Zigbee application helper functions

Table of Contents

• Zigbee default signal handler
  • Complete zboss_signal_handler implementation
  • Zigbee stack startup without autostart
  • Zigbee Base Device Behavior initialization
    • New device scenario
    • Commissioned device scenario
  • Zigbee network formation and commissioning
  • Zigbee network leaving
  • Zigbee network rejoining
• Zigbee stack sleep routines

The nRF5 SDK for Thread and Zigbee provides several helper functions that are intended to facilitate the end user application implementation. They implement the default behavior of a Zigbee device and are used in the Zigbee examples in this SDK.

Zigbee default signal handler

The Zigbee default signal handler is a function called from the zboss_signal_handler() function that is mandatory for each application.

Because most of Zigbee devices behave in a similar way, zigbee_default_signal_handler() was introduced to provide a default logic to handle stack signals.

As a result, the minimal implementation of zboss_signal_handler includes only a call to the default signal handler:

void zboss_signal_handler(zb_uint8_t param)
{
 /* Call default signal handler. */
 zigbee_default_signal_handler(param);

 /* Free the Zigbee stack buffer. */
 zb_buf_free(param);
}

With this call, the device will be able to join the Zigbee network or join a new network when it leaves the previous one.

In general, using the default signal handler is worth considering, because of the following reasons:

• It simplifies the application.
• It provides a default behavior for each signal, which reduces the risk that an application will break due to an unimplemented signal handler.
• It provides Zigbee role specific behavior (e.g. finite join/rejoin time)
• It makes the application less sensitive to changes in the Zigbee stack commissioning API.

Apart from that, the default signal handler serves as a good starting point for a custom signal handler implementation.

Complete zboss_signal_handler implementation

In its complete implementation, the zboss_signal_handler allows the application to control a broader set of basic functionalities, including joining, commissioning, and network formation.

There are in fact cases in which the default handler will not be sufficient and needs to be extended. For example, when the application wants to use the procedure of the initiator of finding & binding or use the production configuration feature.

If the application wants to change or extend the default behavior, use a switch-case statement and then the default signal handler inside the default case:

void zboss_signal_handler(zb_bufid_t bufid)
{
 zb_zdo_app_signal_hdr_t * p_sg_p = NULL;
 zb_zdo_app_signal_type_t sig = zb_get_app_signal(bufid, &p_sg_p);
 zb_ret_t status = ZB_GET_APP_SIGNAL_STATUS(bufid);

 switch (sig)
 {
 case ZB_BDB_SIGNAL_DEVICE_REBOOT:
 /* fall-through */
 case ZB_BDB_SIGNAL_STEERING:
 if (status == RET_OK)
 {
 NRF_LOG_INFO("Joined network successfully");
 // TODO: Start application-specific logic that requires the device to be connected to a Zigbee network.
 }
 else
 {
 zb_bool_t comm_status;

 NRF_LOG_INFO("Unable to join the network. Restart network steering.");
 comm_status = bdb_start_top_level_commissioning(ZB_BDB_NETWORK_STEERING);
 ZB_COMM_STATUS_CHECK(comm_status);
 }
 break;

 default:
 // Call default signal handler.
 ZB_ERROR_CHECK(zigbee_default_signal_handler(bufid));
 break;
 }

 if (bufid)
 {
 zb_buf_free(bufid);
 }
}

Zigbee stack startup without autostart

Whenever zigbee_default_signal_handler() is used, the stack should be started using a call to zboss_start_no_autostart():

zboss_start_no_autostart();

The reason for this is that although the Zigbee stack provides some automatic initialization procedure with zboss_start(), the default signal handler requires the application to start the stack without those mechanisms.

Starting the stack with zboss_start_no_autostart() has the following advantages:

• It allows you to ensure better granularity in customizing the default procedure.
• You gain a better understanding of procedures executed by the stack, because all procedures specified by the Zigbee Base Device Behavior specification are explicitly started inside the default signal handler and generate a separate signal once finished.

Zigbee stack startup without autostart

When the stack is started through zboss_start_no_autostart(), the stack generates the following signals:

• ZB_ZDO_SIGNAL_PRODUCTION_CONFIG_READY – indicating that the stack attempted to load application-specific production configuration from flash memory.
• ZB_ZDO_SIGNAL_SKIP_STARTUP – indicating that the stack has initialized all internal structures and the Zigbee scheduler has started.

The reception of these signals determines the behavior of the default signal handler:

• Upon reception of ZB_ZDO_SIGNAL_PRODUCTION_CONFIG_READY, the default signal handler will print out a log with the signal status and exit.

ZB_ZDO_SIGNAL_PRODUCTION_CONFIG_READY signal handler

• Upon reception of ZB_ZDO_SIGNAL_SKIP_STARTUP signal, the default signal handler will perform the BDB initialization procedure, and then exit.
  

  ZB_ZDO_SIGNAL_SKIP_STARTUP signal handler

  Note

  If you want to perform some actions before the stack attempts to join or rejoin the Zigbee network, you can overwrite this behavior by providing a custom ZB_ZDO_SIGNAL_SKIP_STARTUP signal handler implementation.

Zigbee Base Device Behavior initialization

Once the BDB initialization procedure is finished, depending on the data stored inside the Zigbee persistent storage, the stack will complete one of the following scenarios:

• New devices: Generate the ZB_BDB_SIGNAL_DEVICE_FIRST_START signal for factory new devices.
• Commissioned devices: Perform a single attempt to rejoin the Zigbee network based on NVRAM contents and then generate the ZB_BDB_SIGNAL_DEVICE_REBOOT signal.

Both scenarios will cause different behavior of the the default signal handler.

New device scenario

For factory new devices, the default signal handler will:

• Start the BDB network formation on coordinator devices. Once finished, the stack will generate ZB_BDB_SIGNAL_FORMATION signal, and continue to Zigbee network formation and commissioning.
• Call start_network_rejoin() to start the Zigbee network rejoining on routers and end devices. Once the procedure is started, the device tries to join the network until cancellation. Each try takes place after a longer period of waiting time, for a total maximum of 15 minutes. Devices may behave differently because the implementation of start_network_rejoin() is different for different Zigbee roles. See Zigbee network rejoining for more information.

Once handling of the signal is finished, the stack will generate the ZB_BDB_SIGNAL_STEERING signal, and will continue to Zigbee network formation and commissioning.

Scenario for factory new devices (ZB_BDB_SIGNAL_DEVICE_FIRST_START)

Commissioned device scenario

For devices that have been already commissioned, the default handler will:

• Not perform additional actions if the device implements a coordinator role.
  • This will keep the network closed for new Zigbee devices even if the coordinator is reset.
• Not perform additional actions if the device successfully rejoins Zigbee network.
  • This will not open the network for new devices if one of existing devices is reset.
  • In case of the Zigbee network rejoining is running, it will be cancelled.
• For routers and end devices, if they did not join the Zigbee network successfully, Zigbee network rejoining is started by calling start_network_rejoin().

Once finished, the stack will generate the ZB_BDB_SIGNAL_STEERING signal, and continue to Zigbee network formation and commissioning.

Scenario for already commissioned devices (ZB_BDB_SIGNAL_DEVICE_REBOOT)

Zigbee network formation and commissioning

According to the logic implemented inside the default signal handler, the devices can either form a network or join an existing network:

1. Coordinators will first form a network. Attempts to form the network will continue infinitely, with a one-second delay between each attempt.
   

   Forming a network following the generation of ZB_BDB_SIGNAL_FORMATION
   By default, after the successful network formation on the coordinator node, a single-permit join period of 180 seconds will be started, which will allow new Zigbee devices to join the network.
2. Other devices will then join an existing network during this join period.
   • When a device has joined and Zigbee network rejoining is running, the procedure is cancelled.
   • If no device has joined and the procedure is not running, the procedure will be started.
     

     Forming a network following the generation of ZB_BDB_SIGNAL_STEERING

Zigbee network leaving

The default signal handler implements the same behaviour for handling ZB_ZDO_SIGNAL_LEAVE for both routers and end devices. When leaving the network, the default handler calls start_network_rejoin() to start Zigbee network rejoining to join a new network.

Once start_network_rejoin() is called, the stack will generate the ZB_BDB_SIGNAL_STEERING signal and will continue to Zigbee network formation and commissioning.

Leaving the network following ZB_ZDO_SIGNAL_LEAVE

Zigbee network rejoining

The Zigee network rejoin procedure is a mechanism that is similar to the ZDO rejoin backoff procedure. It is implemented to work with both routers and end devices and simplify handling of cases such as device joining, rejoining, or leaving the network. It is used in default_signal_handler() by default.

If the network is left by a router or an end device, the device will try to join any open network.

• The router will use the default signal handler to try to join or rejoin the network until it succeeds.
• The end device will use the default signal handler to try to join or rejoin the network for a finite period of time, because the end devices are often powered by batteries.
  • The procedure to join or rejoin the network is restarted after the device reset or power cycle.
  • The procedure to join or rejoin the network can be restarted by calling user_input_indicate(), but it needs to be implemented in the application (for example, by calling user_input_indicate() when a button is pressed). The procedure will be restarted only if the device does not join and the procedure is not running.

The Zigbee rejoin procedure retries to join a network with each try after a specified amount of time: 2^n seconds, where n is the number of retries. The period is limited to 15 minutes if the result is higher than that.

• When start_network_rejoin() is called, the rejoin procedure is started and depending on the device role:
  • For the end device, the application alarm is scheduled with stop_network_rejoin(ZB_TRUE), to be called after the amount of time specified in ZB_DEV_REJOIN_TIMEOUT_MS. Once called, the alarm stops the rejoin.
• When stop_network_rejoin(was_scheduled) is called, the network rejoin is cancelled and the alarms scheduled by start_network_rejoin() are cancelled.
  • Additionally for the end device, if stop_network_rejoin() is called with was_scheduled set to ZB_TRUE, user_input_indicate() can restart the rejoin procedure.
• For end devices only, user_input_indicate() restarts the rejoin procedure if the device did not join the network and is not trying to join a network. It is safe to call this function from an interrupt and to call it multiple times.

Warning

The Zigbee network rejoin procedure is managed from multiple signals in default_signal_handler(). If the application controls the network joining, rejoining, or leaving, each signal in which the Zigbee network rejoin procedure is managed should be handled in the application. In this case, user_input_indicate() must not be called.

---

Zigbee stack sleep routines

The Zigbee stack provides the following mechanisms to reduce power consumption:

• On Zigbee end devices, routers and coordinators:
  • The stack calls the nrf_pwr_mgmt_run API (through zb_osif_wait_for_event) if the Zigbee scheduler queue is empty.
• On Zigbee sleepy end devices:
  • If there are no immediate callbacks scheduled through the Zigbee scheduler, the stack calls the nrf_pwr_mgmt_run API (through zb_osif_wait_for_event).
  • If there are no delayed callbacks (alarms) scheduled for the next sleep_threshold period, the stack switches off all peripheral devices related to the Zigbee stack before calling the nrf_pwr_mgmt_run API.

On Zigbee sleepy end devices, the sleep_threshold period lasts approximately 15 ms by default and can be modified by using the zb_sleep_set_threshold API.

If the device implements a Zigbee sleepy end device, the ZB_COMMON_SIGNAL_CAN_SLEEP signal is generated whenever the device can be put into the deep sleep mode. This signal can also be handled using the Zigbee default signal handler. If so, it will allow the Zigbee stack to enter the deep sleep state.

If the default behavior is not applicable for the application, you can customize the sleep functionality by using the following options:

• Overwriting the zb_osif_go_idle weak function and implementing a custom logic for handling the stack idle state.
  

  Implementing a custom logic for handling the stack idle state
• Overwriting the zb_nrf52840_sleep weak function and implementing a custom logic for putting the stack into the deep sleep mode.
  • As part of this option, overwriting zb_nrf52840_periph_disable and zb_nrf52840_periph_enable weak functions to change the set of peripheral devices, whose power is managed by the Zigbee stack.
  • The stack informs the application about the possibility to enter the deep sleep mode by generating ZB_COMMON_SIGNAL_CAN_SLEEP signal.
    

    Implementing a custom logic for putting the stack into the deep sleep mode