A Guide to Wireless Mesh Networks
Date: May 2, 2017
Product Name: Wireless Mesh Networks
Ever generation seems to have their own wireless network standards and connectivity types, each heralded as the solution to high bandwidth connections between networked devices. There’s been standards like N, 2.5GHz, 5GHz, MU-MIMO and many other acronyms that can’t even be pronounced. The latest technology that’s bust onto the mainstream scene is the Wireless Mesh Network.
The first question that people always ask when they hear about wireless mesh networking is ‘what is a wireless mesh network?’. That is actually a very good question and one that is rather easy but long to explain, since it requires a bit of history. In the simplest terms, a Wireless Mesh Networking (WMN) is a very old idea about how to connect various devices that are spread over a specific coverage area, but it is an idea that has been updated to take advantage of new technology. It is also coming down in cost, becoming accessible to a wider base of potential clients.
A typical WiMAX Network Setup (Wikipedia)
Before we continue though there are two main flavors of WMNs – Wide Area and Local Area. If you live in the countryside a hundred kilometers away from the nearest city – and you are not using a satellite-based ISP – you are likely running on a Wide Area Wireless Mesh Network – probably a variation of ‘WiMax’ to be specific. In this article however, we are going to focus in on the so-called Local area. The basic concepts are the same, it is just the scope is smaller and more localized.
So with that out of the way, back in the old days of networks there was no wireless networking. It was all wired. This was perfectly fine when dealing with smaller areas and a limited number of users, but when you scale a network up to even a medium-sized office level the wire run length and number of wires that were required became a nightmare.
This is where dumb ‘hubs’ came into being. An Ethernet Hub allowed a single Ethernet wire to be run from the main server to a floor/location/etc., and then had all that locations Ethernet cables terminated at the hub instead of the main server. The downside is they were just that: dumb. They were nothing more than a little device that allowed multiple shorter RJ45 cables to connect to a single RJ45 backbone. In other words, they are the equivalent of a repeater. They accept data packets and then push them on down the line. If more than one user was pushing data packets at a time they would be queued, but beyond this nothing more was done at the hub level.
If you scale up this wired network concept further, in a bigger environment you could have a multiple hubs connected to one main hub for the entire floor of a building… with each story running back to a server via one single cable. If you were really lucky the backhaul was fiber optic, but most were not so lucky and ran off RJ45. In either case, having all these dumb hubs connected caused numerous issue such as packet collisions that slowed down performance, even more so than just trying to cram dozens or even hundreds of users over one main connection. For those unaware of the term, a packet collision is when two data packets come in at the same time and basically cause a ‘car crash’ requiring both packets to fail and have to be resent.
The answer was to swap out the hubs and install switches. Much like a Hub, a switch allows multiple RJ45 cables to be connected to it, but it is a bit smarter and does not simply broadcast the packets to all of its connected cables. Instead, it knows what is on the end of each of its ports and only transfers the data along the path it is ‘supposed’ to be. This decreased latency, but at a certain point switches do have a tendency to be overwhelmed during peak times.
Thus, for extensive network configurations the answer was to swap out (some of) these semi-smart switches and install routers in their place. Much like a switch and a hub, a router allows multiple connections but has even more processing power and automatically talks to all the other switches/hubs/routers on its network so as to create a routing table. A routing table is literally a file that it could read (and update) to route a given packet through the least congested path from Point A to point Z. For example, if network congestion is bad in say group B, the router could route information through the less congested group C router and so on.
This noticeably improved overall performance and network throughput, but did come at a price: an increased cost of roll out as a router is much more expensive than a switch and a hub. This is why all but the largest networks ever use routers, and instead go the ‘managed’ switch route which offers a bit more “brain power” than a standard switch but at a much reduced cost compared to routers.
So what does this all have to do with a Wireless Mesh Network? In the simplest terms, each mesh access point is the equivalent of an old school router. Each one talks to every other mesh AP on the network, and they all work together to get every data packet from Point A to Point Z in the least amount of time and with the lowest chances of a packet collision. That basically is what a WMN is, which is a wireless version of the multi-router setups that any professional IT worker has dealt with.
The only major difference is that the access points on this network do not talk to each or the device on the network via wires and rather use wireless technology. As such, they are also capable of handling a lot more devices per node than their wired counterparts, as they really are a combination of a wireless access points with the power and brains of a wireless router.
The ASUS HiveSpot
The next question is invariably: why use a WMN? Why not just stick a bunch of wireless routers on a network and have them do their own thing instead of using specially designed equipment? This one question is a bit more complicated to answer.
Firstly, WMNs really grew out of a couple of specific fields. The first is the hotel service industry that wanted to offer ‘free wireless access’ in each room, but did not want to place multiple routers per story of the hotel. The other is large office buildings where typical wireless and wired setups simply will not work. This is because metal and concrete materials impede wireless signal and thus require the installation of many expensive routers, and going the wired route would cause too many headaches and be physically difficult to install.
Basically, it all boils down to cost. Cost of installation, cost of maintenance, and cost of securing all those routers and making sure that none are compromised. In this regard, non-WMN setups can get expensive, tedious and difficult to work with when dealing with anything but simple setups. Meanwhile a WMN on the other hand is made up of fairly simple access point nodes that don’t have massive capacity, can easily be mounted on the ceiling and are really inexpensive when compared to business-grade wireless routers. Installation is as simple as running power to the mesh APs, installing them where you need more coverage for a wireless network, and letting them figure it out amongst themselves how to best route the data packets from Point A to the Point X (with point X of course being the WMNs main router that connects to the server or to an external ISP, etc). Of course, they do still need a certain amount of setup and configuration but it is radically simpler than setting up multiple routers to work seamlessly with each other. Think of this as a hive mind approach with multiple smaller devices doing a more efficient job than several larger complicated ones.
So what does a WMN look like and precisely how does it work to provide long distance performance? As the name suggests, a Wireless Mesh Network looks just like a fishermen’s net or a mesh. Each and every node talks to every other node and each node keeps the others updated on the status of the entire network. Going back to the old school router analogy, each node keeps a table of the quickest pathways in the network so that any device that connects directly to it can get its data packets to their destination as quickly as possible.
Unlike a wired router that relies solely on its router lookup tables – which are only periodically updated – Wireless Mesh Network nodes can also do what is called ‘Extremely Opportunistic Routing’ (ExOR) in that they broadcast a small sample of the data packets to all nodes in range, see which nodes respond, and more importantly which respond the quickest. Then and only then do they choose which node is to broadcast all the data to the end point. These up to the second routes help the mesh navigate around sudden network congestion areas that wired routers never could do, as the additional overhead and the limited number of wires would just have created too much congestion to be worth it.
Unlike modern non-mesh WiFi routers, mesh routers also act much like an old school dumb hub meaning that each node serves as a repeater for the original wireless network. They do not create additional secondary networks like a typical range extender or secondary router will. To talk to each other they use a different network channel, a different frequency (for example 5.8GHz), or even a different IEEE 802 protocol (for example 802.11s) that is only utilized for internal comms. This ‘secondary’ network is not seen by end-users nor end-user devices and instead it is only used for the Wireless Mesh Network nodes to communicate and transfer packets amongst themselves.
To picture all this, imagine a user at one end of the mesh connects to the wireless network. The nearest WMN node is providing the highest signal strength so the wireless device automatically connects to it. When the user then say opens up a YouTube video, the nodes themselves decide which among them is best suited to repeat those packets until they reach the last node that then pushes it on to the users device. However, unlike a dumb repeater, wireless mesh networks automatically do load balancing and will decide in real-time which packets are sent to which node to get to the final destination. All that matters is that they get to the user as fast as possible and in the right order, which the nodes also take into account.
While there is increased latency from the numerous hops that the data packets have to travel to reach their final destination, the overall performance is usually greater. This is because each node is spaced so that the signal strength (and overall connection speed) is optimal. Compare and contrast that with a device which is at the extreme edge of a wireless network, and the speed it is usually able to connect, and you will see that the increased latency is more than made up for in overall performance increase.
It is also worth noting that unlike typical range extenders, any wireless device that connects to the network has no idea which node it is connected to, nor does it care. If the device is in room 1 on floor 2, and the twelfth wireless mesh AP node is the nearest, it connects to that node as the signal strength for the entire network is strongest there. If the device is moved to another floor or even just room, it may hop on to another node.
This all happens seamlessly as the wireless devices that connect to the WMN are passive, and it is the nodes themselves which decide what device connects to what node and how to get the data packets to their destination. Compare and contrast that with an environment that has a single wireless router and even just two range extenders, where the device itself (and the user) has to manually connect to one of the range extenders networks, and then has to manually connect to a different one when the device is moved into what would be more optimal range of the different range extender. Yes, a Wireless Mesh Network can indeed provide rather hassle-free connectivity to a lot of devices and do so almost seamlessly.
The only downside, besides increased latency, is a Wireless Mesh Network must have a special wireless router as the starting point. These routers have custom firmware that allows them to fully harness the power of a Wireless Mesh Network. More specifically, they really are not meant to be connected to directly. Instead, they are usually setup so that they can only talk to all the nodes in the mesh network; much like a spider will sit at the center of its web. While it would be technically possible to use an off-the-shelf wireless router and load a custom firmware on it, the reality is that a WMN needs both nodes and a main router to work optimally. This too does increase the overall setup price as the entire wireless network needs to be upgraded, or better yet created around a WMN setup in the first place.
A Linksys Velop Mesh Node
The next question is always: if WMNs are so great why don’t we all use them even in our homes? Why bother with consumer-grade wireless routers or using multiple range extenders? Well that is a good question. First and foremost, there is no real reason to not do so, but they are costly when dealing with a typical home. That is why WMN setups are mostly found in a professional environment. In other words, WMNs scale up incredibly well and offer increasing return on investment the bigger a network is, but don’t really scale down all that well. The ROI is pretty terrible when dealing with the typical-sized house, where a single router will cover most of the house and maybe one or maybe two range extenders are needed to cover any dead spots. Here the upfront cost of a WMN does not seem like a great deal, especially with wireless technology advancing so quickly!
They do however offer some rather unique advantages to the home user that can pay dividends for knowledgeable consumers. Firstly, while a large home can indeed make use of a single router and bunch of wireless access points, all those access points talking to each other are going to make for a crowded network and can cause bottlenecking pretty quickly. More importantly the typical wall socket access point has rather limited horsepower, and when someone sets up a second wireless AP to talk to another AP which in turn talks to the wireless router, the overall performance is going to be rather lackluster. More importantly, such a setup is serial in nature; Access Point 2 has to always be connected directly to Access Point 1, even when Access Point 3 is not being used and could provide better speeds.
A WMN access point on the other hand basically has as much processing power as the typical router, can reroute data packets around a heavily congested portion of a network using any and all WMN devices within reach, and can easily handle a lot more devices at a time before performance can degrade. The downside however is that WMNs not only cost a lot more, but do not really offer very high wireless performance. Instead, they offer consistent performance over incredibly large areas of space and in really heavily congested areas. All of which is pretty uncommon in the typical home.
Once again, when dealing with the typical home, the typical networking enthusiast will find that two strategically placed AC3200 or greater routers communicating with one another will provide higher performance. But such a setup does not scale up to cover as much area as a WMN setup could do, and can be very finicky in setting up. For example, after a power outage the secondary router may default back to its original state and stop functioning as a wireless Access Point. After that, reconfiguration is in order and that’s typically not something an average home user will be able to easily accomplish. Meanwhile all those WMN devices will reconnect, rebuild their routing tables and be back up and running with very little, or even no user intervention.
The appeal of a Mesh network really does come down to total cost, what the home user is willing to spend, how great an area they need to cover, the total speed they are looking to get, and how much fiddling they are willing to do. However, this equation is changing. Every year the cost of entry is getting much lower, putting WMN devices within reach of the mass market, and as such are quickly skewing the equation in their favor over typical range extenders.
In the near future, we are going to take an in-depth look at some Wireless Mesh Networking devices, which can indeed be a great solution for home users as they have been created with the home – and not business – user in mind. So stay tuned and be on the lookout for these reviews as this may indeed be the Year of the Wireless Mesh Network!