Study On Smart Antenna Technology For OFDM System

Orthogonal frequency division multiplexing uses a cluster of parallel andlower data rate subcarriers to realize high data rate digital communication.Recently years, it has been considered to be one of the key technologies for thenext generation mobile communications systems for its good performance incombating multipath fading, high spectral efficiency and so on. Unfortunately, itperforms poorly in the presence of cochannel interference. The adaptivebeamforming techniques for OFDM systems using antennas array at receivercan not only mitigate delay spread and multipath fading, realize space divisionmultiple access and suppress cochannel interference consequently, decrease biterror rate, but also increase spectral efficiency and system capacity.The paper focuses on adaptive beamforming algorithms of OFDM systemswith antenna array. The job has been done are mainly presented as follows:(1)Two kinds of time domain reference beamforming algorithms areinvestigated and compared, and an improved algorithm is proposed. Thisalgorithm periodically computes the initial weight vector by the preamble errorin the beginning of a frame, and updates the initial weights using the pilot errorduring the frame.(2)Based on the studies of classical SMI adaptive algorithms andsubcarriers grouping scheme, a novel frequency domain beamforming scheme ispresented. To reduce complexity, the scheme computes a set of beamformingweight vector and applies the weight vector to all the subcarriers. In frequencyselective fading environment, in order to mitigate performance degradationarising from different channel fading across subcarriers, frequency domainchannel estimation and equalization for the post-beamforming channel areemployed at individual subcarrier.(3)Based on the model of MC-CDMA system with antenna array, a PS-DRMTA beamforming algorithm is proposed. This algorithm exploits preamble-assisted de-spread re-spread data to update the weight vector, which candecrease the effct of the judge errors, then the weight vector may credibilityconverge to the optimum weight vector. By using variable size gradient descentmethod the algorithm needs less computational load.Performance improvements of the proposed algorithms are investigatedthrough computer simulations.