WiFi Meshing Without the Mayhem

15th Jul 2017

There is an axiom that we all follow in the wireless networking industry which states “Cable wherever you possibly can, mesh only when you absolutely must”. 

There is also the principle in any wireless network design that you need to “get off the air and onto the wire as soon as possible”. 

Why? Well, there is far less contention, and no RF interference, on a fast, switched fibre and copper network infrastructure than there is across the air, therefore this makes absolute sense.

So, with the above in mind, are mesh based wireless networks useable and supportable either as standalone solution, or as a component part of your overall mobility architecture design?

Mesh terminology

First, we must explain a little on the technology and terminology in use. The access point nodes that make up a mesh network are referred to as Root APs (RAPs) and Mesh APs (or MAPs) - they are also sometimes known as Mesh Portals and Mesh Points. 

The RAPs/Portals have a connection to the wired network and act as gateway between the wired and wireless network.

MAPs/Points have no wired connection and pass client traffic in wireless hops in as efficient a manner as possible to/from the hard wired RAPs/Portals. The links between the nodes are formed using the 5GHz band as these provide more non-overlapping channels of significantly lower occupancy than the crowded 2.4GHz band.

The MAPs and RAPs that make up the nodes in your mesh network will use routing and meshing algorithms to form links between each other. I won’t stress the detail on this process here, but all typically follow variations on the theme of “least cost path”, where hop counts, latency, link quality and packet loss are all factored into the equations that dictate the mesh links. 

Should a node fail, or link qualities change in the mesh, the system can dynamically refresh the connectivity between the nodes, ensuring minimum latency and maximum throughput and reliability for network traffic. 

Professional installation is essential

A modern 802.11ac access point on the end of a properly installed, well terminated piece of CAT5e or CAT6 cable, connected to a gigabit full duplex switch port, is of course the preferred way to deploy wireless networks. This gives a reliable, high throughput and low latency point of egress from, or ingress to, the wired network and can support numerous wireless clients running network heavy applications.

By comparison, the reliability and throughput of access points that are backhauling client traffic as part of a mesh network can be prone to variable performance and to remove this variability their layout and implementation needs to be planned and designed with some significant thought and care.

At the heart of the problem is the fact that a wireless LAN connection is not a full duplex one; it can transmit and it can receive, but not at the same time. 

Add to this that the same radio is also providing client access, again half duplex, and the issues become clear.

Optimum hops

The upshot of this is that raw data throughput effectively halves with each hop between access points, so you need to keep the hop count to the wired network as low as possible. 

In reality, this means that you should plan for no more than two hops before you “hit the copper” and a single hop would be even better.  

Fear not though, as a clear line-of-site between antennas counts for a lot here, and the achievable distances for a solid mesh connection can be greater than you would think.

So what and where are the applications for mesh networking?

Although sometimes promoted as an option, the use of a mesh network within a building  is an anathema to me, and to most, I am sure. If you need to get power to the device, you may as well get network to the device.

You will undoubtedly need other connectivity in the future, so bite the bullet and get another wiring cabinet and switch installed and get it connected as a part of your larger fibre and copper network.  

Where mesh comes in to play is to extend wireless networks out of buildings and across far wider external environments that sit outside the range of your wired networks. 

These can be campus environments in education or business, parks and city centres, sporting/leisure facilities and marinas, through to more industrial environments such as sea and airports. 

When extending a wireless network in mesh mode to an external area, there will always be two limiting factors for the number of nodes there can be:

  • You need somewhere to physically mount the devices
  • You need a method of providing power to the devices

Your MAPs are will typically be installed to the exterior of buildings, where multiple points of access to the network exist, and the RAPs are going to be installed on street furniture, light masts, CCTV masts and outlying buildings that may have power available, but no network.

So, what are some of the factors that you should you consider in your planning for an external mesh network?

  • Try to gain an understanding of client requirements and of likely traffic loads. In a private network, where all the clients are known, this is relatively straightforward. For wider networks that have a large element of public access, this is a little more tricky, but look at implementing bandwidth limits, so that everyone can get a decent slice of the pie.
  • Find a good balance for the number of mesh portals and mesh points. Too many mesh points having to backhaul through a single mesh portal becomes a choke point for traffic and a significant risk as single point of failure, so spread the load.
  • Where you are looking at stretching the limit on hop count, consider a hybrid approach using both point to point bridging and wireless LAN hardware to give a solid full duplex backhaul for a remote wireless mesh cluster.
  • Think about your band/channel plan and design a scheme that mitigates known interference and excessively busy channels.
  • In your planning and design effort, simulate node failures and look to ensure that there will always be multiple routes available for traffic to pass between the nodes to and from the wired network.

  • In your planning and design effort, simulate node failures and look to ensure that there will always be multiple routes available for traffic to pass between the nodes to and from the wired network.

Plan to succeed

With appropriate planning, surveying and testing to work into your design and a longer term view of the expectations, demands and requirements that will be placed on your network, it is perfectly feasible to implement large areas of external WiFi connectivity that you, your colleagues, visitors and guests can rely and benefit from.

What Next?

For advice on designing or upgrading your WiFi network and associated systems, to the deployment of security solutions like Next Generation Firewalls and Endpoint Network Security, please contact Ensign Communications for a chat with our technical team.