Scaling tweaks by NomDeTom · Pull Request #9653 · meshtastic/firmware

NomDeTom · 2026-02-15T01:06:26Z

@Jorropo did the thing. Have a look.

🤝 Attestations

I have tested that my proposed changes behave as described.
I have tested that my proposed changes do not cause any obvious regressions on the following devices:
- Heltec (Lora32) V3
- LilyGo T-Deck
- LilyGo T-Beam
- RAK WisBlock 4631
- Seeed Studio T-1000E tracker card
- Other (please specify below)

… scaling behavior

…in different configurations

Copilot

Pull request overview

This pull request attempts to improve the congestion scaling algorithm in Meshtastic by replacing hardcoded throttling factors with a dynamic formula based on LoRa spread factor and bandwidth. The goal is to make the scaling more responsive to actual radio parameters rather than using preset-specific constants.

Changes:

Modified Default::congestionScalingCoefficient() to calculate throttling factors using pow(2, spread_factor) / (bandwidth * 100000) instead of hardcoded values
Added unit tests in test/test_default/test_main.cpp to verify scaling behavior for different device roles and presets
Updated comments to document the new formula values compared to original hardcoded factors

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 8 comments.

File	Description
src/mesh/Default.h	Replaced hardcoded throttling factors (0.075, 0.04, 0.02, 0.01) with formula-based calculation using spread factor and bandwidth
test/test_default/test_main.cpp	Added new test file with 4 test cases for router, client below threshold, client with default scaling, and client with MEDIUM_FAST preset

Copilot · 2026-02-15T01:13:21Z

test/test_default/test_main.cpp

+void test_client_default_preset_scaling()
+{
+    config.device.role = meshtastic_Config_DeviceConfig_Role_CLIENT;
+    config.lora.use_preset = false;
+    // nodesOverForty = 10 -> coefficient = 1 + 10 * 0.075 = 1.75 => 60000 * 1.75 = 105000
+    uint32_t res = Default::getConfiguredOrDefaultMsScaled(0, 60, 50);
+    TEST_ASSERT_EQUAL_UINT32(105000U, res);


This test doesn't set config.lora.bandwidth or config.lora.spread_factor when using use_preset = false. The new formula in Default.h directly accesses these fields, which will be 0 after initializeTestEnvironment() initializes config, leading to division by zero.

The test should explicitly set these fields, for example:

config.lora.bandwidth = 250; // 250 kHz config.lora.spread_factor = 11; // LONG_FAST default

However, note that the formula in Default.h has bugs that need to be fixed first (see comments on Default.h).

src/mesh/Default.h

Copilot · 2026-02-15T01:13:22Z

test/test_default/test_main.cpp

+// Unit tests for Default::getConfiguredOrDefaultMsScaled
+#include "Default.h"
+#include "TestUtil.h"
+#include <unity.h>
+
+void test_router_no_scaling()
+{
+    config.device.role = meshtastic_Config_DeviceConfig_Role_ROUTER;
+    uint32_t res = Default::getConfiguredOrDefaultMsScaled(0, 60, 100);
+    TEST_ASSERT_EQUAL_UINT32(60000U, res);
+}
+
+void test_client_below_threshold()
+{
+    config.device.role = meshtastic_Config_DeviceConfig_Role_CLIENT;
+    config.lora.use_preset = false;
+    uint32_t res = Default::getConfiguredOrDefaultMsScaled(0, 60, 40);
+    TEST_ASSERT_EQUAL_UINT32(60000U, res);
+}
+
+void test_client_default_preset_scaling()
+{
+    config.device.role = meshtastic_Config_DeviceConfig_Role_CLIENT;
+    config.lora.use_preset = false;
+    // nodesOverForty = 10 -> coefficient = 1 + 10 * 0.075 = 1.75 => 60000 * 1.75 = 105000
+    uint32_t res = Default::getConfiguredOrDefaultMsScaled(0, 60, 50);
+    TEST_ASSERT_EQUAL_UINT32(105000U, res);
+}
+
+void test_client_medium_fast_preset_scaling()
+{
+    config.device.role = meshtastic_Config_DeviceConfig_Role_CLIENT;
+    config.lora.use_preset = true;
+    config.lora.modem_preset = meshtastic_Config_LoRaConfig_ModemPreset_MEDIUM_FAST;
+    // nodesOverForty = 30 -> coefficient = 1 + 30 * 0.02 = 1.6 => 60000 * 1.6 = 96000
+    uint32_t res = Default::getConfiguredOrDefaultMsScaled(0, 60, 70);
+    TEST_ASSERT_EQUAL_UINT32(96000U, res);
+}
+
+void setup()
+{
+    // Small delay to match other test mains
+    delay(10);
+    initializeTestEnvironment();
+    UNITY_BEGIN();
+    RUN_TEST(test_router_no_scaling);
+    RUN_TEST(test_client_below_threshold);
+    RUN_TEST(test_client_default_preset_scaling);
+    RUN_TEST(test_client_medium_fast_preset_scaling);
+    exit(UNITY_END());
+}
+
+void loop() {}


The test suite is missing several important test cases:

Test with explicitly set bandwidth and spread_factor: When use_preset = false, both fields should be explicitly set to avoid division by zero and test the actual formula behavior.

Test for LONG_FAST preset: Should test the default preset (SF11, BW=250kHz) with use_preset = true.

Test for edge cases:

Very high node counts (e.g., 100+ nodes)

Special bandwidth codes (31, 62, 200, 800, 1600) that map to non-integer kHz values

Different spread factors (SF7, SF12)

Test for SENSOR and TRACKER roles: These roles also don't scale (like ROUTER), according to Default.cpp lines 45-46.

These additional tests would help ensure the formula works correctly across all supported configurations.

Done it all again. Lets see what it looks like...

src/mesh/Default.h

test/test_default/test_main.cpp

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

src/mesh/Default.h

…2 utility function

…getConfiguredOrDefaultMsScaled

NomDeTom · 2026-02-21T06:23:13Z

Looks like the tests need some work:

Portduino is starting, VFS root at /home/runner/.portduino/default
Running in simulated mode.
DEBUG | ??:??:?? 0 Upgrade time to quality NTP
WARN  | 14:12:40 0 RTC not found (found address 0x00)
WARN  | 14:12:40 0 RTC not found (found address 0x00)
test/test_default/test_main.cpp:45:test_router_no_scaling:FAIL: Expected 133728 Was 60000
test/test_default/test_main.cpp:90:test_client_below_threshold:PASS
test/test_default/test_main.cpp:91:test_client_default_preset_scaling:PASS
test/test_default/test_main.cpp:80:test_client_medium_fast_preset_scaling:FAIL: Expected 96864 Was 96863

…ectedMs function

NomDeTom · 2026-02-21T07:01:17Z

Better:

Processing test_default in coverage environment
--------------------------------------------------------------------------------
Building...
Testing...
Portduino is starting, VFS root at /home/runner/.portduino/default
Running in simulated mode.
DEBUG | ??:??:?? 0 Upgrade time to quality NTP
WARN  | 06:58:59 0 RTC not found (found address 0x00)
WARN  | 06:58:59 0 RTC not found (found address 0x00)
test/test_default/test_main.cpp:102:test_router_no_scaling:PASS
test/test_default/test_main.cpp:103:test_client_below_threshold:PASS
test/test_default/test_main.cpp:104:test_client_default_preset_scaling:PASS
test/test_default/test_main.cpp:105:test_client_medium_fast_preset_scaling:PASS
-----------------------
4 Tests 0 Failures 0 Ignored 
OK
-------------- coverage:test_default [PASSED] Took 14.49 seconds --------------

NomDeTom and others added 3 commits February 15, 2026 00:55

refactor: update throttling factor calculation and add unit tests for…

03c42e6

… scaling behavior

Merge branch 'meshtastic:develop' into scaling_tweaks

8459b2c

refactor: adjust throttling factor calculation for improved accuracy …

3a3090b

…in different configurations

github-actions bot added needs-review Needs human review enhancement New feature or request labels Feb 15, 2026

Jorropo requested review from Jorropo and Copilot and removed request for Jorropo February 15, 2026 01:07

Copilot started reviewing on behalf of Jorropo February 15, 2026 01:08 View session

Jorropo added the cleanup Code cleanup or refactor label Feb 15, 2026

Copilot AI reviewed Feb 15, 2026

View reviewed changes

thebentern and others added 2 commits February 15, 2026 06:45

Update src/mesh/Default.h

5393dd0

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Update src/mesh/Default.h

8265cd0

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

GUVWAF reviewed Feb 15, 2026

View reviewed changes

src/mesh/Default.h Outdated Show resolved Hide resolved

NomDeTom and others added 3 commits February 15, 2026 20:00

refactor: enhance throttling factor calculation and introduce pow_of_…

1adf972

…2 utility function

refactor: improve expected ms calculation in unit tests for Default::…

4182201

…getConfiguredOrDefaultMsScaled

Merge branch 'develop' into scaling_tweaks

1fb5daf

NomDeTom marked this pull request as ready for review February 20, 2026 14:04

refactor: improve scaling logic for routers and sensors in computeExp…

565864b

…ectedMs function

Merge branch 'develop' into scaling_tweaks

b94ab2d

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Comments

Scaling tweaks#9653

Scaling tweaks#9653
NomDeTom wants to merge 10 commits intomeshtastic:developfrom
NomDeTom:scaling_tweaks

NomDeTom commented Feb 15, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Feb 15, 2026

Uh oh!

Uh oh!

Uh oh!

Copilot AI Feb 15, 2026

Uh oh!

NomDeTom Feb 20, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

NomDeTom commented Feb 21, 2026

Uh oh!

NomDeTom commented Feb 21, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Comments

Conversation

NomDeTom commented Feb 15, 2026

🤝 Attestations

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Feb 15, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI Feb 15, 2026

Choose a reason for hiding this comment

Uh oh!

NomDeTom Feb 20, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

NomDeTom commented Feb 21, 2026

Uh oh!

NomDeTom commented Feb 21, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants