Research On Low Complexity CPM Demodulation Algorithm And Its Robustness

With the rapid development of aerospace technology in recent years,long-distance satellite communication and deep space communication have become one of the research highlights,and then the research of Continuous Phase Modulation technology(CPM)which is suitable for this kind of communication scene has been gradually deepened.CPM is a constant envelope digital modulation technology with high spectral efficiency and power utilization.Its advantages are closely related to the modulation parameters,but at the same time these parameters also affect the demodulation complexity.In addition,the frequency offset and phase offset caused by Doppler or imprecise synchronization will also affect the demodulation of CPM signals.In order to take into account the reliability of the system while applying the advantages of CPM signals,it has become a popular direction to connect channel coding technology with CPM systems.Thus the purpose of this thesis is to study the low-complexity CPM signals demodulation algorithm and its robustness under different frequency offset or phase offset,and then to further study the serial cascade CPM system with excellent error correcting codes.The work of this thesis is as follows:1.When the CPM signals are demodulated,the amount of calculation of the branch metric in the Viterbi algorithm and the Max-log-MAP algorithm increases exponentially with the base number and correlation length of the CPM signals.In order to solve this problem,based on the specific transition characteristics of the phase state diagram of the CPM signals,this thesis groups them,then a simplified method to reduce the number of states according to the phase state jump law is put forward.Meanwhile,according to the phase relationship between specific state groups,a simplified method based on phase difference is proposed.These two methods can reduce the calculation amount of branch metric of Viterbi algorithm and Max-log-MAP algorithm by half.In the case of specific parameters,the two algorithms can be used together,and the amount of calculation can be further reduced to 1/4.The simulation results show that when the state reduction simplification method is adopted in the binary CPM signals,there is only 0.1d B performance loss when the bit error rate is10^-5.When the state reduction simplification method is adopted in the quaternary CPM signals,there is only0.3d B performance loss when the bit error rate is10^-5.The direct negative operation based on the phase difference does not change the branch metric calculation results,so there is no performance loss when the phase difference between different state groups simplification is used.2.CPM signals carry information in the phase change,so it is very sensitive to frequency offset and phase offset.In order to solve this problem,this thesis analyzes the influence of fixed frequency offset and phase offset on the calculation results of branch metrics.On this basis,it is concluded that,compared with Viterbi demodulation algorithm,the system has higher phase offset tolerance when using Max-log-MAP algorithm to demodulate.Furthermore,through the approximation,it is concluded that more computation in the Max-log-MAP algorithm does not contribute to the resistance of the system to the fixed frequency offset,and the weak performance advantage is obtained only by the improvement of the phase offset tolerance of the system,and the above conclusion is verified by simulation.Next,the performance of LDPC-CPM cascade system and Polar-CPM cascade system under different phase offset is studied and compared.The soft demodulation is carried out by Max-log-MAP algorithm combined with reducing the number of states simplified method.The simulation results show that when the phase offset is large,the performance of the Polar-CPM system is better,but the improvement of the phase offset tolerance of the system by reducing the code efficiency is smaller;although the LDPC-CPM system is more affected by the phase offset,the phase offset tolerance of the system can be significantly improved by reducing the code efficiency.