Slow-time Randomly Missing Data Reconstruction for Skywave Over-the-horizon Radar

In skywave over-the-horizon radar (OTHR), the slow-time randomly missing data reconstruction is vital for the following signal processing where the data contaminated by lightning discharges or transient meteors are usually discarded and have to be reconstructed properly. The least-square auto-regressive (LS-AR) method is often used as a solver for the randomly missing problem. However, the derivations for the LS-AR method are not convincing and the performance of the LS-AR method would be degraded or even failed in some cases. In this paper, we are aimed at solving these two issues where two corresponding methods are proposed. The former method has a solid theoretical foundation and we will see that the LS-AR method could be considered as a simplified form of it. The latter method based on the LS-AR method behaves robustly for randomly missing positions compared with other types of methods by real data processing.