Study On Fast Radio Burst Localization Technology Based On The Understanding Of Focal Field Features

Fast radio burst(FRB)is a new kind of radio transient source which has been observed in recent years.It is characterized by short duration(?millisecond level)and high peak flux density(?Jansky),and it is considered as a sensitive probe to study large-scale cosmological problems.The timely and accurate positioning of rapid radio storms helps to retain sudden polarization information and trigger joint multi-spectral collaborative real-time observation,which is a prerequisite for studying its origin and physical nature.Except for the two cases of FRBs are repetitive,most of FRBs are one-off,and their radiation locations are very uncertain,which challenges the transient survey capabilities of the radio telescope.Aiming at the problem of expanding the field of view(Fo V)of the radio telescope and improving the localization precision of radio target,this paper proposes a novel detection method of FRB based on the understanding of the focal-field features.The main research contents include:(1)The geometry and texture characteristics of the focal-field distribution of the radio telescope in the time domain,frequency domain,and polarization domain are studied.The code of physical optics method based on Matlab is used to simulate,analyze,and verify the focal-field distribution in the time domain,frequency domain,and polarization domain.The FEKO commercial electromagnetic simulation software is used to realize cross-validation.The correlation between the position characteristics of radio target and the focal-field distribution is comprehensively analyzed.The texture features of the focal-field distribution of the radio target under different frequencies,incident angles,polarization,and translations are studied.The mapping relationship between different radio targets and corresponding focal-field distribution characteristics is established.(2)The performance model of FRB location based on focal-field feature matching is established.By setting up a complete focal-field distribution database in advance,the received signals from feeds will be matched with the database feature.According to the multi-frequency characteristics of FRBs,a multi-frequency joint estimation algorithm is proposed,which transforms the localization problem of FRB into a sparse convex optimization problem.Finally,the estimation of the FRB position is realized by the solved index information.(3)The undersampling,critical-sampling and oversampling problems of the focal plane fed array are studied,and the positioning effect of the method based on the feature matching of the focal plane field is simulated in the case of electrically large horn,electrically small horn and dense array element feeding.Simulation experiments show that for high-energy radio targets such as FRB,the number of feeds has a more significant effect on positioning than their gains.Moreover,for traditional cluster-fed radio telescopes,the focal-field distribution is seriously undersampled,which will seriously affect the accurate localization of radio target.In this paper,the super-resolution reconstruction of the focal-field distribution is realized by combining the generative adversarial network(GAN)technology,and a fast location estimation algorithm of radio burst based on the deep neural network is presented.The simulation results show that the proposed method can realize perfect reconstruction of the focal-field features and achieve higher real-time localization accuracy(much higher than the angular resolution of radio telescope)with the limited feeds.(4)According to the structural characteristics of FAST,a new type of active rotating feed array model is proposed,and the positioning algorithm under the model is established,which is verified by simulation.According to the regular hexagon feed structure,a symmetric three-channel data coding method is designed.The feed is de-noising by Dense Net-GAN,the inner continuity of different sampling data is mined by using two-dimensional long short-term memory(LSTM),and the final radio target is estimated using a capsule network.The simulation results show that the proposed method can achieve higher real-time localization accuracy and realize effective estimation for the radio target off the Fo V of the radio telescope,which partly improves the transient field of view of the radio telescope to some extent.