Getting Up to Speed, Part 1: Optimizing Wi-Fi Mesh Node Communication

Tell a Friend

While metro-scale mesh systems have several factors in common, a number of different architectural approaches have been employed by various vendors. What differentiates one from another is the ability to carry subscriber traffic over multiple hops to a wireline point-of-presence (POP) without throughput degradation.

Some architectures are extremely efficient in carrying subscriber traffic over many hops and may require as few as one Internet connectivity point for every one-hundred wireless nodes. Others devices are inefficient and may require as many as one bandwidth insertion point for every three nodes.

In the latter case, a potentially large number of additional building sites, additional costly wired insertion points, and many nonmeshed point-to-multipoint links are required to backhaul traffic, which significantly drives CAPEX (time to market and capital) and OPEX upward. Management of multiple vendors’ products and multiple complex backhaul links becomes increasingly complex.

The use of a large number of backhaul links turns a fully meshed network into a partial-mesh network. Fully-meshed networks have more redundant paths back to the POP. Thus they will exhibit higher network availability, and a single link outage is not service affecting.

The key architectural difference between those architectures capable of carrying traffic over multiple hops (hence requiring fewer bandwidth insertion points) and those only capable of carrying traffic over a couple of hops (hence requiring many bandwidth insertion points) is the number of radios per node dedicated to carrying traffic from node to node.

Routing algorithms, no matter how well they are designed, cannot alter the underlying physical layer behavior of the single- and dual-radio architectures. This underlying behavior is governed by the following principles:

1. In architectures employing a single radio for node-to-node communications, all neighboring radios must be on the same channel for the mesh to form.
2. At the same time, during the transmit interval of a given radio, all neighbors must be silent.
   

Given the need to take turns, the throughput is cut in half for each node added. Throughput = 1/n (n being the number of nodes in a given segment of the network).

However, architectures with two radios per node for node-to-node communications are not limited in this manner. While communicating radios linking a given node pair operate on one channel, the second radio communicating with the next node operates on a different channel. Accordingly, neighbors can communicate at the same time.

One way to overcome the lack of a second backhaul radio in a node is to add a "capacity injection layer" — effectively a second backhaul. Through the addition of a second backhaul designers can move traffic off the mesh via a point-to-multipoint radio system. This leads to one of the most significant challenges in deploying this technology: identifying appropriate vertical assets for network nodes and backhaul.

Next Week: Getting Up to Speed, Part 2: Vertical Assets and Network Design