InnerWeb
Skip to main content

Implementation notes: a LoRa repeater without internet

The IT administrator sees a building at the far end of the site that neither a cable nor Wi-Fi reaches, yet people work there. A repeater gateway solves this without new hardware on the system side, but it has conditions and pitfalls. Below are the deployment steps and the places where teams most often stumble.

TTL 2 to 3
hops of a repeater
Source: spec card
2x
channel occupancy with one hop
Source: sample calculation

What the repeater variant is

Backhaul variant A is the repeater gateway: it forwards InnerWeb protocol frames to a gateway with internet access, with a hop counter (TTL) of 2 to 3 and frame deduplication, and requires no new hardware on the system side. The gateway at the far end hears the watches, forwards their frames onward by radio, and the gateway with a link makes the HTTP request.

This is the default choice when a site has its first building without full Wi-Fi coverage. Variant B, a mesh network as the aggregation link, is more expensive to deploy and is considered only at a larger scale.

Deployment steps

The order of work matters, because a mistake in the first step only shows up in the fifth.

Step 1: choose a gateway with a link

Identify the target gateway: stable power, Ethernet, PoE or LTE. All frames from the repeaters will arrive at it. Check that the target gateway hears the repeater before you mount the repeater at its location.

Step 2: mount the repeater and set the TTL

The repeater needs power, but not a link. Set the TTL to 2 or 3, depending on the number of hops to the gateway with internet access. For one hop, 2 is enough, and for two hops, 3.

Step 3: check the deduplication

When two gateways hear the watch, the same frame arrives by two routes. Deduplication means the system accepts it once. In the test, press SOS and check that there is one alarm in the panel.

Step 4: measure the link quality

From the gateway confirmations, read the RSSI and SNR for the route watch, repeater, target gateway. A weak SNR on the repeater-to-gateway segment is the most common cause of lost frames.

Step 5: SOS and inactivity tests

Test the alarm from the end building: the time, the confirmation on the watch, and the alarm appearing with the dispatcher. Test inactivity with the watch left on a table for the set time.

Channel load from 50 watches: without and with a repeater

Channel load from 50 watches: without and with a repeaterBefore and after chart: channel load rises from 9.6% without a repeater to 19.2% with one hop05101520%Without repeater: 9.6 %9.6 %Without repeaterWith one hop: 19.2 %19.2 %With one hop
Source: Sample calculation: 230 ms frame, 50 watches, 120 s period

Pitfalls that come up most often

The pitfalls come from radio physics and from the limits of the chip.

  • Two different radio protocols on one SX1262 chip cannot listen at the same time, so a dual-radio gateway means two boards or two radio modules.
  • The default mesh slot overlaps with the 869.4 to 869.65 MHz channel, so a mesh network alongside the InnerWeb protocol needs a separate frequency plan.
  • Every hop lengthens the frame's path and loads the channel. Long repeater chains increase the risk of collisions.
  • A repeater without a stable power supply becomes an intermittent data source. A battery is not a good power source for a gateway.
  • Without a gateway token, an unknown gateway can connect to the system. Set the token list from the start.

Sample calculation: channel load caused by a repeater

A 230 ms frame sent by a watch occupies the channel, and the repeater, while forwarding it, sends another one. For one frame with one hop the channel is therefore busy twice, about 460 ms. With 50 watches and a 120 s period, this gives about 19% channel load in the repeater's surroundings, while without a repeater it would be about 9.6%.

The practical conclusion: behind a repeater that serves a large group of watches, lengthen the pulse period or keep the repeater for a smaller group. The period is set centrally in the panel within 30 to 3600 s.

When a repeater is not enough

If the end building has power and GSM coverage, an LTE module on the gateway can be simpler than a repeater, because it removes a hop. If there are many such buildings and the group of watches keeps growing, it is worth discussing a mesh network as the aggregation link, or LoRaWAN with a ChirpStack network server, which is an option on request.

Video

Frequently asked questions

Does a repeater need an internet link?

No. The repeater needs power and radio coverage to the gateway with internet access. A link is needed only at the target gateway.

Will the system count a frame twice if it arrives by two routes?

No. The repeater variant includes frame deduplication, so the system accepts the frame once.

Can a gateway handle the InnerWeb protocol and mesh at the same time?

Not on one SX1262 chip. You need two boards or two radio modules and a separate frequency plan.

Let's bring up a repeater in a building without a link

We will plan the route of frames from the end building to the gateway with internet access, and the SOS tests. We invite you to a conversation with our team.

Let's talk