Zero Entropy Networks All articles
Audit & Assurance

The Alarm Has an IP Address: Auditing Life-Safety Endpoints Before the Copper Retires

Zero Entropy Networks
The Alarm Has an IP Address: Auditing Life-Safety Endpoints Before the Copper Retires

Photo: Brian Robert Marshall , CC BY-SA 2.0, via Wikimedia Commons

The copper retirement programme is often described as a telephony migration. That description is too narrow. For many UK organisations, it is also a life-safety migration. Lift entrapment alarms, fire alarm signalling, telecare base units, lone-worker systems and emergency help points are being moved from analogue circuits onto IP, SIP or cellular paths. The change is rarely led by the zero-trust team, yet the resulting endpoints land inside the estate that the zero-trust team is expected to assure.

That inheritance deserves an audit before the copper goes, not after the first silent failure.

Start with the circuit, not the dashboard

A zero-trust dashboard will not usually show a lift phone as a life-safety device. It may appear as an unknown MAC address, a SIP user agent, a DHCP lease, or a line item on a facilities invoice. The audit therefore begins with records that sit outside the network team: maintenance contracts, fire risk assessments, alarm receiving centre schedules, lift service reports, telecare provider lists, building management inventories and carrier circuit billing.

Build a register with enough fields to be useful under pressure: device type, location, legal or contractual function, carrier, current technology, power source, backup duration, network path, alarm receiving centre, test frequency, responsible owner and migration date. Include devices that are still analogue. The point is not to create a tidy asset list; it is to expose the estate that will shortly change shape.

Map the dependencies that determine whether help arrives

Life-safety endpoints fail for ordinary reasons. A switch reboots. A DHCP scope is exhausted. A certificate expires. A broadband circuit loses power. A firewall rule is tightened. A SIP registrar rejects an unregistered device. In a conventional IT service, those events produce a ticket. In a life-safety service, they may produce a failure to summon help.

The dependency map should therefore be explicit about power, network and call routing. Record whether the device is powered by mains, PoE, a local battery, a UPS or a generator, and how long that backup has been tested. Record the first hop, the VLAN, the address method, the DNS resolver, the time source and the egress path to the alarm receiving centre or SIP provider. Where a cellular backup exists, verify that it is active, registered and included in testing rather than merely present on a quotation.

This is also the point to distinguish between devices that initiate calls and devices that depend on inbound calls. A lift alarm may dial out; a telecare hub may need both. A fire panel may use a dedicated signalling path that is not interchangeable with a voice line. Treating them as one category is how migrations break.

Segment without silencing the emergency call

Zero-trust principles apply, but they must be applied with a fail-safe bias. A lift phone cannot present a hardware token. A fire panel will not complete an interactive identity challenge. A telecare pendant cannot wait for a conditional access policy to evaluate. These devices need network placement and policy that assume the identity layer may be unavailable at the moment of need.

A practical pattern is a dedicated life-safety segment with a narrow egress allow-list to named alarm receiving centres and SIP providers, no default route to the corporate estate, and logging that preserves evidence without depending on the device to authenticate to a management plane. Device certificates or MAC-based admission can be used where the vendor supports them, but they should not become a precondition for emergency calling. Any exception to normal zero-trust policy should be documented, time-bound and reviewed alongside the device register.

The objective is not to exempt life-safety devices from assurance. It is to assure them against a different failure model: one in which availability of the emergency path outranks the convenience of central policy enforcement.

Test the failure modes that ordinary policy ignores

Audit evidence should include controlled failure testing agreed with the responsible maintenance provider and, where relevant, the alarm receiving centre. Test mains loss and battery runtime. Test broadband loss and cellular failover. Test SIP registrar unavailability, DNS failure and certificate expiry. Test a switch reboot and a firewall failover. Confirm that the alarm receiving centre receives the correct location and device identifier, not merely that a call arrives.

A heartbeat on a monitoring console is useful, but it is not proof that help can be summoned. The audit should ask for the last end-to-end test result, the date of the next test, and the name of the person accountable for acting on a failed result. Where testing is impractical, record the reason and the compensating control rather than leaving a blank field.

Make the register a living control

Copper retirement has a timetable, and the estate will continue to change after it. New lifts, refurbished fire panels, temporary buildings and changed telecare contracts can all introduce endpoints that nobody adds to the zero-trust diagram. The register should be reviewed on a fixed cadence, linked to change management and referenced in procurement. Contracts for life-safety maintenance should require notification before a device is migrated, replaced or reconfigured.

The audit finding is rarely that life-safety devices are insecure in a conventional sense. It is that they are invisible, unowned or tested against the wrong assumptions. In 2026, as copper paths are withdrawn, that invisibility becomes a resilience risk with a very short fuse. The work is unglamorous: find the circuits, map the dependencies, segment with care and prove the call still gets through.

All Articles

Related Articles

The Desk Remembers: Auditing Residue Between Shifts in a Shared Zero-Trust Estate

The Desk Remembers: Auditing Residue Between Shifts in a Shared Zero-Trust Estate

Presence Is Not Proof: Reconciling Door Events With Network Identity

Presence Is Not Proof: Reconciling Door Events With Network Identity

The Consent Layer Beneath the Cable: Auditing Landlord-Controlled Paths in a UK Zero-Trust Estate