Research And Implementation Of Separation Algorithm For Space-based 1090 ES ADS-B Overlapping Signals

With the development of aerospace science and technology,automatic dependent surveillance broadcast technology has become a key technology in the field of air surveillance.The aircraft sends its own height,speed,longitude,latitude and other important information through a fixed format to realize airspace surveillance by broadcasting.Installing ADS-B receiver on LEO satellite can solve the problem of small coverage of ground receiver,so as to realize global coverage and monitoring.However,space-based ADS-B system is faced with a more serious problem of signal collision and overlap at the receiver,which has a serious impact on signal decoding,resulting in error decoding of important information,or even loss.In this paper,the separation algorithm of overlapping ADS-B signals is studied.First of all,through modeling and simulation,this paper clarifies that the probability of two ADS-B signals overlapping accounts for a large proportion of all overlapping cases.The traditional independent component analysis algorithm,projection algorithm and its extension algorithm are introduced in detail.Secondly,this paper proposes a time domain ADS-B overlapping signal separation algorithm.When there is a certain power difference between the two source signals,the overlapping signals are offset by the large and small power signals respectively,and the corresponding cancellation signals are obtained.According to the superposition mode of different pulses,different bit judgment results are obtained according to the amplitude,so as to recover the source signal.The experimental results show that when the source signal has 3d B power difference and the SNR is 5d B,the bit error rate of the algorithm is 10.1%,which is 5.4%and 3.7% lower than the traditional independent component analysis algorithm and projection algorithm respectively.With the increase of power difference and SNR between source signals,the bit error rate decreases to 1.2% when the power difference is 4d B and the SNR is 25 d B.Finally,this paper proposes an algorithm based on complex neural network and Hilbert transform to improve the performance of overlapping signal separation in low SNR situation.After Hilbert transform,the overlapping signals are taken as the input of the complex neural network.The first half of each input is the real part of the signal,and the second half is the imaginary part of the signal.The output is the estimated time-domain waveform of the two source signals.The experimental results show that the complex neural network can effectively separate the overlapped ADS-B signals.When the two source signals have 3d B power difference and the SNR is 5d B,the bit error rate can reach 9%,which is 1.1% lower than that of the time domain cancellation algorithm,and the operation time reaches the millisecond level,and the operation efficiency is improved 100 times.When the power difference between the source signals is 4d B and the SNR is 25 d B,the similarity coefficient reaches 0.9314.