On Interference Rejection in Wireless Multichannel Systems

To meet the growing demands for service provisioning usingwireless access, efficient utilization of the available radiospectrum is a key issue of great importance. To achieve highspectrum efficiency, use of spectral efficient modulationmethods in combination with tight physical channel reuse areanticipated in future cellular communication systems. As aconsequence, the link performance and hence the capacity ofthese systems will to a large extent be limited by therobustness against interference provided by the receivingterminals operating within the systems. Thus, in all futuresystem aiming at serving mass markets, interference rejectionalgorithms---both transmitter and receiver oriented---will benatural components along with other capacity increasing meanssuch as adaptive resource allocation etc. In this thesis, interference rejection and parameterestimation methods suitable for multichannel systems aredeveloped and investigated, i.e., methods suitable forreceivers exploiting different diversity techniques to recoverthe data from a desired user. In particular, statisticalmethods to reject cochannel interference originating fromtransmitters supporting transmit diversity based on space-timeblock codes and methods to retrieve information about unknowninterferers needed to accomplish interference suppression arestudied. To handle the presence of transmit diversity interference, aspace-time processing based extension of the conventionalinterference rejection combining algorithm is proposed. Forsingle antenna receivers, it is shown that noise whiteningtechniques are effective for interference rejection purposes ifthe rate of the space-time block code is smaller than one. Itis further shown that for receivers with more than one antenna,alternatively if space-time block codes with rate smaller thanone are used, taking into account the inherent structure of thetransmit diversity interference is of major importance asinterference rejection by means of conventional processing willseverely degrade the achievable performance as compared tousing optimal space-time processing methods. To retrieve model parameters characterizing unknowncochannel users, a structured semi-blind parameter estimationtechnique utilizing the known training data of a desired useris suggested. Employing spatio-temporal processing, theproposed scheme exploits the temporal correlation present inthe output from a multichannel system to find parameterestimates that fit the range space of a structured