Timing estimation in distributed sensor and control systems with central processing

TL;DR

This paper addresses direct timing estimation in distributed sensor and control systems without relying on clock synchronization, proposing models and heuristics to reduce jitter and improve timing accuracy.

Contribution

It introduces models for timing and timestamping, and presents heuristics for system time construction that enhance timing estimation accuracy in distributed systems.

Findings

Including peripheral unit timestamps reduces jitter.

The proposed models improve timing estimation accuracy.

Numerical example demonstrates practical benefits.

Abstract

We consider the problem of estimating timing of measurements and actuation in distributed sensor and control systems with central processing. The focus is on direct timing estimation for scenarios where clock synchronization is not feasible or desirable. Models of the timing and central and peripheral time stamps are motivated and derived from underlying clock and communication delay definitions and models. Heuristics for constructing a system time is presented and it is outlined how the joint timing and the plant state estimation can be handled. For a simple set of underlying clock and communication delay models, inclusion of peripheral unit time stamps is shown to reduce jitter, and it is argued that in general it will give significant jitter reduction. Finally, a numerical example is given of a contemporary system design.