flat MIMO channels in terms of OFDM or frequency domain equalization. In this connection the generalization of the sorted QR decomposition for several parallel channels realizes again a remarkable reduction in complexity in comparison to the scheme from literature.