| The topic of this thesis is the parameter estimation of MPSK signals by cyclic spectrum method. This paper concerns in the improvement of the traditional cyclic spectrum algorithm and the implementation by Digital Signal Processor (DSP). And then parameters like symbol rate and carrier frequency are estimated.Signal detection and parameter estimation are still difficult, especially in low signal to noise ration (SNR) environments. Theories for these topics are still on researching. Communication signals, especially digital ones, usually have a property called cyclostationary, by which we could solve these problems effectively. MPSK signals also are cyclic stationary, and we could estimate the parameters with the cyclic spectrum—a kind of two dimension power spectrum.Cyclostationary signals are persistent random signals with statistical parameters that vary periodically with time. Most digital signals in communication are more appropriately modeled as cyclostationary because of underlying periodicities due to various periodic signal processing operations such as sampling, scanning, modulating, multiplexing, and coding. The fact that the cyclic spectrum for a signal can be accurately measured, even when the signal is buried in noise or masked by interference, contrasts with the fact that noise and interference have an unremovable (in general) masking effect on the power spectral density. Cyclic spectral analysis grows in importance as a signal analysis tool and the need for computationally efficient algorithms becomes increasingly.Over the recent years several computationally efficient cyclic spectral analysis algorithms have evolved from basic concepts, such as FFT Accumulaiton, but they are difficult for implementation and not suitable for real time processing. Based on time smoothing algorithm, two new methods are proposed in this paper. One of them has lower computationally complexity and could be implemented simply and flexibly with DSP. Carrier frequency and symbol rate could be estimated from the results of this method. Performance of this method is as good as traditional ones. The end of this paper is conclusion and some suggestions for further work. |
