IoT Data Analytics Simulator

IoT Data Stream

Connected devices 8

Window size 20 samples

Anomaly sensitivity 2.5σ

Emission rate 1.0×

Moving average

–

Std deviation

–

Anomalies flagged

Throughput

– pkt/s

How it works

Every connected device emits a noisy sensor reading on its own schedule; each reading travels to the central aggregator as a packet. The aggregator keeps a sliding window of the last N readings across all devices and continuously recomputes the window's mean and standard deviation. A fresh reading is scored against that rolling baseline; if it lands too many standard deviations away, it is flagged as an anomaly instead of being folded quietly into the trend.

mean = (1/N) · Σ xᵢ          (last N samples)
std  = sqrt( (1/N) · Σ (xᵢ − mean)² )
z    = (x_new − mean) / std
anomaly  if |z| > threshold

Connected devices — how many independent sensors feed the stream; more devices raise throughput and mix more baselines together.
Window size — how many recent samples the moving average and anomaly detector look back over; a small window reacts fast but is noisy, a large window is stable but slow to adapt.
Anomaly sensitivity — the z-score threshold; lower values flag more readings as outliers, higher values only catch extreme spikes.
Emission rate — how often each device sends a new reading, controlling the overall data velocity of the stream.

Real-world relevance: this rolling-window aggregation and z-score check is the same core pattern behind production IoT telemetry pipelines — cheap enough to run per-message at the edge, yet effective at catching failing sensors, spoofed readings or sudden equipment faults before they reach a dashboard.

Normal reading Flagged anomaly Moving average

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