How WasabiNet Works
The Mesh-Node Wifi technology employed by WasabiNet allows low-cost devices to create a dynamic mesh of wireless access throughout a coverage area. Each access point, or mesh node, works intelligently with the other nodes to find the shortest and fastest path through the mesh, thereby establishing a decentralized, ad-hoc infrastructure for routing Internet traffic. Furthermore, the nodes dynamically determine alternate routes to overcome disruption (e.g. interference, a node being unpowered), by continually monitoring links with their neighbors, and updating routes to their destination accordingly.
Thanks to their operation on unlicensed frequencies, these Mesh-Node Wifi access points cost only $60 to $100 each. This permits scattering the nodes in generous quantity around a neighborhood at whichever locations, indoors or outdoors, are most convenient to installation. In practice, this also allows WasabiNet a great deal of flexibility in reaching new customers. That is, if a customer is already near the boundaries of the coverage area, one need only add a node to extend the mesh to the customer's location.
A drawback of this flexibility is that mesh networks slow down with increasing density, that is, more radios clustered in the same area cause interference with each other. Furthermore, the effective bandwidth available to customers' traffic decreases with each wireless hop that traffic must traverse. To mitigate this, WasabiNet employs a layered mesh to efficiently distribute traffic while avoiding such congestion. As shown in green in Illustration 1, 2.4GHz Wifi-frequency nodes are grouped in pockets of small, roughly block-wide pocket meshes that can sustain on the order of 10mbit/s of customer traffic at each node, up and down. Each pocket mesh is then served by at least one 5.8GHz node that operates on a sparser, but much faster, backhaul mesh that can sustain on the order of 50Mbit/s of customer traffic at each node. To supply Internet connectivity, at least one node on the 5.8GHz backhaul mesh is connected to a wired Broadband uplink, e.g. DSL or cable.
Illustration 1: Diagram of Layered Mesh Network
An inherent advantage to this mesh topology is the ability to support multiple uplinks at each level, as the individual nodes will automatically chose whichever uplink is on their optimal route. That is, each node on a 2.4GHz pocket mesh, shown in green in Illustration 1, will chose whichever uplink to the backhaul mesh is closest. Likewise, each node on the 5.8GHz backhaul mesh, shown in blue in Illustration 1, will chose whichever wired uplink is closest. Do note this architecture can support multiple, independent wired Broadband uplinks, i.e. multiple DSL or cable connections, which may be sited at whatever locations (or single location) are most convenient to installation and maintenance. Furthermore, when multiple uplinks (wireless and wired) are established at each layer of mesh, this automated uplink selection provides inherent redundancy and failover.
Another advantage to this mesh topology is that, although all nodes in a particular mesh operate on the same channel, they do not interfere equally with every other node within that mesh. That is, while adjacent nodes broadcasting at full capacity to their uplink would compel each other to slow down and conserve bandwidth, distant nodes in the mesh would not experience the same degree interference. This is because the very property that motivated selection of the frequencies 2.4GHz and 5.8GHz for unlicensed use, namely their high atmospheric attenuation, almost permits highly localized usage of spectrum.
Finally, in anticipation of accommodating 1Gbit/s fiber service as uplink to the Internet, one or more of the 5.8GHz backhaul mesh nodes may furthermore be connected to one or more fiber uplinks via 5.8GHz point-to-point links capable of 300Mbit/s. This is shown with the solid and dashed lines at the top of Illustration 1. Because the point-to-point links would be configured for dedicated channels and not participate in any mesh, redundancy would be achieved with multiple, parallel links, either operating simultaneously (e.g. trunked together), or configured for automatic failover from primary to backup.