Spatial and spatio-temporal processing with antenna arrays in wireless systems

Abstract: This thesis treats some modeling and estimation problems inthe fieldof antenna array signal processing. With multipleantennas at thereceive side, the spatial dimension is added,and processing is performed in both the temporal and spatialdomain. The work is motivated by the interest in using antennaarrays, primarily at thebase stations, in wirelesscommunication systems. Among thepotential benefits areincreased range, fading diversity, interferencerejection andspatially selective transmission.The problem of estimating the non-frequency selectivechannels, referred to as spatial signatures, for a uniformlinear array withunknown receiver gain and phase responses isstudied. Sufficientconditions for identifying the spatialsignatures using a plane wavemodel with subspace based methodsare derived, and a closed formESPRIT-like estimator isproposed. Results from simulations andprocessing ofexperimental data show that the absence of receivercalibrationis not critical for uplink signal waveform estimationusing aplane wave model.The effects of angular spread caused by multipathpropagation from local scatterers on direction of arrival (DOA)estimation with theMUSIC algorithm are studied. Under theassumption of small angularspread, an approximate distributionof the DOA estimates is derived. An extension to more generalRayleigh and Ricean fading channels is included. Experimentaldata collected in suburban environments havebeen processed, andthe results agree well with the derived distribution.A generalized array manifold model for parameterized spatialsignatureestimation in environments with local scattering isproposed. Thearray manifold commonly used for plane waves isgeneralized toinclude linear combinations of the nominal arrayresponse vectors andtheir derivatives. To estimate theparameters of the model, severalsubspace based estimators areproposed.Spatio-temporal interference rejection combining fordigitalcommunication systems is considered. The use of a finiteorder linear predictor to take the temporal correlation of theco-channelinterference (CCI) into account is studied, and amaximum likelihoodsequence estimator which handles bothintersymbol interference andtemporally correlated CCI ispresented. Spatio-temporal interferencerejection may providelarge gains over space-only interferencerejection, whichneglects the temporal correlation, especially ininterferencelimited scenarios with angular spread and timedispersion. trainingsequence based estimator of the parametersfor the sequence estimatorwith spatio-temporal interferencerejection is proposed. Resultsfrom processing data collected ina suburban environment with a dualpolarized sector antenna of aGSM base station test bed are presented.