Research On Demodulation Methods Of MFSK Based On Compressed Sensing

Compressed Sensing as a data collection and coding and decoding theory whichappeared in recent years is different from the Nyquist’s sampling theory,which hasbeing used in the past years,and it can reconstruct sparse signal from few samples withhigh probability.It has been used in many fields such as the data compression,theAIC,the channel coding and the channel estimation.In this thesis,the demodulation ofMFSK signal which has good sparsity in the local Fourier matrix is researched based onCompressed Sensing.The concepts of the sparsity and the compressibility of a signal are introduced inthis thesis,which are the fundamental issues of Compressed Sensing.And the basicprinciples of Compressed Sensing,the design of the measurement matrix and the theoryof the restructuring algorithm are also studied. Then the concept of the recognition ofthe type of a signal modulation, especially the theory of the decision-making tree areintroduced. The demodulation of MFSK signal based on Compressed Sensing is showedin this thesis,and the SCD demodulation performance,the coherent demodulationperformance and the noncoherent demodulation performance are made.Results showthe SCD demodulation is better than the noncoherent method,worse than the coherentmethod.However, when achieving the same BER performance, the data size the SCDdemodulation needs is far less than that of the coherent demodulation.The fundamental basis for the demodulation with few data size in compresseddomain is provided by Compressed Sensing.It will bring the method of signal samplingto a new level,and it has broad prospects.