Pilot-Calibrated Hankel Space–Time Waveform Recovery for Broadband LEO Satellite Interference Channels

Broadband co-channel interference in low-Earth-orbit satellite links is challenging when angular proximity and multipath jointly couple spatial and temporal interference structure. This paper proposes a semiblind pilot-calibrated Hankel space–time fusion estimator (PC–HSTF) for desired-waveform recovery. The received array sequence is embedded through an exact finite-memory Hankel lifting, after which target-conditioned space–time estimators, direction-aided spatial estimators, and a statistical-separation estimator are placed in a common desired-user reference using receiver-known preamble/pilot symbols. For the marker-trained Hankel space–time (HST) estimator, we derive a first-order finite-sample loss law for scale-invariant squared waveform correlation when the covariance and target cross-correlation are estimated from the same periodically sampled observations under colored proper Gaussian disturbance. The coefficient decomposes into pilot-spectrum-dependent cross-correlation-estimation and covariance-estimation terms after projection onto the subspace orthogonal to the population recovery direction. Controlled experiments verify the predicted $N^{-1}$ law, and communication-scene diagnostics support the stated analytical regime. Communication-scene simulations show improved interference suppression and waveform recovery in the moderate-to-high signal-to-noise ratio (SNR) regime, while a simpler minimum variance distortionless response (MVDR) estimator remains competitive at the lowest SNRs. The formulation uses only received samples, desired-user markers, and a nominal desired direction; payload truth and true channel/interferer parameters are excluded from receiver processing.