Failure-aware evaluation of sum–difference coarray direction-of-arrival estimation under array mismatch data and code

This dataset and software package accompanies the manuscript “Failure-aware evaluation of sum–difference coarray direction-of-arrival estimation under array mismatch.” The package contains the simulation results, numerical implementations, fixed parameters, and analysis outputs supporting the manuscript. The experiments examine direction-of-arrival estimation under mutual coupling, gain–phase errors, sensor-position errors, and spatially correlated noise. Performance is evaluated using execution validity, failure-penalized estimation error, tail risk, paired comparisons, and runtime measurements. The archived evidence includes: A 50-seed impairment study comparing six implementations. Cross-geometry and signal-setting experiments. An independent 500-seed validation of index-based and distance-dependent complex coupling models. Runtime measurements, numerical reference checks, and spectral peak diagnostics. The historical MATLAB R2023a case used in the manuscript to illustrate the effect of failure treatment on method rankings. MATLAB and Python source files, parameter specifications, the six manuscript figures, a data dictionary, and a manuscript-to-file index are included. Failed trials are retained to preserve the reported estimands. The data are synthetic; trial-generation code and seed specifications are provided. SHA-256 checksums and a file-integrity verification script accompany the archive. The main simulations were conducted using MATLAB R2025b. Software requirements and usage instructions are documented in the README. Historical R2023a results are identified separately from the current experiments.