Krylov Subspace-Based Clutter Channel Impulse Response Estimation in Cognitive Fully Adaptive Radar

A Cognitive Fully Adaptive Radar (CoFAR) operates on the principles of sense, learn, and adapt to its environment. This process requires a robust understanding of the surroundings to optimize radar performance. Enhanced environmental knowledge improves CoFAR's effectiveness, yet it is crucial to use minimum resource for acquiring this knowledge such that the primary radar surveillance functions will not be impacted. This study introduces a Krylov subspace-based reduced-rank filtering approach for efficient Clutter Channel Impulse Response (CCIR) estimation in CoFAR. Unlike conventional methods, which are computationally demanding, our approach leverages Krylov subspaces to focus on the most informative subspace, enabling accurate, resource-efficient estimation while avoiding full-dimensional processing. By iteratively applying the conjugate gradient algorithm, the proposed method converges to near-optimal (clairvoyant) performance in only a few iterations, reducing computational overhead. This adaptation makes the proposed algorithm ideal for a CoFAR. Computational complexity analysis, comparing the Krylov subspace-based approach with closed-form least square, QR, and singular value decomposition-based methods, demonstrates that the proposed algorithm achieves the lowest complexity for CCIRE. Furthermore, validation against representative clutter data from RFView modeling and simulation confirms the analytical advantages, demonstrating the efficacy of our approach for real-time, adaptive radar environments.