Radio behaviour behind the reading
Bluetooth Low Energy (BLE) beacons broadcast at 2.4 GHz, a frequency shared with Wi-Fi, Zigbee, microwaves and a growing number of personal devices. In an empty room, a beacon’s signal attenuates predictably as distance increases, and the received signal strength indicator (RSSI) follows a reasonably smooth curve. Add a few hundred people, and that curve breaks down.
The human body is largely water. At 2.4 GHz, water absorbs signal energy. A dense crowd does not simply block a line-of-sight path; it creates a shifting, lossy medium that attenuates signals passing through it and scatters signals around it. The result is not a uniform drop in signal strength but increased variance: the same beacon, at the same distance, might return an RSSI of -65 dBm one moment and -82 dBm a few seconds later as people move between it and the receiving phone.
This matters because proximity systems convert RSSI into estimated distance. When RSSI fluctuates unpredictably, distance estimates become unreliable. A zone trigger set at -70 dBm might fire late, early, or not at all, depending on crowd density at that instant. The problem intensifies near entry points, along narrow corridors and around popular exhibits or product displays where visitors congregate.
Device density compounds the issue. Each smartphone in a crowded venue is itself a 2.4 GHz radio, scanning, connecting and occasionally transmitting. In a retail store on a Saturday afternoon or a conference hall between sessions, the spectrum is contested. Beacons do not collide in the way Wi-Fi packets do—BLE advertising channels are narrow and transmissions are brief—but the background noise floor rises, reducing the effective range at which a receiver can reliably detect a beacon packet.
Why standard path-loss models fall short
Log-distance path-loss models assume a relatively stable environment. They work acceptably in warehouses, offices and sparse galleries. In crowded spaces, the model’s environmental factor becomes a moving target. Some deployments attempt to compensate by adjusting the factor for expected occupancy, but this trades one approximation for another because crowd distribution is rarely uniform. A measured calibration in a space filled with standing people will not hold when those people are walking, seated or clustered around a single point of interest.
A repeatable venue test
Retail environments
During peak trading hours, a supermarket aisle or fashion floor can hold dozens of people within a single beacon’s range. If the system is designed to trigger a notification when a shopper enters a three-metre zone around a display, crowd-induced RSSI drops may cause the trigger to fire only when the shopper is one metre away—or not fire at all if the signal falls below the receiver’s detection threshold. Practical responses include raising the beacon’s transmit power (at the cost of battery life), lowering the RSSI threshold for the zone, or accepting a larger, less precise zone and designing content that works at that coarser granularity.
Museums and galleries
Museum crowds tend to cluster around popular exhibits while leaving others empty. A beacon mounted above a high-traffic piece will experience severe, intermittent attenuation as groups stand directly beneath it. Meanwhile, a beacon above a quieter exhibit operates in near-open-air conditions. If the same RSSI-to-distance calibration is applied across both, the system will overestimate distance at the crowded exhibit and underestimate it at the quiet one. Zone-based triggers, where the system cares only that a device is somewhere near an exhibit rather than at a precise distance, are more robust than point-estimation approaches in this context.
Events and conferences
Conference halls present the most extreme case. Thousands of devices, dense crowds, temporary partitions and overlapping Wi-Fi access points create a hostile RF environment. Beacon deployments here often shift from precise indoor navigation to coarse zone detection: “you are in Hall A”, “you are near the registration desk”. Attempting sub-five-metre accuracy in a packed hall is usually impractical. The design question becomes whether coarse zoning delivers enough value to justify the hardware, or whether static signage and printed maps are the more honest solution for that particular use case.
Placement adjustments for crowded operation
- Mount beacons above head height where feasible. Signals travelling through empty air above the crowd reach receivers more consistently than signals that must pass through bodies at torso level.
- Avoid placing beacons where crowds naturally press against walls or fixtures. A beacon at shoulder height on a popular exhibit’s plinth will be repeatedly shadowed.
- Increase beacon density in high-traffic zones so that a receiver is likely to detect at least one beacon with a relatively clear path, even if others are blocked.
- Use wider zones and longer averaging windows in the receiving software. A one-second RSSI average smooths out momentary fades; a five-second average smooths further but adds latency to zone entry detection.
What changes invalidate the result
Calibrating in an empty venue and deploying into a full one
This is the single most common error. A site survey conducted on a Tuesday morning in an empty museum hall will produce RSSI readings that bear little relation to Saturday afternoon conditions. If the deployment relies on those empty-venue calibrations, zone triggers will behave unpredictably once real visitors arrive. Calibrate at representative occupancy, or at minimum test the system under load before committing to a full rollout.
Assuming all crowds affect signals equally
A standing audience at a keynote absorbs and scatters signal differently from a moving queue, a seated theatre crowd or a milling exhibition-floor crowd. The distribution and movement of bodies change the RF environment continuously. A calibration that works for one event format may not transfer to another in the same physical space.
Over-relying on filtering to fix a physical problem
Software filters—moving averages, median filters, Kalman filters—can smooth RSSI readings, but they cannot recover signal energy that has been absorbed. If the crowd is dense enough that the beacon’s signal is genuinely too weak to reach the receiver, no amount of filtering will help. The correct response is to change the physical deployment: move the beacon, add another beacon, raise transmit power or accept a larger zone.
Ignoring the receiver side
Not all smartphones have equal BLE sensitivity. A high-end device held in a hand above the crowd may detect a beacon that a budget phone in a jacket pocket, partially shielded by the wearer’s body, misses entirely. Any accuracy or reliability claim tied to a specific device and carrying position will not generalise across a real visitor population.
Key checks before going live in a crowded environment
- Test at representative density. If the venue regularly reaches a specific occupancy, verify zone triggers and navigation behaviour under those conditions, not just during setup.
- Compare head-height versus ceiling-mount readings. Measure the RSSI difference between a beacon at 1.5 metres and the same beacon at 3 metres in a crowd. The data will reveal whether raising the beacon materially improves consistency.
- Check the noise floor. Use a BLE scanner to observe the background noise level in the space during peak hours. If the noise floor is close to your beacon’s RSSI at the intended detection distance, the system will be unreliable.
- Verify zone triggers across multiple device types. Test with at least two different phone models, one high-end and one mid-range, carried in typical positions (hand, pocket, bag).
- Review battery-life projections against raised transmit power. If you increase power to compensate for crowd attenuation, recalculate battery replacement intervals using the manufacturer’s data for the new setting.
Crowded spaces do not make beacon deployments impossible, but they demand honest acknowledgement that the RF environment is fundamentally different from an empty room. Designing for that reality—coarser zones, higher mounts, density-aware testing—produces systems that behave predictably when it matters, rather than only during commissioning.
