Study On Real-Time Signal Reconstruction And Upconverting Method Based On DSP

In this paper, on the base of different characters of deterministic signal andnon-Gaussian distribution random signal, the algorithm of signal reconstructionand the method to upconvert are analyzed in theory and applied to DSP systemswith TMS320VC33 as the core, which can realize real-time deterministic signalreconstruction with high ratio of signal to noise (SNR) and high frequencyresolution, both the algorithm design and the experimental study are included.The real-time algorithm to reconstruct time-domain signal from givenpower spectrum has practical value, it can be used in the hardware-in-the-loopsimulation of some signals, such as radar clutter signal, random vibration controlsignal and noise signal. In order to ensure the high precision, it presents rigorousrequest to the algorithm. This paper uses the typical SNR request of 50~60 dB inclutter signal rebuilding, based on which the reconstruction algorithm isrealized.During the reconstruction process of deterministic baseband signal, in orderto make sure the time-domain signal to be real and symmetrical, the initialpower spectrum is symmetrilized. In this case, it just needs simple organizationof short sequences to make long sequence, and can reach the purpose ofdecreasing computation and compressing data. During the reconstructionprocess of random baseband signal, the zero memory nonlinearity method isused, which uses white noise to excitated digital filter with required powerspectrum character. Corresponding to different system structure, this method canrealize the reconstruction of many kinds of non-Gaussian distribution randomsignal, and to some extent depress the pseudorandomicity and periodicity causedby the organization of short sequences.The upconverting process of baseband signal is realized in DSP systems bydigital quadrature modulation method, multiphase filter is used to fulfill theinterpolation extension and digital filtering simultaneously. Through the D/Aconverter and lowpass filter, the final signal can be reached. Experimentalresults indicate that the SNR in reconstruction algorithm of deterministic signalis up to 50 dB, the frequency resolution can reach 25 Hz, and the SNR inreconstruction algorithm of random baseband signal can reach 60 dB, thefrequency resolution is 100 Hz.