| Automatic Dependent Surveillance Broadcast(ADS-B)is one of the new main technologies for Air Traffic Management.The aircraft realizes information sharing and air traffic surveillance by broadcasting location,identity,speed and other information.The current ADS-B surveillance system is mainly based on ground-based system.As the ground stations have a small coverage and its cost of deployment is quite high,it's not conductive to covering a large area,ocean and remote areas for ground-based ADS-B system.Therefore,the development of the satellite-based ADS-B system appear as the trend of the times.Satellite based ADS-B system has a series of new functions,including a wider range of coverage,lower signal to noise ratio and less power difference between signals.These new functions make ADS-B signals collisions at the satellite-based ADS-B receiver more severe.This paper mainly studies multiple overlapping signal separation algorithm for satellite-based ADS-B.The multiple overlapping signal is modeled and analyzed.The necessity and importance of signal separation are demonstrated by calculating the relationship between the probability of overlapping and the amount of aircrafts.In this paper,three different algorithms are used to simulate and separate the multiple overlapping satellite-based ADS-B signals,and the algorithm is optimized and analyzed.First of all,the extended projection algorithm is studied,this algorithm is used to separate the satellite-based ADS-B signals,and get the estimation of the source signals by means of matrix recombination and rank decomposition.Then,the independent component analysis(ICA)algorithm is studied.In order to reduce the work of calculation,before the separation work,the overlapping ADS-B signal is pretreated with zero mean and whitening.Considering that the FastICA algorithm based on negentropy has local maximum and its convergence depends on the initial value,this paper import loose gene in the classical FastICA algorithm,a loose modified method is proposed and applied to the ADS-B signal separation algorithm.Finally,Joint Approximative Diagonalization of Eigen matrix(JADE)algorithm and Loose Modified FastICA algorithm were used to simulate the multiple overlapping signal separation experiments,the correlation of ADS-B source signal and estimated signal separated by loose modified FastICA algorithm is analysed,the feasibility of the algorithm is verified,the performance and convergence speed of the loose modified FastICA algorithm and classical FastICA are compared.Experimental results show that these three algorithms all can solve the ADS-B signal overlapping problem,and the the loose modified FastICA algorithm could not only reduce the computational quantum,be more suitable for the hardware to perform.but also implement convergence in large-scale,improve the performance of the algorithm. |
