A Robust Tensor Decomposition Model for Traffic Data Imputation with Capped Frobenius Norm in Smart City

Data missing is a frequent issue in smart city systems, resulting in poor accuracy and reliability in related applications. Traditional models for traffic data imputation are often under the assumption of outlier-free data, limiting their effectiveness in real-world scenarios with outliers. In this work, we devise a robust tensor completion method for traffic data imputation (STTC-CF) based on tensor ring decomposition and Capped Frobenius norm to enhance robustness against missing data and outliers. Subsequently, the half-quadratic (HQ) optimization technique is utilized to transform the original problem into a tractable form. The solution to this reformulated problem is attained through alternating optimization combined with the alternating direction multiplier method (AO-ADMM). Extensive testing on three real-world traffic datasets demonstrates that our proposed method surpasses several state-of-the-art algorithms in traffic data imputation accuracy across various simulated scenarios. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.