A new delay jitter smoothing algorithm based on Pareto distribution in Cyber-Physical Systems

Abstract

Delay jitter, which can affect the performance of controlled physical system and result in system instability, is an important issue for supporting instantaneity in Cyber-Physical Systems (CPSs) to guarantee continuous, real-time and dependable operation. Thus, it is significant to establish an algorithm to smoothing the delay jitter. In this paper, first, we compare several delay jitter smoothing algorithms, discuss their merits and demerits, and then propose one new algorithm over the existing algorithms, called derivative least square (DLS), which can eliminates the reverse way delay variation spikes. Second, we address the problem of most delay jitter smoothing schemes, namely they used uniformly distributed random numbers to simulate network delay data, which deviates from real network delay. To demonstrate the validity and reliability of DLS, except uniform distribution, Pareto distribution, of which the cumulative distribution function (CDF) is more close to the CDF of real network delay, is applied to conduct the experiment. The experimental results show that compared with the other three existing algorithms, DLS can be more effective and reliable in predicting the play-back delay.