Research And Realization Of The Algorithm For Modulation Identification And Parameter Estimation In Wideband OFDM Testing

OFDM (Orthogonal Frequency Division Multiplexing) has been developed quicklybecause of its high spectrum efficiency and the property of anti-multipath. It hasbecome the core technology of the4th generation mobile communication system andthe broadcast television network. In the field of radio spectrum monitoring,communication countermeasure and non-cooperative communication, the problem ofmodulation recognition and parameter estimation of OFDM signal has attractedextensive attention.In this dissertation, the OFDM signal based on IFFT/FFT is firstly analyzed, andthe mathematical model of modulated signal of OFDM is established, then the detectionscheme of OFDM signal is designed. The main problems to be solved are as follows:the detection of OFDM signal, the estimation of the related modulated parameters andthe recognition of the modulation mode of subcarriers.In the part of the detection of OFDM signal, based on the Gaussianity of OFDMsignal in time-domain, the matrix of the modified4th order accumulation and thenormalization4th or6th order statistics are chosen to construct the characteristicquantity for detection. Three types of characteristic quantities are compared in Rayleighchannel to gain the detection performance, and the modified4th order accumulation ischose to detect the OFDM signal.In the part of parameter estimation, firstly, the OFDM signal model based onprotocol analysis is proved to be2nd order cyclicstationary, so the cyclic spectrum canbe used to estimate of the carrier frequency and chip width of OFDM signal. Secondly,two types of estimation algorithms of cyclic spectrum based on smooth in time domainare studied, and the operand of FAM is improved. At last the structures of two types ofefficient algorithm, named FAM and SSCA, are analyzed and the reliability andcomplexity of them are compared. The chip width of signal is derived by cyclicspectrum estimation. The oversampled signal is extracted at a certain times. The coderate matched to the original symbol is derived. The total and the effective length ofOFDM symbol is estimated by utilizing the correlation between cyclic prefix and the corresponded useful symbol. This method is practical fairly.In the part of the recognition of subcarrier modulation mode, the multipath channelis firstly modeled as a slow-fading one. Because of the independence of thesub-channels in the OFDM system, each of them can be modeled as a flat fadingchannel. At last, the scheme for the subcarrier recognition is proposed, which iscompleted by extracting the instantaneous characteristics of the amplitude and phase ofsignals on the subcarriers.While in the DSP implementation, the running speed of FFT calculation has beenimproved by the method of lookup table, at the same time methods to handle theoverflow of the butterfly unit are proposed, and the optimization of the key module, i.e.,FFT, in FAM structure is completed. The realization of the optimization of fixed pointFAM on the fixed point chip of TMS320C6416is completed, which can double theefficiency of FAM realization. Finally, this dissertation combined Agilent vector signalgenerator E4438C with SystemVue, which can generate OFDM signal to test and verifyparameter estimate algorithm.