Research On Low Complexity Noncoherent Detection Of CPM

As a constant envelop modulation scheme with high spectral efficiency and high power utilization,Continuous Phase Modulation(CPM)is widely used in deep space communications,satellite communications and other wireless communication systems.CPM can generate a variety of modulation modes according to different modulation parameters.The best detection method for CPM is coherent Maximum Likelihood Sequence Detection(MLSD).However,coherent detection requires carrier phase synchronization,and the condition is very demanding,and it is not suitable for the channel with severe carrier phase change.Because noncoherent detection does not need to consider phase synchronization,the performance of the noncoherent detection is stable and reliable,and the structure is simple and easy to implement.However,the detection complexity is also related to the selected modulation parameters.In this thesis,it mainly studies noncoherent detection with low complexity of CPM.Firstly,this thesis analyzes and simulates the effects of modulation parameters on the power spectral density,the minimum Euclidean distance and the bit error rate of the CPM.The traditional noncoherent detection techniques,such as Noncoherent Maximum Likelihood Sequence Detection,are theoretically simulated and analyzed.Then,in order to realize low complexity noncoherent detection of CPM,this paper proposes to combine Multiple-Symbols Differential Detection with Reduced State Sequence Detection algorithm to use both the performance gain brought by multiple symbols and the advantages of reducing state.Secondly,two improvements are proposed for Noncoherent Maximum Likelihood Block Detection(Non-MLBD).The adjacent blocks are considered judgment overlapped,using the block part before a judgment has been for a decision after decision symbols block part determine the symbol,reducing the number of combinations of a reference signal,reduces the computational complexity,and simulation is made on the effects of overlap length on the detection performance.Secondly,this thesis proposes to output part of the judgment block symbols before the output of the intermediate signal.Because it is more reliable than other symbols before the intermediate symbol,the simulation proves that the algorithm has 0.4d B performance gain.Finally,the Laurent decomposition is applied to Non-MLBD,using several Laurent decomposition power concentrated amplitude pulse components as the reference signal with the received signal to relate,reduces the number of matched filters,getting low complexity detection of the Non-MLBD.