Short answer
You need an uncontended internet link, ideally dedicated fibre, with enough headroom for your busiest hour of calls. FirstNet plans for roughly 100 Kbps per concurrent call plus a 20% buffer, and checks the link before any voice deployment.
In detail
The targets FirstNet's voice engineers work to:
- One-way latency under 150 ms, jitter under 30 ms and packet loss under 1%.
- Voice traffic prioritised end to end with QoS (DSCP EF), from the LAN switches and router through to the provider handoff. Partial QoS does little.
- Bandwidth sized to peak concurrent calls, not headcount, and checked against current utilisation.
As a worked example, 20 concurrent calls at 100 Kbps is 2 Mbps, so with the 20% buffer the link needs about 2.4 Mbps of headroom for voice.
If your link is anything other than dedicated fibre, FirstNet can run a 24-hour line monitor that measures jitter, latency and loss before you commit. Where a gap shows up, the usual fix is dedicated internet access (DIA) or SD-WAN with multi-link bonding. When you also take FirstNet connectivity, FirstNet engineers QoS end to end; best-effort internet works, but call quality cannot be guaranteed.
Source: FirstNet Voice service page →
Didn’t answer your question?
