Constellation Range • BLE inference dashboard

Presence, attention, and inferred people in the signal field.

This page lets you switch between the raw BLE story and a people-based estimate. The hard layer shows detected device identities and signal behavior; the inference layer applies a devices-per-person assumption to estimate how many people were represented within the installation’s BLE field, from close viewers to peripheral passersby.

Event Window
View Mode

What This Is Showing

This dashboard blends the measured BLE signal with a transparent inference layer so the exact data and the story you want to tell can live side by side.

1. Detect devices

The scanner records BLE device identities, signal strength, entry/exit events, and near-field proximity events. This is the hard measurement layer.

2. Infer engagement

RSSI, dwell duration, and proximity events are combined to classify each device as brief, nearby, active, passive, or likely fringe traffic.

3. Estimate people

In inferred mode, device counts are divided by your selected average devices-per-person assumption to produce a rough people-in-field estimate, not a literal headcount.

Quick Read

Key Read

Presence and Engagement by Hour

Unique detected devices show reach; engaged devices show active orbit; active-engaged devices are the strongest proxy for intentional attention.

Unique devices
Engaged devices
Active engaged

Engagement Mix

A rough participation model built from RSSI, proximity events, and dwell.

Visibility Duration

Percentiles show where the longest dwell sessions begin, not the average stay.

Signal Strength Profile

RSSI is treated as relative closeness only. Tighter distributions suggest a more coherent event field. Near a window, weaker bands can include people outside looking in.

Very weak
Weak
Mid
Strong
Very strong

Rough Distance Read

These are directional ranges, not exact measurements. Glass, bodies, reflections, and device hardware all affect BLE strength.

−56 to −51 dBm

Very close. This is the strongest near-field band, often roughly within about 1 to 3 metres of the installation.

−60 to −57 dBm

Close presence. A good proxy for people in the installation’s immediate orbit, roughly around 2 to 5 metres.

−63 to −61 dBm

Mid-near field. Likely nearby in-room presence or viewers a bit farther back, roughly around 4 to 8 metres.

−66 to −64 dBm

Edge-of-field presence. This can include people farther from the work, roughly around 7 to 12 metres.

≤ −67 dBm

Weak fringe detections. These are the best candidates for passersby, window bleed, or peripheral presence beyond the main viewing zone.

Important caveat

Read this dashboard as people represented in the BLE field, from close viewers to peripheral passersby, not as a literal attendance count.

Hourly Table

Use this to map the event rhythm: setup and ambient flow early, dinner peak, afterparty rebound, and late-night taper.

Hour Unique devices Engaged Active engaged Avg RSSI RSSI p10 RSSI p50 RSSI p90

Interpretation Notes

Data source: messages-cr_ble-2026-05-30 (1).csv. Noise filter removed 30 high-chatter devices. The anonymous scanner IDs were treated as scanner-session segments rather than separate physical nodes.