A phone call connects two people. Push-to-talk (PTT) connects one talker to an entire group — a channel, a talkgroup, a fleet — simultaneously, and it must do so the instant someone presses a button. That single design requirement reshapes almost everything: signalling, latency, spectrum use and what happens when the network fails.
The defining metric is call setup time, often called "access time" or "PTT latch time". In consumer voice, a second or two of setup delay is unremarkable. In emergency communications, a firefighter needs the channel open before the sentence is finished. TETRA, the digital trunked-radio standard developed under ETSI, specifies access times under half a second; analogue land-mobile radio can be faster still, because there is no digital handshake at all — press the button, the radio transmits. Modern critical networks built for reliability rather than throughput treat this latency budget as a hard constraint, not a quality-of-service preference.
01The Architecture Behind Instant Group Audio
Traditional PTT operates on a shared channel: everyone in the group listens to the same frequency; whoever presses transmit captures it. No server, no routing, no round-trip to infrastructure. That simplicity is also resilience — if the repeater or base station disappears, radios can switch to direct mode and continue talking peer-to-peer.
Cellular PTT — whether 3GPP's Mission Critical Push-to-Talk (MCPTT) over LTE and 5G NR, or earlier proprietary "push-to-talk over cellular" services — works differently. The voice path routes through a server, which means infrastructure dependency, and the latency budget includes an LTE/NR uplink, a dispatch server, and a downlink multicast or broadcast. 3GPP addressed this in its Mission Critical services specifications, defining floor control (the mechanism that grants one talker exclusive access), late entry (joining a call already in progress), and group affiliation (attaching a subscriber to multiple logical groups without requiring multiple physical channels). Getting floor control right is subtle: the server must prevent simultaneous talkers, handle contention gracefully and release the floor quickly when the talker finishes.
Broadcast efficiency matters too. Delivering the same audio to hundreds of subscribers as individual unicast streams wastes capacity; eMBMS (evolved Multimedia Broadcast Multicast Service) and its 5G successors allow a single downlink transmission to serve the whole group at once, but coordinating unicast and broadcast paths dynamically adds its own complexity.
The defining metric is call setup time, often called "access time" or "PTT latch time".
From this piece
The underlying tension in every PTT system is the same: group voice demands broadcast-like efficiency, near-zero setup latency and survivability without infrastructure — requirements that pull in different directions depending on whether the air interface is a dedicated narrowband channel or a shared broadband cell.