A payment terminal is a security product that happens to take payments. The transaction is the easy part. Tamper response, key management, connectivity that survives a bad day and a certification cycle measured in months are what decide whether the product ships.
| Transport | Suits | Watch for |
|---|---|---|
| Ethernet | Fixed lanes and counters with existing cabling | Merchant network changes nobody tells you about; segmentation requirements |
| Wi-Fi | Mobile terminals within a venue | Roaming behaviour between access points mid-transaction, and shared networks the merchant controls |
| Cellular | Mobile, outdoor, pop-up and merchants without fixed connectivity | Coverage inside buildings, and per-device data cost across a large estate |
| Bluetooth to a host | Terminals paired to a till or handset | Pairing state after a reboot, and where the security boundary actually sits |
Most estates use more than one, and the difficult engineering is consistent behaviour across all of them rather than any single transport.
A terminal is installed by someone whose job is retail, on a network the merchant changed last week, in a location chosen for the counter rather than for signal. It is then expected to work every time, because a failed payment is a lost sale and an unhappy queue.
That shapes what matters. Clear diagnostics the merchant can act on, a fallback that does not require a technician, and remote visibility so support can triage without a site visit are worth more than any performance figure.
The estate management layer is the product as much as the terminal is. Provisioning, key rotation, firmware update, configuration and health monitoring across thousands of devices decide the total cost of ownership, and they are frequently specified last.
Active tamper response — the device must detect intrusion and erase its keys on battery, unpowered, in storage — plus controlled key injection, and a security certification cycle measured in months where any change to the secure boundary can require re-evaluation.
Usually more than one. Ethernet for fixed lanes, Wi-Fi for mobile terminals in a venue, cellular where there is no fixed connectivity, Bluetooth for terminals paired to a till. The engineering difficulty is consistent behaviour across all of them rather than any single transport.
Because security evaluation takes months and changes touching the secure boundary can require re-evaluation. Planning the certification route at architecture rather than at the end is the highest-return scheduling decision in this category.
The estate management layer rather than the unit price. Provisioning, key rotation, firmware update, configuration and health monitoring across thousands of devices in the field determine support cost, and they are frequently specified last.
It must behave predictably: defined offline handling, deferred authorisation where the scheme permits, and clear operator feedback. A terminal that becomes ambiguous when the network drops causes more damage than one that simply declines.
Written for engineers. Share it with one.
Concept to a supportable fleet.
SOLUTIONHardware-anchored keys.
INDUSTRYEstate scale, not store scale.
TECHNOLOGYWhere the key boundary sits.
PRODUCTEstate operation.
CASE STUDYSecure silicon in production.
Tell us the specification, the constraint and the deadline. Programmes that cross silicon, radio, embedded and AI are where Faststream is strongest.