Wireless networks have long embraced a centralised model that holds the potential for bottlenecks, latency and a single point of failure. But wireless mesh networks are emerging as an alternative to wireless switching. Mesh networks distribute intelligence from switches to access points by incorporating a grid-like topology.

A mesh network allows nodes or access points to communicate with other nodes without being routed through a central switch point, eliminating centralised failure, and providing self-healing and self-organisation. Although decisions on traffic are made locally, the system can be managed globally.

Today’s wireless LAN mesh networks use standards-based 802.11a/b/g, but they can be extended to any radio-frequency technology, such as UltraWideband or 802.15.4 Zigbee. Because network intelligence is contained within each access point, no centralised switches are needed — just intelligent access points with network processors, switching capability and system software.

For a network to intercommunicate in a mesh topology, the nodes’ self-discovery features first must determine whether they are to serve as access points for wireless devices, as backbones for traffic coming from another node, or a combination of roles.

Next, the individual nodes locate their neighbours using discovery query/response protocols. These network protocols must be parsimonious so they don’t add much overhead to the traffic — that is, they cannot require more than 1 to 2 per cent of available bandwidth.

Once the nodes recognise one another, they measure path information such as received signal strength, throughput, error rate and latency. These values must be communicated among the neighbour nodes, but this information must not take up very much bandwidth. Based on these signal values each node then selects the best path to its neighbours so the optimum quality of service is obtained at any given moment. The network-discovery and path-selection processes run in the background, so that each node maintains a current list of neighbours and frequently recomputes the best path. If a node is taken off the network — for maintenance, rearrangements or failure — the adjacent nodes quickly can reconfigure their tables and recompute paths to maintain traffic flow when the network changes. This self-healing attribute, or failover, sets mesh topologies apart from hub-and-spoke networks.

Each node is self-managed, yet is part of an organised network that can be managed and configured as a single entity from a centralised point. Using SNMP, a systems administrator can set and monitor individual elements, nodes, domains or an entire network. Discovery protocols simplify the task by seeking out and locating individual nodes for display on management screens.

Because mesh networks rely on management, control and discovery messages, they must be able to secure this traffic and user traffic. In-band messages, secured within encrypted tunnels, remain free from eavesdropping and similar attacks. Standards-based security techniques, such as 802.1x and Advanced Encryption Standard encryption, ensure that only authenticated wireless devices and nodes are in the connection and that they are properly encrypted.

Mesh topology is inherently reliable and redundant, and can be expanded quickly. A wireless mesh network doesn’t require elaborate planning and site mapping. Nodes can be up and running as soon as administrators mount them.

With intelligent points on the network dispersed, mesh networks can organise themselves, select the best path for user traffic, route around failures or congestion, and provide secure connections. The decentralisation provides for unlimited growth and stability. Mesh architecture just might be the future of wireless networks.

Bob Jordan is co-founder and VP of marketing at Strix Systems

Events in the industry getting you stirred up? Don't just steam about it; share it with your peers. Send e-mails of around 300 words to David_Beynon@idg.com.au