Skip to content
NS/ Sulakvelidze

SimTwin

A discrete-event simulator for 5G/LTE telemetry, built to answer one question: what does a digital twin look like when the network feeding it is lossy, delayed, and shared with everything else on the cell?

The hard problem of a digital twin is not building the twin. It is knowing what the twin believes about the plant, and what happens when the channel feeding it is lossy, delayed and shared with everything else on the cell. SimTwin measures that gap.

The simulator does four things at once: it builds a large fleet, injects faults, impairs the network — loss, latency, duplication, outage — and maintains the twin from arrived traffic alone. Then it asks whether the injected fault is recoverable from telemetry alone, and at what distance.

The simulator on this page runs in the browser, on a worker thread. The numbers it prints describe the mechanism, not the paper’s measurements — those are listed separately, with their source.

1 000

printers simulated in one fleet

70 363

messages per second through the broker

42

tests, 5 of them against a live Mosquitto broker

10 / 11

fault families separable at |d| ≥ 1.45

The four figures above are the paper’s measurements, read off the manuscript. What runs below is the browser port, and it shows the mechanism.

Second product view of the G3D Starter desktop fused-filament 3D printer.

PL.01

The G3D Starter — a desktop fused-filament printer, built in Tbilisi.

The simulator models a fleet of fused-filament printers; this is one of the production platforms it models — the plant, not the twin.

idle

The twin, running

This is the simulator from the paper, ported to the browser. Set a fleet, impair the channel, inject a fault into the plant, and watch what the twin is left believing. The numbers are the mechanism, not the paper’s measured figures.

48printers
8%
4t
2%
Replay QoS
Injected fault