Radar Range Super-Resolution Using Sparse Bayesian Learning with a Pulse-Compression-Response Dictionary

Conventional range processing is fundamentally limited by signal bandwidth and therefore cannot reliably resolve multiple targets located within the same range-resolution cell. Existing sparse-reconstruction-based range super-resolution methods typically construct dictionaries from ideal delay-induced frequency responses, which do not faithfully characterize the pulse-compressed observations used in practice, thereby making these methods susceptible to model mismatch and the associated performance degradation. To address this issue, this paper proposes a pulse-compression-response-dictionary-based sparse Bayesian learning (PCD-SBL) method for radar range super-resolution. By jointly incorporating the transmitted waveform, system bandwidth, sampling rate, pulse-compression processing, and the candidate range grid, the proposed method constructs a pulse-compression-response dictionary consistent with the actual observation mechanism and formulates a sparse measurement model in the range domain. Based on this model, a hierarchical sparse Bayesian learning framework is then developed to iteratively update the sparsity-controlling hyperparameters and noise precision, enabling the estimation of multiple closely spaced target ranges within a single range-resolution cell. Simulation and measured-data results demonstrate that the proposed method alleviates the adverse effect of model mismatch and provides reliable range super-resolution for closely spaced swarm targets.