Idealized truth data for system modeling and testing

Modeling and testing Air Traffic Control (ATC) surveillance systems rely heavily on stimulus data. Currently, stimulus data sets are generated either from recoded sensor measurements or artificially generated scenarios. Both may contain hidden errors, contributing to the total error of the system. These hidden errors appear as jumpy, non-continuous positions, velocities, or accelerations. It is important to address the discontinuities in the higher order terms, not just position and velocity. This paper will discuss a technique used to generate idealized truth data. Idealized truth data uses the original data as a guide to generate an achievable scenario that does not contain discontinuities in position, velocity, and acceleration. Idealized truth data is not the original stimulus data filtered to perfect accuracy. The approach presented uses common mathematical techniques to eliminate discontinuities in velocity and acceleration from the original stimulus data, resulting in a scenario that could have happened with effectively no error. Though these techniques may be applicable to many systems, a case study applying these techniques to an aircraft surveillance tracker under test is provided.