A.N. Korolev, A.F. Kotov, D.A. Fofanov
Radar developers are working continuously on improvement of its characteristics, resolution in particular. This problem may be formulated shortly as a superresolution . Modern receivers are being developed as an optimal for a single echo signal, that is why solving a superresolution problem for such receivers is a challenge of great interest.
Some results of investigations in this direction are given in the paper. At the first stage an absence of noise is supposed and interaction of two received signals is analyzed.
Parameters of echo signal are proposed to be determined in a following way. A sum of determined signals with known parameters should be fed to the receiver’s input. By means of varying these known parameters one should try to obtain maximum coincidence of the resulting response with that of being investigated. Root mean square error is proposed as a measure of coincidence.
An approximation of receiver output principal maximum can be applied to reduce computational complexity. Parabolic and cosinusoidal approximations are shown to be suitable. Approximation functions parameters can be found with by least squares algorithm on basis of the response samples. High quality of approximation can be achieved by increasing of sample rate.
Results of simulation of signals parameters estimation procedure are given. Responses are shown to be successfully resolved even when signals are as close as 10% of width of LFM autocorrelation function. RMSE that is used as a measure of coincidence turned out to decrease slowly in case of a Doppler shift because of specific features of LFM signals that were used in simulation.
A threshold used in processing may differ from one that is used in detection routine. Information from receiver may be saved for the whole response length. Data is being sent to processor after detection threshold exceeding and during this processing samples may be compared with arbitrary threshold that depends on processing features.
Simulation results analysis testified a principal possibility of such additional processing in case of resolution problem, even when the receiver is not optimal. This methodology allows developing of super resolution algorithm that doesn’t require radar hardware upgrade.