Optimization and Convenient Evaluation Model of Differential Coherent Post Detection Integration

Liu Wenxiang,Li Jingyuan,Ge Rui,Wang Feixue
2011-01-01
Abstract:The problem of pseudo-random (PN) code acquisition is common in many spread-spectrum systems, such as navigation systems, satellite communication systems, radar ranging systems, etc. Considering Doppler frequency error of carrier, the problem of PN code acquisition can be modeled using the problem of binary hypothesis testing. The method of combining coherent correlation and non-coherent post detection integration (PDI) is usually adopted in the above problem. Non-coherent post detection integration can be implemented as envelop detection or square-law detection. The signal total integration time T is divided into N segments, in each segment coherent correlation with T/N integration time before detection. After non-coherent detection, the partial results are processing by means of post detection integration. At last the accumulated variable forms into decision statistics. In some references the optimization of coherent integration time of envelop detection based PDI is analyzed. Other references [5] analyzed the performance of square-law detection based PDI and concluded that envelop detection based PDI is a little better than square-law detection based PDI.The differential coherent post detection integration (DC-PDI) is a new method proposed in recent years. The statistical characterization of the decision variable of DC-PDI is difficult to be modeled strictly under H1. Previous researches have investigated its performance and compared it with non-coherent post detection integration (NC-PDI). However, in those researches the decision statistics have been model by various theoretical approximations. The effect of those approximations on performance analysis and parameter optimization cannot be evaluated over the whole range of each involved parameter.In this paper the theoretical model and numberical simulation are combined to analyze the detection performance. Based on studying the statistical characterization of DC-PDI, the processing loss formula was introduced and the detection probability of DC-PDI under the constant false probability was determined by Monte-Carlo simulation. The optimal coherent integration time that minimizes the processing loss was researched in different cases using the Monte-Carlo simulating method, and the optimal processing loss of DC-PDI is compared with that of NC-PDI (using envelop detection), concluding that DC-PDI is better than NC-PDI.Furthermore, an approximate analysis model of DC-PDI processing loss with an accuracy of +/- 0.4dB, that is convenient to evaluate the detection performance of DC-PDI, was given after obtaining the curve fitting formula of differential coherent detector loss. The optimal coherent integration time that minimizes the processing loss was also re-calculated using the approximate analysis model, and the optimal processing loss difference between the approximate analysis and Monte-Carlo simulation is less than 0.3dB.
What problem does this paper attempt to address?