Large data volume and computation burden are the main characteristics of real-time ultra-wide band synthetic aperture Radar (UWB SAR) processing. The nonlinear chirp scaling (NCS) algorithm adopted in the real-time processing needs large data volume. In order to reduce the data volume demand, an improved NCS algorithm is proposed. Meanwhile, the flow of Real-time UWB SAR processing is given, and the effect of forward velocity error is also analyzed, and it can be removed by real-time pulse repeat frequency (PRF) adjusting. The performance of loss of sight (LOS) compensation in improved NCS algorithm based on the data measured is analyzed. Finally, the performances of the algorithm are verified by the simulation and real UWB SAR data