Mesh Network Topology for IoT Applications

Mesh Network Topology for IoT Applications


Network topologies come in different shapes and sizes. Each network topology has its own advantages and implementation areas. While networks of the past have mostly been wired, wireless networks have gained significant traction in the last fifteen odd years. In this article, different network topologies are highlighted with more focus on mesh networks. We will also see how mesh networks fit in the Internet of things (IoT) domain.

Different network topologies

Let us briefly take a look at the different kinds of topologies.

Bus Topology

Bus topology. The bus topology relies on a single cable called bus, which joins multiple nodes. A single server is used to control the bus. The communication can be either unidirectional or bidirectional. The server sends out a packet on the bus and the appropriate node accepts it. The biggest hassle with this topology is that if either the bus or server fail, the whole network crashes. Also, there is a limit to the length of the bus and number of nodes that can be added to the network.

Ring Topology

Ring topology. A ring topology has a wire connected to two adjacent nodes. The packet is unidirectional and each node checks if the packet being transferred is meant for itself. If not, it passes the packet forward. Again, the problem here is that failure of single node results in the following nodes denied access to the network.

Star Topology

Star topology. A star topology has a single server connecting to multiple nodes in a one-to-one manner. As a result, the server gets to choose who to send packet to and who not to. While this may seem like a good network, and though it is for most applications, the disadvantages are quite glaring. For one, if the server fails then the whole network goes for a toss. Also, constant communication with one node can lead to other nodes waiting for a long time for their turn. This topology needs multiple management algorithms not only to avoid starvation, but also to communicate based on priority. Another major drawback is the range of the network. As each node is connected only to the server, they have to remain in the range of the server. For example, a wireless star network can be found in offices where laptops are connected to the routers. Take a laptop away from the router and the range falls rapidly.

Mesh Topology

Mesh topology. In mesh topology, every single node is connected to multiple other nodes. Packets jump from one node to the other until it reaches the server. As the path of data is never fixed, the failure of one node does not result in failure of the entire network. The data just re-routes itself through a different path. This is known as self-healing. Consequently new nodes added to the network become a part of the it without any kind of configuration, or self-forming as it is called. There are a high number of packets travelling in the network simultaneously, which helps in increasing the chances of packets reaching their destinations. However, high redundancy also results in collisions and data corruption. The network itself can be further divided into smaller networks to manage data being sent around.

Terminologies explained in detail

Packets. Data is sent in form of group of bytes. These groups are called packets. Each protocol has a fixed number of bytes per packet. The size of packet affects the performance of the network. A packet is divided into two parts, the header and the body. The header contains the details that are required to identify the packet. It usually contains the source and destination address and gives some indication of what data is contained in the packet. The body of the packet contains the actual data. For instance, in a street lighting scenario, the body would indicate the status of light such as ‘ON’, ‘OFF’ or ‘DIM’.
Hops. When data jumps from one device to another, it is known as data hopping, packet hopping or just hops. These hops decide the range and reliability of the network. Imagine there are five street lights to a gateway device. The data from the first street light hops via other street lights sequentially to reach the gateway.
Nodes. Any device that is going to receive or send a packet is called as a node. A node could be a utility meter, a light, an electrical appliance, a computer or rather anything with the ability to transmit and/ or receive data. A node can be wired or wireless depending upon the network it is in.

How mesh works

A mesh network has every single node connected to multiple nodes. All these nodes transmit packets wirelessly and they are in range of each other. This ensures packets transmitted by one node is received by multiple nodes. The packets received are then forwarded, until they reach their expected destination. As there are multiple copies of the same packet in the network, the chances of the packet reaching its destination are quite high, even if individual nodes in the network fail. The probability of packets reaching their destination is also increased because mesh networks do not have fixed paths for packet hopping. Failure of nodes does not result in the network crashing. A single point of failure can occur if the mesh network works on the coordinator router architecture like the ZigBee network. This means that if the coordinator of a network fails then the whole network crashes and you are back to square one.

Data hopping through one node

Data re-routing after node fails

New nodes can be added to the network without facing hiccups. These nodes add themselves to the network without much effort and begin transmitting and receiving data just like the remaining network. This is the self-forming feature of the mesh network. Adding new nodes does not affect the remaining nodes. However, each protocol has its own limit of nodes that it can handle. For example, the ZigBee mesh can handle around 100 nodes per network.

Mesh implementation comes with advantages…

Self-healing. In case one node malfunctions, the packets re-routes on its own to reach the appropriate destination.
End-to-end reach. The range of the mesh can be extended up to multiple kilometers with the use of appropriate algorithms and technology.
Scalable. New nodes can be easily added to the network without disturbing the existing architecture.
Adaptable. The mesh network can be adapted to a wide variety of networks from where we want to get data. This includes networks for home automation, smart lighting, street lighting, smart metering, medical wearables and industrial automation.

…and disadvantages

Architecture. The architecture of the mesh network differs from protocol to protocol. As a result each protocol comes with its own sets of disadvantages. To add to it, most protocols are not interoperable with each other. This results in having to deploy a single mesh and then finding if it suits you or not. For example, the ZigBee mesh has coordinator router architecture. This means that there is one coordinator per network that has multiple routers. On the other hand, the DigiMesh protocol has only routers in the network and there are no coordinators. Needless to say, these two protocols are not interoperable. Each protocol has its own set of benefits and drawbacks.
Redundancy. There is a high redundancy in the amount of data that is sent and received in the mesh. Algorithms are needed to effectively manage this redundancy, otherwise it affects the speed of the network. This limits the area of usage of the mesh protocol. For example, sensor or meter data can be collected without much hassle, but audio or video data cannot be transferred by mesh network; at least not the current implementations of mesh network.
Lack of interoperability. Different mesh algorithms do not work with each other. For example,  ZigBee protocol will not work with DigiMesh, which will not work with Smartt Mesh. This does not let the same network use different protocols. This means if you have implemented DigiMesh for home automation then you cannot use ZigBee based sensors.

Implementations in IoT

The mesh network can be used in multiple scenarios as talked about in the advantages. Here we will throw light on two completely different scenarios.
Let’s take the example of medical wearables. Assume that these are battery operated devices monitoring the patient’s heartbeats. These sensors would be spread across the floor of a hospital, each with a unique ID to identify the patient in question and the data about the heartbeats. A mesh network would help in collecting data and transferring them to a central location to be monitored by a doctor, intern or nurse. Even if the patients moved about on the floor, the data would form new routes and reach the server. A central monitoring station would mean that the person who is monitoring the patients would not have to move about. Also, this data can be backed up on the cloud in order to maintain the history of the patient’s condition. Thus the mesh network now becomes the best method of communication for remote monitoring of patient data.
This is another example for industrial automation. Large industries have a wide variety of human operated or automatic machines that communicate with each other continuously. Needless to say, wiring becomes a huge hassle in industries. To top it off, laying down new wires is an equally tough task. A mesh network can be successfully deployed here in order to enable the machines to communicate wirelessly. High range technology along with repeaters would help in overcoming the solid metal obstacles that would hamper communication. There would be no cost of extra wiring and communication would happen without change of the protocol of the machine. Once again, the status and the data of all the machines could be monitored from a central server. All this data could be uploaded to the cloud in order to give remote monitoring capabilities.

Mesh network will undoubtedly be the future of IoT communication

So, what does this mean for the networking domain in general and the machine to machine (M2M)/IoT field specifically? Mesh network is without a doubt the future of computer networking. The M2M/IoT domain uses a large number of sensors to collect and upload data to the internet. Mesh network is an efficient way to do it. The future will progress in the direction of interoperability and battery efficiency.
At the present moment, there is no single standard for mesh network and going forward this will have to change if we are to collect vast amounts of data and upload it on the Internet. As sensors will be battery operated, the power management and energy efficiency part will play a big role in the mesh algorithms of the future. A winning combination of mesh algorithm will be one that can be truly plug-n-play using nothing more that market bought components, replete with collision management algorithms, power management features and user friendly user interface similar to most of the websites on the Internet.

---