IoT Edge Offloading: Latency vs. Cloud Simulator

Offloading Policy

Sensor packet rate 4.0 / s

Cloud offload fraction 35%

Cloud network latency (one-way) 80 ms

Edge server capacity 5.0 / s

Live Metrics

Avg. total latency

0 ms

Edge queue depth

Edge utilization ρ

Energy rate

0 mJ/s

Sensor / IoT device Packet routed to edge Packet routed to cloud Edge compute node

How it works

Each IoT device generates data packets as a Poisson process with rate λ (packet rate slider). A fraction p of packets (offload slider) is routed over the network to the cloud; the rest are processed on the local edge server, which has a fixed service capacity μ (packets/second).

The edge server behaves as an M/D/1 queue — Poisson arrivals, deterministic service time 1/μ, single server, first-in-first-out:

λ_edge = λ · (1 − p)
service time  Ts = 1 / μ
utilization   ρ = λ_edge / μ
mean wait     Wq = ρ · Ts / (2(1 − ρ))   (Pollaczek–Khinchine, D-served)
total edge latency = Wq + Ts

Cloud-routed packets skip the local queue but pay a network round trip: total cloud latency ≈ 2 × one-way latency + a small fixed cloud processing time (the cloud is modelled as effectively unbounded capacity). Each packet also carries an energy cost — edge processing burns compute energy locally, while cloud offloading burns extra radio-transmission energy to send the payload out. This is the real trade-off edge computing exists to manage: keep ρ below 1 and latency stays low and bounded; push too much load onto a capacity-limited edge server (ρ → 1) and queueing delay explodes, so some of that traffic is better offloaded — at the cost of network latency and transmit energy.

Packet rate — how fast every sensor together generates data.
Cloud offload fraction — the routing policy: what share of packets bypass the edge server entirely.
Cloud latency — the network round-trip cost of reaching a remote data-center.
Edge capacity — how much compute the local edge server actually has; watch the queue and latency climb as load approaches it.

Edge node Cloud node

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.

IoT Edge Computing: Offloading & Latency Simulator