One-way delay variations adversely affect the performance of several real-time applications that require accurate prediction of packet delivery times. This thesis study the random behavior of one-way delay variations based on extensive real-time experiments that we have carried over the Internet using fifty strategically located wired and wireless hosts. Specifically, in this thesis we examine whether forward and reverse one-way delay variations can be adequately represented by specific distribution functions. Then, we explore the degree of dependency in both forward and reverse one-way delay variations by investigating their autocorrelation and cross-correlation.
Furthermore, we propose a scheme that uses several one-way delay variation models
to smooth out delay jitter in Cyber-Physical Systems (CPSs) with no clock synchronization.