Network level concepts

This section describes important concepts relating to the work of the ZigBee stack.

ZigBee nodes

There are three general types of node that can exist in a ZigBee network:

  • Coordinator

  • Router

  • End Device

Note: These roles exist at the network level - a ZigBee node may also be performing tasks at the Application level, independent of the role it plays in the network.

For example, a network of ZigBee devices measuring temperature may have a temperature sensor application in each node, irrespective of whether the node is an End Device, Router, or the Coordinator.

The roles of these node types are described in the sub-sections below.

Coordinator

All ZigBee networks must have one (and only one) Coordinator.

At the network level, the Coordinator is mainly needed at system initialization - it is the first node to be started and performs the following initialization tasks:

  • Selects the frequency channel to be used by the network (usually the one with the least detected activity)

  • Starts the network

  • Allows child nodes to join the network through it

The Coordinator can additionally provide other services such as message routing and security management. It may also provide services at the Application level. If any of these additional services are used, the Coordinator must be able to provide them at all times. However, if none of these additional services are used, the network will be able to operate normally even if the Coordinator fails or is switched off.

Parent topic:ZigBee nodes

Router

A ZigBee PRO network usually has at least one Router. The main tasks of a Router are:

  • Relays messages from one node to another.

  • Allows child nodes to join the network through it.

Note: An important feature of the Router is that it cannot sleep, as it must always be available for routing.

Parent topic:ZigBee nodes

End Device

The main task of an End Device at the network level is sending and receiving messages. An End Device can only communicate directly with its parent, so all messages to/from an End Device pass via its parent.

An End Device can be battery-powered and, when not transmitting or receiving, can sleep in order to conserve power. The parent device buffers messages destined for a sleep-enabled End Device. The End Device collects these messages once it is awake (also see Section 3.2.2 Network topology below).

Note: End Devices cannot relay messages and cannot allow other nodes to connect to the network through them. In other words, it implies that they cannot have children.

Parent topic:ZigBee nodes

Parent topic:Network level concepts

Network topology

The ZigBee PRO standard was designed to facilitate wireless networks with the Mesh topology.

A Mesh network consists of a Coordinator, Routers, and End Devices. The Coordinator is associated with a set of Routers and End Devices - its children. A Router may then be associated with more Routers and End Devices - its children. This can continue to a number of levels. The relationships between the nodes must obey the following rules:

  • The Coordinator and Routers can have children, and can therefore be parents.

  • A Router can be both a child and a parent.

  • End Devices cannot have children, and therefore cannot be parents. The communication rules for a Mesh network are as follows:

  • An End Device can only directly communicate with its parent (and with no other node).

  • A Router can directly communicate with its children, with its own parent, and with any other Router or Coordinator within radio range.

  • The Coordinator can directly communicate with its children and with any Router within radio range.

The resulting structure is illustrated in the figure below.

Mesh topology

Mesh topology

In ZigBee PRO, the maximum depth (number of levels below the Coordinator) of a network is 15. The maximum number of hops that a message can make in traveling between the source and destination nodes is 30 (twice the maximum depth).

A routing node (Router or Coordinator) can communicate directly with other routing nodes within radio range. This specific property distinguishes a Mesh network from a Tree network. This property enables very efficient and flexible message propagation. It also implies that alternative routes can be found if a link fails or there is congestion.

Note: An End Device, which is able to sleep, is unable to receive messages directly. A message destined for a sleep-enabled End Device is always buffered in its parent node if the End Device is asleep when the message arrives. Once the End Device is awake, it must ask or ‘poll’ the parent for messages. In the Mesh topology, a ‘route discovery’ feature is provided, which allows the network to find the best available route for a message. Refer further details in Section 3.5.2, “Route discovery”.

Note: Message propagation is handled by the network layer software and is transparent to the application programs running on the nodes.

Parent topic:Network level concepts

Neighbor tables

A routing node (Router or Coordinator) holds information about its neighboring nodes. This information is stored in a Neighbor table containing entries for the node’s immediate children, for its own parent and, in a Mesh network, for all peer Routers with which the node has direct radio communication.

It is possible to define the maximum number of entries in a Neighbor table. If this parameter is set to a low value, it will result in a ‘long, thin network’.

The structure and configuration of a Neighbor table are described in Appendix B.5.1.

Parent topic:Network level concepts

Network addressing

In a ZigBee network, each node must have a unique identification. For this purpose, each node has two addresses:

  • IEEE (MAC) address: A 64-bit address, allocated by the IEEE, which uniquely identifies the device. No two devices in the world can have the same IEEE address. It is often referred to as the MAC address. In a ZigBee network, it is sometimes called the ‘extended’ address.

  • Network address: A 16-bit address that identifies the node in the network and is local to that network. Thus, two nodes in separate networks may have the same network address. It is sometimes called the ‘short’ address.

In ZigBee PRO, the network address of a node is dynamically assigned as a random 16-bit value by the parent when the node first joins the network. This is known as stochastic addressing due to the randomness of the address allocation. Although random, the parent ensures that the chosen address has not already been assigned to one of its neighbors. In the unlikely event of the address already existing in the network beyond the immediate neighborhood, a mechanism exists to automatically detect and resolve the conflict. The allocated network address can be retained by the joining node, even if it later loses its parent and acquires a new parent.

The Coordinator always has the network address 0x0000.

While an application on a node may use IEEE/MAC addresses or network addresses to identify remote nodes, the ZigBee PRO stack always uses network addresses for this purpose. To facilitate translation between IEEE/MAC addresses and network addresses, an Address Map table may be maintained on the node, where each table entry contains the pair of addresses for a remote node.

In the NXP implementation of ZigBee PRO, the IEEE/MAC addresses (of other network nodes) are stored in a single place on a node, called the MAC Address table. This avoids the need to repeat the 64-bit IEEE/MAC addresses in other tables, such as the Address Map table and Neighbor table, and therefore saves storage space. Instead, a 16-bit index to the relevant entry in the MAC Address table is stored in the other tables.

It is also possible to define a 16-bit ‘group address’ which refers to a set of applications (or endpoints that may be located across several nodes. For details, refer to Section 3.4.1, Multiple applications and endpoints).

Specifying a group address in a data transfer results in the data being broadcast to all nodes in the network but, at the destinations, the data is only passed to those applications, which are covered by the group address. Refer to Section 6.3, Managing group addresses for more details of using group addresses.

Parent topic:Network level concepts

Network identity

A ZigBee network must be uniquely identifiable. This allows more than one ZigBee network to operate in close proximity - nodes operating in the same space must be able to identify which network they belong to.

For this purpose, ZigBee uses two identifiers, as follows:

  • PAN ID: A 16-bit value called the PAN ID (Personal Area Network Identifier) is used in inter-node communications (implemented at the IEEE 802.15.4 level of the stack) to identify the relevant network. A value for the PAN ID is selected at random by the Coordinator when the network is started. When other nodes join the network, they learn the network’s PAN ID and use it in all subsequent communications with the network.

It is possible that the PAN ID generated for a newly installed network clashes with the PAN ID of another network already operating on the same radio channel, in the same neighborhood. In this case, ZigBee PRO automatically resolves such a conflict by generating another random PAN ID for the new network. This continues until a value is obtained that does not clash with the PAN ID of any other detectable network.

  • Extended PAN ID: A 64-bit value called the Extended PAN ID (EPID) is used in forming the network and subsequently modifying the network, if necessary. This identifier can be pre-set to a random value in the user application that runs on the Coordinator. Alternatively, the identifier can be pre-set to zero. In this case, the Coordinator adopts its own 64-bit IEEE/MAC address as the Extended PAN ID when the network starts. This is a sure way of obtaining a globally unique value (see Section 3.2.4).

When a Router or End Device first tries to find a network to join, it uses the Extended PAN ID in either of following ways:

  • If an Extended PAN ID has been pre-set in the user application for the Router or End Device, the node joins the network that has this Extended PAN ID (provided this network is detected).

  • If there is no pre-set Extended PAN ID for the Router or End Device, the node joins the first network detected, irrespective of the Extended PAN ID. The joining node then learns the Extended PAN ID of its network. It later uses this identifier to rejoin the network if, for some reason, it loses contact with the network (the node is orphaned).

For more information on joining a network, refer to Section 3.3.2.

Note: At the Application level, you only need to be concerned with the Extended PAN ID, as the allocation and use of the PAN ID is transparent to the application.

Parent topic:Network level concepts

Parent topic:ZigBee PRO architecture and operation