Reference topology
The four layers of a local-first LoRa switching site, the role of each layer, and the failure mode each layer is designed to hold.
| Document revision | r0.1-draft |
|---|---|
| Status | draft |
| Applies to | LS-WS-1G, LS-WS-2G, LS-WS-3G, LS-RC-16A, LS-RC-25A, LS-DIN-4CH, LS-GW-100, LS-GW-300, LS-SN-TH, LS-SN-MT |
| Regions | global |
| Published | Oct 15, 2026 |
| Review due | Apr 15, 2027 |
| Author | Engineering documentation (owner unassigned) |
| Reviewer | Engineering sign-off pending (reviewer unassigned) |
The reference topology is described as four layers with a defined responsibility. The design intent is not that the cloud is unreachable; it is that the site keeps working when it is. Each layer therefore has a stated failure mode, and the test of the topology is what still functions when the layer above it is removed.
The four layers
Layer 1 — Endpoints
Switches, relay controllers and sensors sit at the point of control and each hold their last known state locally. An endpoint that can only act on a command received from above is not an endpoint in this sense.
Layer 2 — Private gateway or edge controller
The gateway terminates the radio links and holds the schedule, rule engine and node registry on the premises. It is owned by the site and is the single device the site must protect. LS-GW-100 is the compact case; LS-GW-300 adds field protocols and a historian buffer.
Layer 3 — Local policy and failover
Schedules, scenes and alarms resolve on the edge while the uplink is unavailable. The layering matters here: if a rule can only be evaluated once data reaches the cloud, the rule is not a local rule, regardless of where it is stored. See Failover and local policy.
Layer 4 — Optional cloud management
Remote oversight, reporting and firmware distribution, when connectivity and consent allow. This layer is management-only or optional by design: outbound notifications and dashboards queue until the uplink returns.
Placement rules
- One gateway per radio neighbourhood, positioned where its antenna can see the bulk of the endpoints. Endpoints are placed for their own load first and their radio link second; the gateway is placed for radio coverage first.
- A repeater or second gateway is a coverage remedy, not a capacity remedy. Add coverage devices when the link does not close, not when the node count looks high.
- Keep metal enclosures, concrete, and fused or reinforced floors between the gateway and the endpoints to a minimum on the planned path; these are the variables that decide whether a link closes.
Placeholder — pending engineering sign-off: the reference hop count, the permitted repeater depth and the air-time budget per gateway are not published yet and depend on the SKU generation.
Limitations
- The topology does not define a maximum node count. The practical bound is air-time and the polling or reporting interval, which are configured per site and are not a fixed figure.
- A single gateway is a single point of failure for schedules and rules. Where that matters, a site must plan the gateway's power and its replacement; this page does not prescribe a redundant arrangement.
What this document does not cover: the survey figures behind a placement decision are in Coverage and site survey, and the commissioning order is in Quick start.