Research On The DOA Estimation Of Wideband Signals

As an important part of signal processing, array signal processing has extensive application in many fields such as communication, radar and sonar. The direction of arrival (DOA) estimation is an important research subject of array signal processing. About the estimation of narrow-band signals, many experts and scholars have proposed many effective algorithms, and the research of narrow-band signals has been already mature. Because we can obtain more target information from wide-band signals than narrow-band signals, and the information can afford information guarantee in many fields such as the estimation of basic parameters (speed, distance and orientation), target recognizing, character-obtaining and counterworking. Lately years, the algorithms research of the DOA estimation of wide-band signals become a new hotspot in array signal processing field.The algorithms of the DOA estimation are all about the narrow-band signal firstly, but the signals which the array received are always wide-band signals in the processing systems of the communication and radar. To processing the wide-band signals, a simple method which called incoherent signal-subspace method (ISM) decompose the wide-band signal into many narrow-band signals, process each narrow-band signals, and integrate all results. The ISM method can estimate the DOA of wide-band incoherent signals, but cannot estimate the DOA of coherent signals. The coherent signal-subspace method (CSM) use focusing matrix to accord with every frequency observation in a certain subspace, obtain the focused observation, and then estimate the DOA of signals. The CSM method focus the signal subspace to reference frequency by using focusing matrix, the subspace is reflected the every narrow-band power spectrum matrix. And average every focused narrow-band power spectrum matrix, then we can obtain the focused and averaged covariance. The CSM method has better estimation precision and has lower differentiate threshold value, and the focusing transform is same as smoothing in frequency field. So the CSM method can estimation the coherent signals. But the method needs a direction estimation and focusing frequency in advance to conform the focusing matrix. And this method is easy affected by the signals. We always use these two methods to estimate the DOA of the wide-band signals. But the ISM method can only estimate the incoherent wide-band signals. The CSM method can estimate the coherent signals, but it needs a direction estimation and focusing frequency in advance to conform the focusing matrix. So this method is easy affected by the signals and need large operation, cannot process the signals in real time. This paper propose a new method to estimation the DOA of wide-band signals, which resolves the problems that the above two methods cannot resolved. This new method modifies the ISM method, so this method is called modified ISM method.The ISM method cannot estimate the coherent wide-band signals, but it can estimate the incoherent signals very well. If we can decrease the relativity of the wide-band coherent signals, and use the ISM method to estimate the DOA, we can estimate the DOA of the wide-band coherent signals. To decrease the relativity of the wide-band coherent signals, we can decompose the wide-band signal into many narrow-band signals and decrease the relativity of every narrow-band signal. Then estimate the DOA of every narrow-band signal and average these DOA estimations. So we can estimate the DOA of the wide-band signals. This is the basic principle of the modified ISM method.There are many methods to decrease the relativity of the coherent signals. The common methods are spatial smoothing and spatial filtering. Although the spatial smoothing can resolve the problem of coherent sources, but aperture of array is reduced, because the aperture of the sub-array is smaller than the originally array. The reducing of the aperture increases the beam-width and decreases resolving power. The spatial filtering can resolve the problem of the aperture of array, but this method needs more operation and preliminary estimation. Also this method depends on the preliminary results. Then, this paper proposes a simple and effective method which is forward backward averaging. This method uses a replaced matrix to obtain the backward array covariance matrix. To average the array covariance matrix and backward array covariance matrix and obtain a new matrix which is a full rank matrix. If we can estimate the DOA of the new matrix, we can estimate the DOA of the coherent signals. The new matrix is a full rank matrix, so we can decrease the relativity of the coherent signals. Simulation result and theory analysis show that the method has better ability to process the coherent signals.The modified ISM method is based on the forward backward averaging to estimate the DOA of the coherent wide-band signals. Firstly, according to the idea of the ISM method, decompose the wide-band signal into many narrow-band signals and decrease the relativity of every narrow-band signal by using the forward backward averaging. And estimate the DOA of every narrow-band signal by using MUSIC method and average these DOA estimations. Then we can estimate the DOA of the wide-band signals. Simulation result and theory analysis show that the modified ISM method can estimate the DOA of the coherent wide-band signals, and this method also has better resolving power and less operation. So we can validate that this method is effective.