Access points interconnected with peer-to-peer wireless links create a backhaul infrastructure called a wireless mesh network. These networks extend service across large geographic areas, such as campuses or metropolitan areas, facilitating expanded broadband wireless applications.
An IEEE technical group is working to develop the 802.11s standard for wireless LAN (WLAN) mesh networking. At its March plenary session, the group announced the baseline document for the standard. The group expects to have an initial draft by July and a ratified 802.11s standard by early 2008.
The group is defining capabilities in several areas, including:
- Topology discovery.
- Path selection and forwarding.
- Channel allocation.
- Traffic management.
- Network management.
Mesh networks introduce some new terminology. The mesh architecture defines a mesh point as a node that supports mesh services. A mesh point that supports access-point services as well as mesh services is called a mesh access point. A variant of a mesh point that connects into the wired network is called a mesh portal.
The existing 802.11 media access-control layer is being enhanced to support mesh services. Mesh networking will work with existing 802.11 radio technologies. Mesh services will be compatible with existing WLAN clients.
In a mesh network a mesh node must be able to discover its peers and associate with them. A node should also be able to select an optimal path through a mesh network to forward data frames. The standards group is working on a path-selection protocol called Hybrid Wireless Mesh. The specification is being designed in such a way that vendors can use their own protocols for path selection, so interoperable and proprietary mesh deployments will be possible.
The 802.11s group intends to take advantage of security mechanisms specified in 802.11i (completed in 2004), but extensions will be necessary, because 802.11i provides only one-hop link security, and mesh networks require multihop or end-to-end security. Additional work will define how mesh nodes can mutually authenticate themselves and create secure associations. Each node would act as a supplicant and authenticator for adjacent nodes. Distributed and centralized 802.1x authentication schemes will need to be supported. Re-authentication must occur rapidly for roaming nodes to preserve session persistence.
Engineering traffic to avoid congestion within a multihop wireless mesh network is a challenge. Local congestion on a mesh node can affect neighbouring nodes using the same channel. Extensions to the QoS mechanisms defined in 802.11e are being considered to support hop-by-hop congestion control. The standards body is also looking at ways to implement rate control to alleviate congestion.
Deploying a mesh network with thousands of nodes requires a scalable and comprehensive centralized network management system. It must manage bandwidth, security and QoS policies across a network. Planning and designing a network are essential prerequisites for a successful deployment. A mesh network is dynamic in nature, with topology changes happening in real time. Monitoring of a network with rapid corrective action becomes critical to deliver performance and reliability.
Mesh networking will usher in a new paradigm to support an expanded range of wireless services and applications.
Saraf is director of product marketing and Murphy is director of software engineering for Trapeze Networks