COVER4 The staff lane for emergency orders
Emergency orders to every staff member, confirmed by name, even when cell and Wi-Fi are down.
The operator arms and sends a signed order. Each staff member answers with one press on a tag or one tap on a phone. By the deadline, the control center shows who confirmed, who did not and why, and keeps it in a tamper-evident record.
Cover4 runs beside the venue's PA, evacuation system, fire alarm and radios. It does not replace them.
01 The problem
When an order goes out to thousands of staff, no one can prove who got it.
Security, stewards, medical and production staff are the people who carry out a hold or an evacuation. They are who Cover4 reaches first.
Staff alerting apps
Push, SMS and in-app alerts depend on cell, Wi-Fi or internet, which are the first networks to overload in a crowd or a storm. A 2026 preprint measured stadium data uploads failing up to 46% of the time on game day.
Radios
A radio reaches whoever is on the channel and listening. It keeps no record of who heard the order and who did not.
The result
At the moment it matters most, a supervisor counts heads over the radio.
- ~4,000staff on a stadium event day Estimate
- ~2,500staff across a large three-day festival Estimate
Sources: stadium uploads, arXiv 2607.16008 (preprint), checked October 30, 2026. Staff counts: Cover4 venue model, an estimate.
02 How it works
One signed order out. A by-name answer back by the deadline.
- 01
Arm and send
The operator holds STOP, HOLD, EVACUATE or SHELTER for one second to arm it, picks a zone, team or person, then presses Send. Nothing goes out on its own.
- 02
Carry
Fixed nodes around the site carry the signed order over the venue network when it is up, and over Cover4's own radio network when it is not. No cell, Wi-Fi or internet is needed.
- 03
Confirm
Each staff tag buzzes, and the phone shows the order full screen if the person carries one. One press answers: acknowledge, I'm safe, or I need help.
- 04
Count and escalate
At the deadline each person is confirmed, unreached or coverage unknown, with the reason. Everyone not confirmed escalates, and the exchange is kept in a record on site.
The site keeps
- Its PA, evacuation system and fire alarm.
- Its radios and channels, for talk.
- Its alerting platform and staff app. Cover4 can run inside that app through the SDK.
Cover4 adds the staff lane
- Orders that reach every staff tag and phone at once, targeted by zone, team or person.
- A by-name answer from each person by the deadline, with the reason for every gap.
- HELP with a typed message, placed at the nearest node.
- Orders signed at the site, so a fake STOP goes nowhere, and a tamper-evident record.
More detail on signing, encryption and the record: Security.
03 Proof so far
Measured on our bench, simulated at site scale, labelled either way.
BENCH means our own boards and test builds with one to three phones. SIMULATED means our site simulator at the worst end of every range we model. Nothing on this site is a field result.
| Stage | Date | Result |
|---|---|---|
| Bench | 2026-10-26 | One STOP, end to end, on phones never paired and with Wi-Fi off: confirmed by name in 3.1–4.6 s, tap included. |
| Bench | 2026-10-29 | A relay died right after taking a reply. Another relay delivered it: confirmed by name at 16 s. |
| Bench | 2026-10-29 | The primary control center was killed mid-alert. The backup took over in 15 s and re-sent the order to the 5 people not yet confirmed. |
| Bench | 2026-10-29 | A bench jammer was flagged within seconds, at 73–79 dB over the noise floor. |
| Bench | 2026-10-30 | A relay was jammed until it stopped relaying. The reply went around it: confirmed by name in 2 s. All nodes hopped across 16 channels in step. |
| Bench | 2026-10-30 | HELP with a typed text, from a phone never paired and with Wi-Fi off, reached the operator, and the operator's reply came back. |
| Bench | 2026-11-01 | With the phone's Bluetooth off, a staff tag alerted on its own, and its signed confirmation was counted. |
| Bench | 2026-11-03 | Staff replies in assigned time slots: 30 of 30 heard and acknowledged, two boards. A protocol check, not a range test. |
| Simulated | 2026-11-05 | Festival, Grant Park footprint: 100,000 people, 2,500 staff each with a tag, phones off, every link 20 dB worse than plan. 0 of 24,232 taps uncounted over 10 runs; 99% reached the control center within 7.1 s. |
| Simulated | 2026-11-05 | Stadium, 67,594 seats, 4,000 staff: 0 of 38,785 taps uncounted over 10 runs, including with any one receiver cluster lost 30 s after the STOP; 99% within 10.8 s. |
| Simulated | 2026-11-05 | 110 NFL, NBA, MLB and NHL venues, any one receiver cluster knocked out 30 s after the STOP, 3 runs per venue: 83 counted every tap with 3 clusters and 27 needed a fourth. |
| Not yet shown | open | Any field test or venue walk. A reply reaching a jammed gateway over the 2.4 GHz fast lane (fix written, re-test pending). More than three real phones at once. |
04 Who it is for
Sites where thousands of staff must act on one order.
Each page below is written for that operator: the problem on their floor, the plan for their kind of site, and what they keep.
05 Next step
Start with a site walk and one pilot.
Tell us the site and the orders that matter: holds, evacuations, roll calls and help requests. We reply with a technical call. The first pilot is the field test, judged on the by-name record it produces.
partners@getcover4.com · Chicago, Illinois