The move to all-IP emergency communications is often presented as a technical migration. That framing is too small. It is also a regulatory, operational, and evidence problem.

EENA's August 2026 document listing on emergency calls in the all-IP era captures the question regulators and public-safety authorities should be asking: what fails, what do tests show, and what should regulators demand? That is exactly the right structure for NG112 planning.

All-IP emergency calling can unlock richer capabilities: multimedia, better data exchange, improved accessibility, more flexible routing, stronger logs, and cleaner integration between providers and emergency service networks. But it also removes assumptions that legacy systems quietly depended on.

What changes when voice stops being the anchor

Legacy emergency calling was not simple, but many operating models were built around relatively stable voice assumptions. The originating network knew a lot about the subscriber or line. Routing was often tied to network topology, numbering, or fixed access. Circuit-switched fallback provided a safety net in many mobile environments.

All-IP changes that surface area.

A modern emergency call may involve:

  • IMS emergency calling from a mobile network.
  • Voice over Wi-Fi behavior.
  • Roaming users on 4G or 5G networks.
  • Cloud PBX or enterprise UC clients.
  • SIP interconnect between providers.
  • PIDF-LO or other structured location objects.
  • Emergency apps and accessibility services.
  • eCall migration from older network assumptions.

That is not one migration. It is a mesh of dependencies.

Testing has to get more realistic

A lab test that proves a call can be placed is not enough. Emergency communications teams need scenario testing that reflects real failures.

A useful test programme should include:

  • Domestic mobile emergency calls over modern voice services.
  • Roaming calls from common visitor profiles.
  • Calls from cloud enterprise platforms and VoIP providers.
  • Calls with precise location, approximate location, stale location, and no location.
  • Calls using real-time text or total conversation.
  • Routing failover when a PSAP or network element is unavailable.
  • Call-back behavior after session interruption.
  • Logging and recording review after each test.

The last item is often neglected. If a team cannot reconstruct why a test call reached a destination, it will struggle to reconstruct a real incident.

Why regulators should ask for evidence

Regulators do not need to operate every network function themselves, but they do need confidence that the emergency chain works across ownership boundaries. That requires evidence from operators, providers, vendors, and emergency authorities.

Regulatory assurance should move toward questions such as:

  • Which emergency access paths have been tested end to end?
  • Which caller-location methods are accepted, validated, and displayed?
  • How are all-IP emergency calls handled while roaming?
  • What fallback exists if IMS emergency calling is unavailable?
  • Are accessibility modes tested as primary scenarios?
  • Are provider interconnects tested with realistic location and policy data?
  • How are unresolved defects tracked after testing?

These questions are uncomfortable because they reveal gaps. That is the point.

VoIP providers are in the middle

VoIP providers and cloud communications platforms are no longer peripheral to emergency calling. In enterprise environments, they may be the first domain that knows the user's account, device, policy, and candidate location. If that information is not accurate or cannot be conveyed downstream, the emergency chain starts weak.

Providers should be ready to show:

  • How customer location records are validated.
  • How nomadic or remote users are treated.
  • Whether emergency calls carry structured location where supported.
  • How calls are routed when location confidence is low.
  • What customer-facing reports or logs are available after a test.
  • How changes to SBCs, trunks, numbers, and policies are controlled.

The US E911 model has made many enterprises familiar with structured location workflows. Europe is different country by country, but the operational lesson transfers: emergency routing is only as dependable as the data and interconnect behind it.

Editorial perspective

All-IP emergency communications should be safer than legacy voice in the long run. It can carry more context and support more people. But the transition period is dangerous if organisations treat IP as a like-for-like transport swap.

It is not. It is a redesign of the emergency communications evidence model. Regulators should ask for proof. Operators should expect to provide it. PSAPs should demand logs that explain what happened. Enterprises should stop assuming their provider has solved location on their behalf.

The all-IP era will be judged by operational evidence, not architecture diagrams.

Sources