Reconstruction of the execution times dynamics of real-time tasks by fuzzy digital filtering

Real-time systems (RTS) process their activities through tasks which in turn consist of a set of instances, each one of this real-time task (RTT) have six temporary constraints: arrival time, start time, execution time, end time, over time and deadline. Of these six constraints, the execution time depends directly on the software and hardware of the computer, because of that this temporary constraints varies due to external and internal factors, this variation may cause at least some instance does not satisfy the deadline, in this sense it is important to propose a model to reconstruct the behavior of the execution times for the purpose to determine the proportions of the Real-time System, propose appropriate techniques of fault tolerance and improve operating modules. Therefore in this paper, is proposed a model to reconstruct the execution times from measurements made at a program algorithm developed in the real-time operating system QNX Neutrino 6.5. This reconstruction is done by a model type autoregressive–moving-average integrated with parameter estimator constructed with a digital filtering procedure diffuse (FDD). To validate the reconstruction is used the mean square error which converges to a region close to zero and indicate that the reconstruction is good.