Wireless mesh technology enables 802.11 wireless networks to cover areas far beyond the limits imposed by traditional WLAN technology. Proprietary solutions have been offered for years, but the IEEE 802.11s standard, currently under development, will make multi-vendor mesh networks possible.
Most existing 802.11 networks operate in infrastructure mode. All communication travels through an access point (AP) before being forwarded to its final destination. The 802.11 standard also defines ad hoc mode. An ad hoc network has no AP. Stations communicate directly from one to another.
802.11s extends the ad hoc concept. Packets may travel through intermediate stations on the way to their ultimate destination. The result: 802.11 mesh networks can extend across a larger area than non-mesh 802.11. A station can communicate with a station too far away for direct transmission. Packets travel hop by hop via intermediate stations until close enough to the destination for the final transmission.
An 802.11s network can operate using any of the 802.11a, b, g or n standards. The Media Access Control (MAC) layer and packet formats have been designed to prevent interference with non-mesh networks so mesh and non-mesh networks can be co-located. No modification of traditional 802.11 equipment is required to enable co-location.
All 802.11s functions operate at either the MAC or Link Layer. None of the higher layers of the IP network stack are modified, so applications do not require modification.
- Mesh Point (MP): An 802.11s station supporting node-to-node communication with other 802.11s stations.
- Mesh Portal (MPP): An 802.11s station with a wired network interface in addition to its 802.11s capabilities. It provides a path for packets from the mesh to a wired network.
- Mesh Access Point (MAP): A MAP acts as an access point for non-mesh nodes. It may also connect to a wired network and act as an MPP.
A network can be made up of just MPs communicating among themselves, with no connection to a wired network. A more useful configuration contains an MPP. MPs could be spread out over a large area, possibly outdoors or in a large facility such as a warehouse. Use of conventional non-mesh stations would require locating APs throughout the area and extending the wired network to each. 802.11s requires a single wired connection to an MPP. Distant mesh stations communicate to the MPP via intermediate stations.
In another possible configuration, MAPs could be used to provide access across a large area to non-mesh 802.11 stations. Each non-mesh station would associate with the nearest MAP. Packets would then travel hop by hop through intermediately placed MPs to an MPP at the edge of the network. A single wired connection to the MPP could support non-mesh stations too far from the wired connection for a standard AP.
Finding a route through the network
To the upper layers of the IP stack, Layer 3 and above, the mesh appears to be a single switched network. 802.11s replaces the spanning tree used in traditional switched networks with Hybrid Wireless Mesh Protocol (HWMP). HWMP creates routes through the mesh using two techniques: proactive routing and on-demand routing.
Proactive routing provides an efficient routing method based on the fact that most mesh networks will include at least one Mesh Portal, and many packets will be routed through a portal. An MP can be configured to be a network root. An MP configured as a root sends periodic Portal Announcements and waits to hear announcements from other MPs. If there are multiple MPs configured to be a root, they negotiate to designate one as the network root.
The selected root portal then sends Root Announcements that are forwarded hop by hop through the network. Each MP creates a tree-structured routing table based on received Root Announcements from neighboring MPs. An MP will often receive multiple announcements that have traveled different paths. It will create a routing table based on metrics in the announcement that are updated by each MP along the path.
802.11s nodes use on-demand routing to find destination nodes when there is no proactive route to the destination. The source node sends requests for the destination to neighboring nodes. The requests are forwarded along until one reaches the destination. The destination node then sends a response that is forwarded back to the source.
Requests are sent to all neighboring nodes and may travel many paths to the destination. Duplicate copies may arrive at the destination node. Similarly, duplicate copies of the response may be received by the initial sending node. Radio Metric Ad Hoc On Demand Distance Vector routing (RM-AODV) is used to choose among the possible paths. RM-AODV is a modified version of AODV, enhanced to include airtime, a metric based on observed transmission rate, amount of traffic and interference along a path. AODV is defined by RFC 3561.
Wireless security with 802.11s
Mesh networks add complexity to the process of establishing secure access because nodes maintain paths to many neighboring nodes, not just to a single AP. 802.11s uses the 802.11i and 802.1X standards. Each node uses 802.11i to negotiate parameters and key pairs with each adjacent node.
While the 802.11s standard is not final, implementations are under way. The low-power features in the draft are ideal for the environments in which the One Laptop Per Child (OLPC) organization expects its units to operate. OLPC has developed an implementation for laptop-to-laptop communication based on the draft. Also, the open80211s Consortium is creating an open source implementation.
About the author:
David B. Jacobs of The Jacobs Group has more than 20 years of networking industry experience. He has managed leading-edge software development projects and consulted to Fortune 500 companies as well as software startups.