Study On Array Super-resolution And Altitude Measurement In VHF Radar

As an important research branch in modern signal processing, array signal processing has applied to many fields, such as radar, communications and sonar. Direction-of-arrival(DOA) is the most important research field in array signal processing, and has received considerable attention with the increasing development of array signal processing technique in the past three decades. Affected by the complex environment and rugged terrain, few super-resolution algorithms can obtain satisfying results, DOA estimation algorithms with high accuracy, suitable engineering realization and robustness become the objective to many researchers.As a kind of meter-wave radar, it has good anti-stealth effect, anti-arm capacity, less attenuation compared with microwave radar, and large detection range. The meter-wave radar has more great inspection ability on slow velocity target. Limited by aperture, meter-wave radar has wide beam, which usually causes bad angular resolution.Furthermore, high sidelobe excludes good anti-interference capacity. Especially when the beam irradiates on the ground, the direct path echo comes into the mainlobe as well as the reflection multipath echo, in which case the measurement accuracy of elevation deteriorates sharply. This dissertation starts from low angle estimation in meter-wave radar, mainly discusses low complexity, de-correlation and high estimation accuracy. The main content of this dissertation is summarized as follows.1. A novel method for DOAs estimation of multipath signals is studied to enhance the adaptability of the traditional high-resolution methods and to reduce the computational burden caused by eigen-decomposition procedure. The eigen-subspace is iteratively obtained by gradient algorithm for each snapshot, and then it is transformed from complex-valued space to real-valued space. Finally, Unitary ESPRIT or Root-MUSIC based on real-valued space is used to estimate the angles. Simulations and real data processing results show that, introduction of the eigen-subspace iteration and real-valued space processing makes the algorithm suitable for DOAs estimation of multipath signals without eigen-decomposition, thus reducing the computational complexity significantly.2. Chapter 3 focuses on dual-iteration algorithm based on row-column synthesizing and angle estimation with high-resolution algorithms. The traditional high-resolution methods for 2D DOA estimation using uniform rectangular array usually involve eigen-decomposition and multi-dimensional spectral search, which are procedures with high computational complexity. For single target with reflection multipath, row-column synthesizing is introduced to decrease the dimension, and real-valued 2D Root-MUSIC and RELAX are combined to estimate the DOA. Starting with some initial 2D angles, this method synthesizes the received data of planar array into vectors along both the row and column directions. Then real-valued Root-MUSIC or RELAX is used to obtain the elevation and azimuth angles respectively. Then, the weighted vectors for row-column synthesizing are reconstructed using the estimated 2D angles, and the 2D angles are obtained iteratively until they reach the given precision. The dual-iteration algorithm not only reduces the error of weighted vectors, but also improves estimation accuracy iteratively. This method is a simple procedure with a high convergence rate, and it has lower computation burden and comparable accuracy compared with the traditional processing methods, which are validated by real data processing results.3. Beamspace MUSIC method based on two-dimensional spatial smoothing is discussed. On the basis of spatial smoothing for uniform linear array, a de-correlation method for uniform rectangular array is presented. Spatial smoothing is implemented along both dimensions firstly to de-correlate the coherent signals. Then the de-correlated array data in element space is transformed to that in beamspace with much smaller size. Finally, 2D angles of coherent signals are obtained by virtue of beamspace MUSIC. The novel method not only de-correlates multiple coherent signals effectively, but obtains more robust estimation properties with much lower computational burden compared with the traditional high-resolution methods.4. Making use of iteration algorithm, sparse analysis based on discrete Fourier transform(DFT) gradually reinforces signal component while suppressing noise component by solving the objective function under some constraint, thus improving the resolution compared with digital beamforming(DBF). When applied to DOA estimation, the sparse analysis generates sparse solution corresponding to the DFT of the extrapolated data for a longer array, which can achieves resolution beyond Rayleigh limit. The analysis shows that, without eigen-decomposition and de-correlation, sparse analysis obtains comparable estimation accuracy with MUSIC. To spatial-temporal parameters estimation, sparse analysis carries out temporal sparse solution to get the frequency estimates and to separate the signals of different frequencies, then generates spatial sparse solution of each separated signals to enhance the spatial resolution and obtain DOA. Finally, a novel method for 2D DOA estimation based on sparse analysis using uniform rectangular array is proposed. Sparse solutions along azimuth direction and elevation direction are obtained in turn, then the solutions with different angular frequencies can be separated, 2D DOA can be estimated from each separated sparse solution. A modified method is presented to overcome the blind angular region problem occurred in the algorithm.5. A sparse planar array with two sizes of spatial invariances along both dimensions is analyzed to improve the angle estimation accuracy of sources in low angle region with reflection multipath. With two sizes of spatial invariances along both dimensions, the array estimates the azimuth and elevation angles using Unitary ESPRIT. The first spatial invariance with a displacement of half-wavelength yields unambiguous coarse estimates of high-variance, while the second spatial invariance with much larger a displacement of subarrays obtains cyclically ambiguous fine estimates of low-variance. The final estimates are obtained by disambiguating fine estimates with coarse estimates. A novel pairing scheme is proposed to pair both the azimuth and elevation, coarse and fine angle estimates of the same signal. Employing real-valued processing and array aperture extension, the new DOA measuring method de-correlates the multipath signals and enhances angle resolution with no extra antennas and computational complexity.6. A synthetic steering vector altitude measurement method based on highly refined multipath signal is studied. Taking reflection coefficient and phase difference between the two paths into consideration, we analyze synthetic steering vector MUSIC and ML algorithm based on 2D search on terrain parameter and angle. Based on second radar, a method calculating terrain parameter is proposed, which avoids actual measurement of terrain parameter and simplifies 2D search procedure to 1D angular search. Finally, an adaptive method which uses angular measurement error to adjust terrain parameters is analyzed, avoiding search procedure in calculating terrain parameter. The real data collected by some VHF radar demonstrates the validity and feasibility of the proposed methods.