Delay Compensation-Based Time Synchronization Under Random Delays: Algorithm and Experiment

Robust Synchronized Data Acquisition for Biometric Authentication

“…maximum likelihood estimation [15], [19], linear least squares regression [20]) to estimate a more accurate clock skew for correction. However, these two methods need considerable computational overhead [21], [22], and the required time for calculating the skew estimate (i.e. processing delay) also varies in different (i.e.…”

“…5 in [15]) of timestamps reduces computing accuracy [20]. Thus, the calculation of linear least squares regression moves to the cluster head with adequate computing resources [20], or is implemented via FPGA [21].…”

“…Hence, a Proportional (P) controller [16] is utilised to solve the above issue. In [21], the moving average solution is adopted for clock skew estimation, and a Proportional-Integral (PI) controller is used in [12] for eliminating the impacts of drifting clock skew. Furthermore, the results of [24] and [25] show that applying the PI controller on clock offset adjustment can automatically remove the effects of varying processing delay, however, no skew correction input is employed to the local clock in these two works.…”

“…In addition, although the works mentioned above consider the consequences of drifting clock frequency, only the theoretical analysis of synchronisation protocols is presented (e.g. [21], [24], [25], [26]). This means that parameters (e.g.…”

“…This means that parameters (e.g. of the PI controller and moving average solutions) are determined empirically [12], [21]. In [16], the H ∞ method is used to design the P controller parameters in a pairwise network.…”