Airborne radar has the superior ability of battlefield monitoring and early warningagainst low-altitude penetration target, and thus, has quite important militaryapplications. The radar beam of the airborne platform is generally pointed to the groundfor detecting slow moving target and the radar echo contains many ground scatterings.The Doppler spectrum of the radar echo spreads seriously because each clutter scattererhas a different radial velocity against the moving radar platform. In this case, thepotential moving target is submerged in the clutter spread environment and can not bedetected. Space-time adaptive processing combines muti-arrays and multi-pulses toprovide the joint processing capacity, which can effectively suppress the clutter andmake the slow moving target visible from the surrounding clutter environment. With thefast development of modern airborne radar system, common STAP algorithms face withmany challenges, including fast-changing scattering properties, discreteclutter/interference, advanced array configurations, unknown gain/phase error betweenarray sensors and etc. We combine these nonideal factors and call them heterogeneousclutter environment. This dissertation is focused on how to effectively suppress theheterogeneous clutter and has several achievements as follows.First, focused on slow convergence rate of clutter covariance matrix (CCM)estimation in the common algorithms, we propose SR-STAP approach using single ormultiple stationary snapshots. It provides high-resolution space-time spectrumestimation only with a few snapshots. With the equivalence between CCM and thespace-time spectrum, accurate CCM estimation is available so that the convergence rateof the CCM estimation is effectively accelerated. The Mountaintop real data is used toillustrate the effectiveness. The simulated experiments are also given to demonstratethat SR-STAP is more robust than Knowledge-Based method in the case of priorknowledge mismatch. Due to these advantages, SR-STAP has great potential for theSTAP application in actual fast-changing clutter scenarios.Second, focused on discrete clutter, D3SR is proposed. It does not need the trainingcells and can effectively suppress the discrete clutter/interference only with the test cell.The mismatch of clutter characteristics from the training cells is totally avoided and D3SR can effectively improve the output SCR in the test cell. D3SR maintains the fullsystem DOF so that it can achieve better performance of output SCR and MDV thancurrent D3method.Third, focused on the nonlinear and range-variant relationship between the angleand Doppler frequency of the clutter in the advanced airborne array configurations,SR-RBC is proposed. It effectively compensates the stationarity between different rangecells and solves the problem of range-variant clutter. Compared with common RBC,SR-RBC can estimate high-resolution clutter distribution with full system DOF andprovide better stationarity in the training cells. In addition, SR-RBC can be alsoextended to deal with other advanced array configurations such as bistatic array, and ithas great potential in the future airborne radar system.Finally, focused on the nonideal factors, i.e., unknown gain/phase error betweenarray sensors, the adaptive sparse recovery algorithm is proposed. It exploits the usefulinformation with stationary snapshots, and thus, adjusts the array error matrix to matchbetter with the actual array environment. The DOA performance is greatly improvedand behaves robust. In addition, since it does not need the statistical information butseeks the sparse solution of underdetermined equation to obtain the high-resolutionestimation, it can also effectively distinguish the coherent signal sources. |