MM8108 Link quality optimization

Software
1

Hi everyone,

I’m looking for advice on the best MM8108 mesh configuration for a small deployment and would really appreciate some tuning guidance.

Use case / requirements

  • Topology: mesh, 3–5 devices

  • Environment: open space, clear line of sight

  • Minimum throughput target: 3 Mbps per link

  • Hardware: VT-USB-AH-8108 + Raspberry Pi CM4 (host)

  • OS: OpenWrt on the Pi CM4

Questions

  1. For this scenario, which MAC-layer settings are most important to tune on MM8108 to reliably reach ≥3 Mbps in an open-space mesh (e.g. MCS selection, aggregation, contention parameters, beacon/DTIM, etc.)?

  2. How can I select or change the rate control algorithm (or its parameters) for MM8108 in a mesh setup? Is there a recommended rate control scheme for a small number of nodes with good SNR and line of sight?

  3. On the OpenWrt side, what should I pay the most attention to:

    • mesh vs AP+STA mode choices,

    • channel width and frequency selection for Wi‑Fi HaLow,

    • queue/tx power/driver options that have the biggest impact on throughput and latency?

  4. Is it advisable to use the “thin-mac” firmware for this use case (VT-USB-AH-8108 + CM4 + mesh, 3–5 nodes), or is it better to stick with the full MAC firmware? What are the key trade-offs in terms of performance, flexibility, and complexity?

  5. If anyone has a reference configuration (OpenWrt version, Morse Micro driver/firmware version, mesh settings, and typical throughput you observed) for a similar open-field test with MM8108, could you share it?

Any configuration snippets, example wireless/network files, or high-level guidelines would be extremely helpful. Thanks in advance!

V.

2

I would avoid forcing MCS manually - you are more likely to experience increased packet loss by doing this. Depending on your range, you may want to consider reducing bandwidth from 8 MHz to 4 MHz. The receiver sensitivity will improve slightly by doing so, but at the cost of maximum throughput.

The most important factor for throughput will be your mesh configuration. See below.

The only supported rate control algorithm is “MMRC”, included in our driver.

Pay attention to your choice in mesh. This will depend on the topology you desire, and I can’t make a concrete recommendation as it will depend on your intended architecture.

802.11s Mesh is likely to be more flexible, as it is a true tree structure. We use HWMP in the kernel with mesh11sd on our OpenWrt fork to manage this. As this is what we test, if you’re using 802.11s Mesh, this is what we recommend.

With 802.11s mesh, you will halve your throughput for every hope your traffic needs to traverse! For 3-5 nodes this is likely to be fine for your use case. For larger deployments this might be problematic.

There is also another community who have been making use of B.A.T.M.A.N on with 802.11s mesh modes instead. Have a look at OpenMANET · GitHub for their development. They appear to be having success with this deployment. We are supportive of this community effort but do not test it ourselves :slight_smile:

If you can follow more of a “tree” style topology, EasyMesh might be an option. We also test this on our OpenWrt distribution. This enforces a more rigid mesh topology, but enables higher throughput.

No this will give you no benefit

No, full mac software does not support meshing. You want the standard, soft MAC firmware (mm8108b2-rl.bin).

See User Guides here:

And if you are using our OpenWrt fork, we have wizards available to help you configure your mesh networks :slight_smile:

3

Dear @ajudge

A huge thank to you for a very comprehensive answer.

In case I would like to try modifyding the BCF for deeper experiements, what is your suggestion?

Thanks,

V.

4

Unfortunately we do not release the tools for modifying BCFs.