Study Of Array And Signal Reconstruction For Distributed Multi-ship Based Surface Wave Over-the-horizon Radar

On the basis of both shore-based and shipborne surface wave over-the-horizon radar (SWOTHR), Distributed Multi-Ship based SWOTHR (DMS-SWOTHR) system is just beginning to develop, which employs more transmitters and receivers mounted on various ships in a battle group. Compared with the first two radar systems, DMS-SWOTHR is more advanced, but it has a much more complex configuration. Besides outstanding agility and maneuverability, DMS-SWOTHR can flexibly configure and reconstruct the entire system resource according to the detection requirement, the signal environment and the target type, and hence it can provide the optimal matching for the operational mission and the maximum using efficiency for the detection task. Also, DMS-SWOTHR is a highly complex and delicate Doppler resolution system that implements target detection and extraction by distinguishing the weak motion difference between the target and the background clutter in the Doppler domain. By far, the referable data of the researches on DMS-SWOTHR system are still absent, and this radar system has to be confronted with great challenges.Based on the above background, this dissertation begins with the fundamental researches on the distributed mobile radar system model and the characteristics of target and clutter, and then the further researches on the array and signal reconstruction techniques and the coherent signal processing techniques based on a reconstructed large array are made. After obtaining the dynamic geometry model, the received signal model and the Doppler-broadening expression of the first-order Bragg sea waves, this dissertation concentrates on the following problems both in theory and in simulation: how to extract the ships whose Doppler frequencies appear in the spreading domain of the first-order Bragg lines, how to perform the array reconstruction, and how to make the existing spatial super-resolution algorithms be applicable to the reconstructed large array.Firstly, in the DMS-SWOTHR system with a single transmitter and multiple receivers, since the receivers are mounted on different ships and separated from each other physically, this dissertation analyzes the basic theory of sea clutter for the shore-based narrowbeam bistatic SWOTHR, including the mechanisms of first- and second-order sea clutter and their mathematical models. According to the analysis, the sea surface is a type of special scatterer in high frequency bands. The first-order Bragg scattering is caused by gravity-waves with specific wavelength that is related to the radar operating frequency and the bistatic angle; the second-order sea clutter power is much weaker than the first-order power but its mechanism is much more complex; and the directional sea wave height spectrum has a remarkable effect on the first- and second-order sea clutter.Next, the dynamic geometry model of the transmitter platform, an arbitrary receiver platform and a moving target is derived. Based on the model, this dissertation analyzes the spreading mechanism of the first-order Bragg line in DMS-SWOTHR and presents a mathematical spreading model. The Doppler frequencies of sea echoes are simultaneously modulated by different radial velocity components projected from the radar platform motion, which results in the spreading of the first-order Bragg line; moreover, the time-varying Doppler frequencies are imparted due to the unavoidable motion difference between the transmitter and receiver platforms. Hence, the Doppler-broadening spectrum of the first-order sea clutter becomes very complex, which further weakens the ship target detection. The main interference for the detection of ship targets in the spreading domain is the first-order sea clutter from the direction different from the targets'azimuth angles but with the same Doppler frequency considering the platform motion. According to the synthetic aperture effect of radar platform motion, the azimuth of the first-order sea clutter falling into an arbitrary Doppler cell is known, and in a statistical sense, it is different from that of the target in the same cell. These conditions offer the technical possibilities for the target extraction. When the azimuth angles of the target and the first-order sea clutter are equal, changing the radar operating frequency or the motion of the radar system is effective.The above-mentioned geometrical relation is further employed to obtain the received signal model. As viewed from the phase expression after demodulation process and conventional range transform, the mathematical forms are the same for the ship target with a constant radial acceleration, the ship target with a relative centripetal acceleration, and the first-order sea clutter interference with a time-varying Doppler shift. Based on the bistatic radar equation, the relative amplitudes of the first-order sea clutter from different directions within the range cell are not equal when neglecting the statistical factor. Although the second-order sea clutter are also affected by the platform motion, the second-order spectrum, compared with that in the shore-based case, is still a continuum and changes indistinctively. Thus, the second-order sea clutter continuum and the atmospheric noise in each range-Doppler cell are both considered as the additive noise whose amplitude and phase are modeled by Rayleigh and uniform distributions, respectively. This dissertation gives the simulation results under different conditions based on the above signal models.Then, for the radar echoes containing time-varying and non-time-varying signal components in the Doppler domain, this dissertation proposes a scheme that is a recursive procedure for target extraction. The main idea is that the orthogonal weighting technique is performed to cancel the broadened first-order Bragg lines, so that not only the uniformly moving targets could be first detected due to little coherent integration loss (CIL) but also the cross-terms to be produced in the subsequent steps are greatly reduced in advance, and then the product high-order ambiguity function (PHAF) based spectra are obtained to estimate the corresponding radial motion parameters of the nonuniformly moving targets, respectively. In this scheme, an arbitrary target, once extracted fully, has to be removed for the purpose of suppressing the deterministic noise generated by the cross-terms. The simulation results shows that at low peak SNR, the PHAF-based method exhibits a threshold effect, but at high peak SNR, the performance is very close to the corresponding Cramer-Rao lower bound (CRLB). In addition, the estimation accuracy degrades because of the error propagation as the"peeling"algorithm proceeds.Finally, based on the special configuration of DMS-SWOTHR, this dissertation proposes a radar resolution cell oriented array reconstruction method. The coherence conditions for a given signal source are analyzed in both the range and Doppler domains, respectively, when impinging on multiple uniform linear arrays (ULAs) mounted on the DMS-WOTHR platforms, respectively. If the conditions are both satisfied, all the receiving arrays can be regarded as the uniform linear subarrays (ULSAs) of a distributed array. To implement the cross-platform coherent signal processing by using the array reconstruction technique to synthesize a larger receiving array, the virtual interpolated array (VIA) transform is introduced. The existing robust VIA transform techniques can cause unacceptable interpolation errors over the entire field of view because of high sparseness of the distributed array. For that, this dissertation proposes a preestimation-based VIA transform method by specifying a union of nonoverlapping narrow subsectors as the interpolated sector to cover only the source locations preestiamted roughly on an assumption that at least a single ULSA is available for localizing highly correlated sources within the range-Doppler cell. This method not only guarantees the interpolation precision in the interpolated sector but also avoids the situation of strong interference sources in the same cell impinging on the array outside the sector in multisource scenarios. In addition, it skips noise prewhitening and employs more subarrays of the virtual ULA for forward-backward spatial smoothing (FBSS) that plays a key role in noise floor reduction as well as correlated source decorrelation. Monte Carlo simulations show that the proposed method used in DMS-SWOTHR performs much better in both the azimuth resolution and estimation precision at low SNRs.