Even though you might have heard more and more about it these past months, mesh network have actually been around for a while. It is not a temporary tech hype. They have considerable benefits that will bring us a step closer to a seamlessly connected world of people and things. But what is the technology behind the fancy words? This article addresses common questions about mesh networks. Most people often need to visualise the context to fully understand concepts that one has never heard of before. So to better explain what a mesh network topology is, this article will explain the bigger picture, backed up with visual diagrams.
The OSI Model — 7 Layers
To start with, we are talking about internet, networks and the connections between devices. It is thus important to introduce the OSI model which is a reference model for how systems communicate over a network. This process of communication can be divided into seven distinct groups of related functions. In a network, all devices, also called nodes, use those seven layers to communicate with each other. Each layer in the OSI model serves the layer above it, which in turn serves the one above itself. So for instance, when information is exchanged between nodes the process will work like this: a flow of data circulates down through the layers of the machine that sent the message, then it travels across the network and finally flows up through the layers of the recipient device. Why is this relevant to this article? In the context of mesh networks, this model is in use within the systems active on the network. Consequently, the most important layers to be discussed are the media layers, layer 3 and below, which are described later.
1. Topologies — Ring, Bus, Tree, Star and Mesh
Before looking at how the nodes operate with the OSI model, it is first important to define what a topology is and explain what a mesh network has to do with it. A topology refers to the virtual layout (but it does not have to be the physical layout) of the interconnected devices on a computer network. For example, computers on a high school network can be arranged in a circle in a classroom, but it is not a ring topology there. The different existing topologies are organised as follows: there is the bus, ring, tree, star, and mesh topologies. Naturally, these can also be combined to form hybrid topologies, but we are not going to address those in this article.
We want to understand what a mesh topology is and to explain, we will proceed with comparing it to today’s most used topology. That topology is the star, in which the devices are connected to a central access point (centralised network). In contrast, on a mesh network, nodes connect directly to each other (decentralised network). The two diagrams below illustrate these two examples.
So what does this imply? In the star topology, the central point has to handle all the traffic in the network. It also has to forward the info to the destinations on behalf of the sources. In contrast, on the mesh network topology, the nodes allow point-to-point or peer-to-peer (P2P) communication. This eliminates the need for a central entity. As you can imagine, these different layouts offer both advantages and disadvantages in the usage of the network. These are going to be discussed in detail later.
2. Mesh Networks
From topologies, let’s dig deeper and talk more specifically about mesh networks. To start with, a mesh refers to an interlaced structure. In networks, it refers to the many interconnection of nodes that can establish links to connect to others. Since all nodes are connected in a fluctuating web, devices can act as routers and forward traffic to others. This enables the content to hop between them until it reaches a destination.
3. Multi-transport vs. Single-transport Connectivity
Now that we have defined the larger context, let us take a closer look at the nodes themselves. We are referring here to layer 1 of the OSI model. Something that differs between mesh network providers is whether they support multi-transport or single-transport connectivity. In the former, this means that mesh networks can be implemented on several types of radios (Radio Access Technology RAT) simultaneously (ex: Bluetooth, WIFI, mobile carrier, etc…). In this case they are said to be multi-transport. These devices range from having solely access to Bluetooth, Wi-Fi and phone carries as for instance today’s smart phones, all the way to nodes that support all possible connectivity platforms. This is made possible by downloading onto devices software development kits, also called SDK’s, that support mesh multi-transport.
The implications are self-evident: A node that has access to different types of radios is much more practical. Yet, it can also be costlier and more difficult to develop. However evidently, in the future, one can expect only multi-platforms to be available.
4. Routing Protocols
The way the traffic flows or how the nodes communicate is referred to as the routing protocol. It refers to layer 3 of the OSI model. Mesh network routing protocol taxonomy presents itself as three major types of protocols: proactive, reactive, or hybrid. These types of protocols differ for the most part on the paradigm they use to perform network discovery. This ultimately has consequences on performance and scale.
The proactive protocol keeps a constant discovery process. Its nodes automatically inform each other of route (path) changes, as for instance, a node failure which causes a flow rerouting. It responds well to link breakages. It is therefore self-healing: it is more resilient and capable of recovering from a failure. Proactive protocols perform better in static scenarios, in which the network paths rarely or never changes. In the scenario of fast-changing paths with high node speed (dynamic environments), it will use up more resources, causing network traffic, increasing collision and lowering bandwidth, among other things. To keep costs low, it is therefore necessary to choose a protocol adapted for the environment.
Instead the reactive protocol establishes routes on demand. For each connection they have to inquire the whole network to search for the correct path. As a result, reactive networks scale better, but take more time to establish connections because paths may not be known beforehand.
The hybrid protocol inherits characteristics from proactive and reactive protocols, hence the name. They are often used for more specific cases where the proactive and reactive downsides are very pronounced or unwanted. Hybrid protocols adjust to conditions where either technique is favourable.
5. Full vs. Partial Mesh
Within mesh networks, there are two connection arrangements: full mesh topologies or partial mesh topologies. These can be related to layer 3 of the OSI model. When the network is said to be full, each node is directly connected to all other nodes in the network. In contrast, a partial network is when only some nodes are connected to all the others, and others are only connected with the nodes with which they exchange the most data. Again, the two pictures bellow illustrate these two examples.
Again, what does this imply? In the event that one node fails, the network has self-healing characteristics. The latter reroutes the network traffic accordingly, with no disruption in the service or hassle for the user. Full mesh topology is denser because it requires more nodes. Having more nodes means that the network is more redundant, but it also means that it gets more expensive to build your own mesh. It is usually used for backbone networks whereas partial mesh topology, which is less dense, is more often found in side networks connected to a full meshed backbone.
6. Why Today
The concept of mesh networks first appeared in the 1980s in military experiments, and it became commercially available in the 1990s. This technology has been around for a while, so why are we only starting to hear about it today? This is because in the past, mesh networks had to be wired. The topology could be expensive and complex to set up on a big scale because each node had to be physically connected to the other nodes. Today, there have been considerable advances in wireless communications. This means that wires are not necessary anymore. Moreover, short-range wireless personal network (WPAN) specifications have removed physical and financial barriers that were present in the past. Yet, for the biggest part, it is hardware, radio, and spectrum requirements; cost; and availability that has made it practical to commercialise mesh networks only now. These are the reasons why we are seeing such a boom in the commercialization of this technology.
We have some visibility on the technicalities of the technology. Now we can explore its potential advantages and disadvantages.
- No more problem of single point of failure, which is the issue in star topologies (and even worse on bus topologies). If one node can no longer operate, the network has the ability to reroute which enables it to still communicate between the remaining nodes.
- Taking the network down is impossible unless there is some kind of worldwide catastrophe that wipes out all electronic devices in the world.
- The network works with minimal infrastructure and can therefore be deployed faster at a lower cost than traditional infrastructure.
- Since the devices in a mesh network can retransmit signals further, they have an ability to connect thousands of sensors over a wide area (ex: cities). Other instances include operating in areas with large crowds (ex: concerts, festivals, and so forth) or connecting devices in remote areas (ex: in mountains or in public transport under the ground), and many, many more.
- There is no centralised authority in a mesh network. For that reason, some people compare it with what the Internet was back in the day: localised, anonymous, citizen-based, secure communication.
- Market and regulatory forces make mesh networking difficult to deploy.
- Mesh networks can replace Wi-Fi providers, phone carriers and other middle men that provide connectivity to people. As a consequence, the middle man loses so for them there is no financial incentive to develop this technology.
- Mesh networks are tough to manage and troubleshoot. For big networks, one needs a strong mesh technology to make it worthwhile. This can be difficult to find.
- Battery life impacts the availability of the nodes. For instance, a phone empty on battery could disrupt the network, causing more routing overhead and less reliability.
- The cost of deployment can sometimes be problematic in certain scenarios. However, it can be redeemed by downloading a software development kit, also called SDK’s, which enables you to become a participant node in the whole mesh instead of building it from scratch.
9. Final Thoughts on Areas of Application
There is no authority in a mesh network. This decentralisation opportunity opens up the possibilities of hundreds of new forms of technologies and business ideas that will disrupt markets. Especially with the up-and-coming field of IoT — Internet of Things, mesh networks will start to take huge dimensions. Use cases range from smart metering to object clustering. People also predict that mesh networks will be found in sectors where implementations of robust safety rules are on the rise. For instance, in logistics, mining, oil and gas, utilities, and energy. An increase in the usage of mesh networks in light commercial application is expected too. Examples include large warehouses, agriculture, distribution centres, but also vehicle-to-vehicle connections, etc…. This is because the areas to cover with traditional Wi-Fi are too large and expensive to connect with traditional infrastructure. But most importantly, this technology has a huge potential for humanitarian purposes. In the event of storms or earthquakes, local infrastructures often get damaged which causes people to loose communication means. Mesh networks enable connectivity to not be affected in these situations. Another example is manual pumps for water. When they get damaged, people can spend months with no water access. With IoT and mesh networks, local communities can have the pump repaired in only a couple of days.