Research On Blind Demodulation Of Continuous Phase Modulation Signals

Non-collaborative communication is the technique used by the receiver to demodulate a signal while the sender has no knowledge of it. The receiver has to complete mainly two tasks: one is to estimate the various modulation parameters of the signal and the another is to complete the demodulation of the signal received with the parameters estimated.Continuous phase modulation(CPM) is a non-linear modulation signal with constant envelope and continuous phase. With high spectrum efficiency and power efficiency, CPM has many modulation parameters and all those parameters are interactive. And it has posed a challenge to blind demodulation of CPM.The estimation algorithm and demodulation algorithm of modulation parameters are elaborated in this thesis, and the blind demodulation of CPM is completed based on the interdependence between the two concatenate. Parameters blind estimation algorithm, based on the existing joint blind estimation of those parameters, introduces the concept of approximate entropy and annular statistics. Since approximate entropy and annular statistics are CPM's strong distinguishing attributes, thus they can be used to identify and classify modulation order and frequency shaping pulse by using support vector machines. Multiple symbol differential demodulation algorithms is adopted in demodulation, for it is, with its excellent adaptability to all kinds of CPM signals and unresponsiveness to initial phase information, suitable in blind demodulation without accurate initial phase information. It has been revealed through stimulation that the performance of demodulation can be improved by increasing the number of differential symbol, but the computation will be more complicated accordingly. At the last part of the thesis, the performance of the concatenated systems by concatenating parameters blind estimation algorithm and demodulation algorithm is analyzed, with the signal's parameters estimated through parameters blind estimation algorithm. The stimulation results show that the system has good demodulation performance and allow for some inaccuracies in those estimated modulation parameters, like symbol period, carrier frequency and demodulation index.