Research On Key Technology Of Direction Finding For Mobile Communication Spectrum

With the development of mobile communication technology,the bandwidth of spectrum is gradually expanded to support faster transmission of massive data.At the receiver of antenna array in mobile communication,DOA estimation is used to extract useful signals and estimate directions,which is significant in many applications.Early techniques of DOA estimation are mainly based on narrow-band signals.However,wide-band signals are increasingly broadly used.Besides,coherent signals often exist in mobile communication environment due to multipath effect.Therefore,this thesis focuses on DOA estimation of narrow-band coherent signals and wide-band coherent signals.Main research results are as follows:Firstly,a grouping DOA estimation method is proposed in respect of narrowband multipath signals.After blind separation of independent sources,the method decoherences the homologous multipath signals into groups,uses Gerschgorin Disk Estimation for source number detection,and finally estimates DOA.Simulation results show that the method can distinguish multipath sources and is also suitable when the multipath number is unknown.A matrix reconstruction algorithm based on marginal sensor correlation denoising is proposed for decoherencing.Based on the spatial independence of noise,the algorithm divides the array into two groups,each consisting of a single marginal sensor and the residual subarray.In this way,the self-sensor noise component is eliminated due to complete mutual-sensor correlation.Simulation results show that RMSE of the algorithm can be reduced by 2.8° at the same signal-to-noise ratio,and the success rate can be increased by10%.Moreover,the performance raises more when signal directions are close.Secondly,a multi-focus subspace algorithm based on adjacent frequency is proposed in respect of wide-band coherent DOA estimation,which is equivalent to expanding the frequency smoothing to three times and making more use of the information at each subband.Simulation results show that RMSE of the algorithm can be reduced by 4.4° at the same signal-to-noise ratio,and it can still maintain excellent performance under conditions of degree separation within 10°and bandwidth up to 60%.