Research On Reducing Computational Complexity Of CPM Signal Demodulation

The continuous phase modulation(CPM) has a wide range of applications because of its constant envelope and high bandwidth efficiency. However, the complexity of CPM signal demodulation increases exponentially with the increasing of modulation index and correlation length, which results in a high computational complexity and limits the application of CPM signal. This paper researches a method to reduce the computational complexity of CPM signal demodulation based on the Laurent decomposition.First, the paper describes the general expression and the phase state of CPM signal, then discusses how CPM works, CPM signal's characteristics and coherent demodulation in detail. The paper also introduces the Laurent decomposition and the optimal coherent detection method and sub-optimal coherent detection method.Finally, this paper introduces a method to reduce the computational complexity of CPM signal demodulation which using the feature that CPM signal's phase can be separated when it's modulated by REC pulse. When signal is modulated by separable phase response, the memory of phase is weakened, then it can be equivalent to a full response CPM signal, we can demodulate the pseudo-symbol which is the sum of original symbols and then get the original symbols by soft-decoding. The disadvantage of separable phase is that it increases the bandwidth of the signal which leads to the reducing of bandwidth.Then the paper proposes a new phase response combining separable phase response and traditional phase response, which is partial separable and can balance the computational complexity of demodulation and spectrum utilization. The nature of the partial separable phase response is a combination of separable phase response and traditional phase response, The number of symbols to participate in non-linear modulation can be changed by choosing correlation length, the symbols that has separable phase can be expressed by pseudo symbols, thereby CPM signals can be divided into two parts, Then the response can greatly reduce the complexity of the CPM signal demodulation with sacrificing only a small part of performance.