With this application, you can use the mesh networking API that Mbed OS provides.
The application demonstrates a light control application, where devices can control the LED status of all devices in the network. The application can be built for the unsecure 6LoWPAN-ND, Thread or Wi-SUN network.
See the 6LoWPAN overview for the definition of star and mesh networks. These same principles apply also to Thread and Wi-SUN protocols.
mbed import mbed-os-example-mesh-minimal
cd mbed-os-example-mesh-minimal
See the file mbed_app.json for an example of defining an IEEE 802.15.4 channel to use.
To optimize the flash usage, select a proper configuration for Nanostack. The configuration depends mostly on the preferred use case.
Select the protocol the network is based on:
- 6LoWPAN-ND.
- Thread.
- Wi-SUN
Select the device role:
- Mesh network. A router. (default)
- Star network. Nonrouting device. Also known as a host or sleepy host.
Modify your mbed_app.json file to see which Nanostack and Mbed Mesh API configuration to use.
Example configuration files are provide under configs/ directory. You may override the mbed_app.json with either of these.
| configuration file | Use for |
|---|---|
configs/mesh_6lowpan.json |
6LoWPAN-ND based mesh network. |
configs/mesh_thread.json |
Thread based mesh network. |
configs/mesh_wisun_S2LP.json |
Wi-SUN based mesh network. |
An example of the mbed_app.json file:
...
"target_overrides": {
"*": {
"nanostack.configuration": "lowpan_router",
"nsapi.default-mesh-type": "LOWPAN",
"mbed-mesh-api.6lowpan-nd-panid-filter": "0xffff",
"mbed-mesh-api.6lowpan-nd-channel-page": 0,
"mbed-mesh-api.6lowpan-nd-channel": 12,
"mbed-mesh-api.6lowpan-nd-channel-mask": "(1<<12)",
"mbed-mesh-api.heap-size": 14000,
"mbed-trace.enable": false,
"platform.stdio-convert-newlines": true,
"platform.stdio-baud-rate": 115200,
"atmel-rf.provide-default": true,
"mcr20a.provide-default": false,
"target.device_has_add": ["802_15_4_PHY"],
"target.network-default-interface-type": "MESH"
},
The following tables show the values to use in the mbed_app.json file for your devices in different networks.
- For a 6LoWPAN-ND based network, use
nsapi.default-mesh-type: LOWPAN. - For a Thread-based network, use
nsapi.default-mesh-type: THREAD. - For a Wi-SUN-based network, use
nsapi.default-mesh-type: WISUN.
nsapi.default-mesh-type: LOWPAN
| Device role | nanostack.configuration value |
mbed-mesh-api.6lowpan-nd-device-type value |
|---|---|---|
| Mesh router (default) | lowpan_router | NET_6LOWPAN_ROUTER |
| Nonrouting device | lowpan_host | NET_6LOWPAN_HOST |
nsapi.default-mesh-type: THREAD
| Device role | nanostack.configuration value |
mbed-mesh-api.thread-device-type value |
|---|---|---|
| Mesh router (default) | thread_router | MESH_DEVICE_TYPE_THREAD_ROUTER |
| Nonrouting device | thread_end_device | MESH_DEVICE_TYPE_THREAD_SLEEPY_END_DEVICE |
By default, the Thread application uses the static network link configuration defined in the mesh API configuration file. If you want to commission a Thread device, see how to commission a Thread device in practice.
The Thread stack learns the network settings from the commissioning process and saves them to RAM memory. Therefore, the learned network settings are lost when you restart the device next time. To prevent re-commissioning, you can save the Thread configuration settings to an SD card as follows (only in K64F):
- Change
storage-devicetoMESH_NVM_SD_CARDin the./configs/mesh_thread.jsonfile. - Enable commissioning as descibed in the referred instructions.
- Compile and program the application.
- Insert an erased or FAT-formatted SD card to the
K64Fmemory card slot. The application will initialize the SD card with the appropriate file system on first use. - Commission the device to the Thread network.
- When you restart the device next time, the device reads the Thread configuration settings from the SD card and tries to attach to an existing network.
nsapi.default-mesh-type: WISUN
| Device role | nanostack.configuration value |
|---|---|
| Mesh router (default) | ws_router |
The networking stack used in this example requires TLS functionality to be enabled on Mbed TLS. On devices where hardware entropy is not present, TLS is disabled by default. This results in compile time failures or linking failures.
To learn why entropy is required, read the TLS Porting guide.
See Notes on different hardware for known combinations of development boards and RF shields that have been tested.
You also need to check how LEDs and buttons are configured for your hardware, and update .json accordingly.
To run a 6LoWPAN-ND network, you need a working RF driver for Nanostack. This example uses the Atmel AT86RF233 by default.
To change the RF driver modify the mbed_app.json file by setting preferred RF driver provide_default value to true, For example, to use MCR20a RF driver:
"atmel-rf.provide-default": false,
"mcr20a.provide-default": true,To run a Wi-SUN network, there is currently only one supported Radio, STM S2LP.
To use the S2LP, modify the mbed_app.json file by setting:
"s2lp.provide-default": true,Note that the provide-default for each radio driver defaults to false, so there is no need to list all the possible drivers in the mbed_app.json.
mbed compile -m K64F -t GCC_ARM
A binary is generated in the end of the build process.
We are using the Atmel AT86RF233, which you can purchase. Place the shield on top of your board, and power it up.
Drag and drop the binary to the target to program the application.
This example supports the following border router:
The border router supports static and dynamic backhaul configuration. The static configuration is good for testing, but the dynamic one works if your network infrastructure is supplying an IPv6 address. Make sure that you use the appropiate mode.
Remember to connect the Ethernet cable between the border router and your home/office router. Then power on the board.
By default the application is built for the LED control demo, in which the device sends a multicast message to all devices in the network when the button is pressed. All devices that receive the multicast message will change the LED status (red LED on/off) to the state defined in the message. Note, that the Thread devices can form a network without the existance of the border router. The following applies only to the case when the border router is set-up.
As soon as both the border router and the target are up and running you can verify the correct behaviour. Open a serial console and see the IP address obtained by the device.
Note: This application uses the baud rate of 115200.
connected. IP = 2001:db8:a0b:12f0::1
You can use this IP address to ping from your PC and verify that the connection is working correctly.
On some limited platforms, for example NCS36510 or KW24D, building this application might run out of RAM or ROM. In those cases, you might try following these instructions to optimize the memory usage.
The custom Mbed TLS configuration can be set by adding "macros": ["MBEDTLS_USER_CONFIG_FILE=\"mbedtls_config.h\""] to the .json file. The example Mbed TLS config minimizes the RAM and ROM usage of the application. The configuration works on K64F, but it is not guaranteed to work on every Mbed Enabled platform.
This configuration file saves you 8.7 kB of RAM but uses 6.8 kB of more flash.
You can disable the LED control example by specifying enable-led-control-example": false in the mbed_app.json
This saves you about 2.5 kB of flash.
Nanostack's internal event-loop is shared with Mbed Client and therefore requires lots of stack to complete the security hanshakes using TLS protocols. In case client functionality is not used, you can define the following to use 2kB of stack
"nanostack-hal.event_loop_thread_stack_size": 2048
This saves you 4kB of RAM.
Nanostack uses internal heap, which you can configure in the .json. A thread end device with comissioning enabled requires at least 15kB to run.
In mbed_app.json, you find the following line:
"mbed-mesh-api.heap-size": 15000,
For 6LoWPAN, you can try 12kB. For the smallest memory usage, configure the node to be in nonrouting mode. See module-configuration for more detail.
Nanostack's internal heap can be moved within the system heap. This helps on devices that have split RAM and compiler fails to fit statically allocated symbols into one section, for example, the NXP KW24D device.
Mesh API has the use-malloc-for-heap option to help this.
Add the following line to mbed_app.json to test:
"mbed-mesh-api.use-malloc-for-heap": true,
For devices with small memory, we recommend using release profiles for building and exporting. Please see the documentation about Build profiles.
Examples:
$ mbed export -m KW24D -i make_iar --profile release
OR
$ mbed compile -m KW24D -t IAR --profile release
If you have problems, you can review the documentation for suggestions on what could be wrong and how to fix it.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent