Subband Technologies Research And Its Applications In Wideband Signal Processing

In the recent years, with the rapid development of multimedia and communicationreceiving technology, the rate of wireless communication is higher and higher, the band ofreceiving signal is wider and wider. While the communication capability is increasing, sometraditional narrowband signal processing technologies are challenged by the widebandprocessing. The equalization of single carrier wideband singal and wideband multi-channelreceiving are the two outstanding problems. On the one hand, because of the long impulseresponse of the wideband transmission channel, the equalizer order should be high using thetraditional equalization technologies. The high order equalizer has bad convergenceperformance and huge calculation complexity, which made it hard to be realization. How toeffectively realize the equalization of wideband signal becomes one of the key questions toimprove the demodulation performance in wideband communication systems. On the otherhand, the quantity and type of signals included in the wideband data become more and moreas the increasing of the receiving bandwidth. Parallel digital down conversion is used by thetraditional multi-channel channelization method, which has huge computation and can’t adaptin dynamic channelization. How to effectively realize wideband multi-channel receivingbecomes another key question to improve the performance of wideband receiving systems.Subband technology can avoid the wideband processing difficulties by decompositionand reconstruction of the wideband data. It provides a new solution for wideband signalprocessing. It has significant theoretic and practical meaning to research the subbandtechnology and its applications. This paper gives out the design methods of subband filterbanks. It can decrease the design complexity and meet the reconstruction performance of thesubband system. Then the paper focuses on the subband applications in wideband signalequalization and channelization. Some new methods are proposed which can improve theperformance and efficiency of the wideband signal processing.The innovation points of the thesis include the following four aspects.1. The design methods for uniform subband filter banks are researched. The prototypefilters generally have high order and bad alias error performance using trandition iterativedesign methods. In order to resolve this problem, a modified iterative design method forprototype filter is proposed. The new method can decrease the alias error and the length of thefilter by adding the first-order alias term for the cost function. The cost function is formulatedas a quadratic function of prototype filter coefficients. A mathematics closed-form solution isdeduced for the new cost function. Its minimum point is obtained by iterative approach avoiding the difficulty of direct solution. This modified iterative method and the interpolationdesign method are combined in order to decease the parameter number of the optimizationalgorithm. Therefore it can greatly decrease the computation in the iterative design process.2. The indirect design methods of nonuniform subband filter are usually difficult toobtain its optimum solution. Aiming at this problem, a new iterative design method fornonuniform subband filterbank is proposed based on two-stage merging processing. In thismethod, the outside basic filter banks are designed firstly, then the inner merging filter banksare designed. The cost function is constructed by suppressing of alias terms. The nonlinearcost function is predigested and expressed by a quadratic function of the prototype filtercoefficient vector. Then the closed form solution of the cost function is deduced so that theoptimal solution can be achieved by iterative approach. The new iterative method has nospecial limitation on the decimation rate. As to the traditional design methods, it has betterconvergence performance and lower computation complexity.3. The realization methods are given out for the uniform and nonuniform subbandpartition, which are used in the subband equalization. Two parameters for partitionperformance evaluation are defined. The condition is deduced when the equalizer coefficientmatrix is diagonal-matrix. The error analyse is carried out for subband equalization withparallel structure. A new subband equalization scheme with adding adjacent term is proposedto decease the alias error when the decimation rate is large. The new method can suppress thealias error and loose the limitations for the stopband attenuation performance of the subbandfilter banks. Simulation results prove that the new subband equalization structure has betterconvergence performance compared with the subband equalization parallel structure.4. Aiming at the problems of the conventional subband channelization structure, a newnonuniform subband channelization scheme based on the features of subband signals isproposed. This new scheme uses the matched suband filters to merge the subbands. Iteliminates the limitations on the decimation rate. Moreover, its subband combination processis carried out only for the subbands which have the special features. So the new scheme canimprove the channelization efficiency for special signals. The subband signal model for singlecarrier signal is established. The baud rate and carrier frequency estimation methods arededuced based on the subband signal. The above estimation methods can be consulted byother parameter estimation questions under spectrum distortion conditions.