Getting to grips with the IoT

The Internet of Things (IoT) has become the buzz phrase of the decade, but there is still confusion as to what it really means. One can argue that it is little more than a collection of small end-nodes connected to the Internet, and at one level that is an accurate description. In most cases, however, it is a lot more complex than that.

For a start, what are these end-nodes? True, they could be just a sensor attached to an industrial machine measuring and reporting temperature or pressure or whatever. But the end-node could also be mobile, attached to a package being transported through countries. Here, it could be making its connection through a number of different ways, including cellular, which brings up issues of coverage plus the different cellular standards such as 2G, 3G and so on. Or it could be using a short-range communications standard such as Bluetooth, ZigBee or Wifi to a node on the vehicle that then connects to the Internet, and that could be through cellular or satellite. This is actually quite a common situation, as transport and logistics have become major IoT applications.
Another complicating factor is security; a small sensor connected to a machine may not need much security but it may be attached to a network that does need security and thus is a potential route into that network. As an example, consider a garage forecourt. The tyre air pump may be a thing connected to the Internet, and on the face of it does not really need security. However, its connection may be through the same forecourt system that is also processing payments.
The other part of the complexity issue is the environment in which the nodes are working, the types of machines to which they are connected, and the type of data that needs to be sent. The application may call for regular bursts of small amounts of data or large bursts at longer intervals, or any mix of this. The distances may be long or short or somewhere in between.
There are also new standards coming through to cover the IoT, such as LoRa, a wide area network standard. This is becoming an option for utilities collecting meter readings or for, say, local councils checking where the bins need emptying. Here the data packets are small. But for surveillance application where the things are CCTV cameras, live video streaming may be needed and thus a 4G connection may be the answer. And there is no rule that the connection has to be wireless; many IoT applications will use wire to make the link.
Robustness is another consideration. Again, an understanding of the application is needed. If some data are lost, then depending on the situation they may need to be sent again or it could be fine to move onto the next packet. Some applications will need redundancy and other technologies to increase reliability. Others may not. That decision has to be based on cost and use-case.

Signal path
There are many hardware topologies for the IoT and the one chosen will depend on the application with factors such as the type of network available down to how many things actually need connecting. A typical signal path will start with some type of sensor and this will be linked to a gateway either through wireless technology such as ZigBee, Bluetooth, Wifi or other radio system, or through a physical wire. The link will be to a gateway, which in turn connects to the Internet. Again, there is a choice of how that connection is made, whether through a broadband wired system, through a cellular network, through Wifi, or via a satellite network.
The data are now in the cloud, but that again can have different meanings. People can develop their own cloud infrastructure. They can also use off-the-shelf cloud products, such as the Digi Cloud. This manages all the connections to the remote devices and they can be grouped in various ways, such as by type, use, or geography, or even for a particular customer. A good cloud system will allow users to service their devices via over-the-air updates.
Whatever cloud method chosen, it is basically a structure connected to the Internet, whether the structure is a data centre or server room, or even multiple server rooms. Here, applications are needed to make use of the data. After all, collecting these data is of little use if something sensible cannot be done with them. This is the driving force of the installation. There are lots of things that can be connected to the Internet, but there is little benefit in just making the connection.
Consider a kettle. On the face of it, connecting that to the Internet seems pointless. But say that kettle is in the home of an elderly relative who is known to make a cup of tea at least every couple of hours during the day. If there is a gap of say three or four hours without the kettle being turned on, then that could send an alert to a relative or carer to check up on the person.
With the Digi Cloud are a number of application kits that can help users get a feel of what can be done. Each kit comes with a gateway for connecting to the Internet, a cloud account and a GUI for controlling the IO on the board. The board will have buzzers, motion sensors and so on. This can act as a starting point from which a user can develop a custom application.
The danger though is gathering too much information without a proper plan of what to do with it. A good application would present an overview of what is happening with the ability to drill down to more detail if and when something significant happens. In a factory set-up, for example, the screen could show an overview of the factory workflow all coloured green, but if a sensor registers something beyond a pre-set threshold it would turn red, then the user could click on that and receive more information about what is happening in that particular area.
Companies such as Digi have designer services to help with these software and hardware applications. The Digi Cloud also contains a number of APIs that will interface with standard HTML type applications to let users develop their own web interfaces. It also links to Salesforce.com and other off-the-shelf tools. Basically, the cloud infrastructure is an intelligent database that can be shaped to what the user needs with some standard tools that can help.

Software chain
From a software point of view, this same signal path is obviously to some degree dependent on the hardware chosen. To connect a sensor directly to the Internet, it needs to be addressable with an internet address. But if it is just collecting data and sending it via ZigBee or Bluetooth, then that might not be necessary. Security protocols may be needed, again dependent on the application.
Data storage also needs to be considered. For low amounts of data, there may be no need for local storage, or maybe just a small system. Often, though, the storage is done in the cloud or, if needs dictate, on a local server at the user’s premises. Generally, with a cloud service, there is no need for local storage as the cloud also provides the security and backup needed. This can become complex if the data are stored on several servers, as in the example above with some data on Salesforce.com or other applications, and some on the user’s own servers stored locally. Cloud services such as Amazon’s tend to be more about storage, whereas cloud services such as Digi’s tend to be about managing the data.
With all this IoT talk, people should be asking why they are collecting the data and whether it is appropriate for these data to be connected to the Internet. They should look at the risks involved and be asking whether or not the data are useful rather than collecting data for the sake of it. Given how much data can now be stored in a small space, there is a real temptation to collect more and more data, whereas the real issue is doing something useful with these data.

Wireless technologies
The main connection means include Wifi, Bluetooth, ZigBee and proprietary RF in the short range, and cellular technologies for the longer distances. The benefits of Wifi include having a lot of devices already supporting it. It has a high throughput and existing infrastructure. There are also no roaming charges and it can use MIMO technology. Latency can be an issue with Wifi when it comes up from sleep mode. It doesn’t connect initially and that can eat up battery life.
Bluetooth is lower power and lower cost, and there are no cellular charges. No infrastructure is needed and it has a reasonable throughput. The frequency hopping is good for finding a free channel. The downside is that only about eight devices can be connected. To get more, multiple PANs need to be set up.
The ZigBee mesh network does not have this problem as a thousand devices can be on the network. It also needs no infrastructure, is low power and relatively simple. There is good latency when coming up from sleep. Network management can be a problem if too many devices are allowed to communicate at the same time. But it is nice and reliable, and it is self-healing; if a device goes out it will try to heal the network by finding another route.
The cellular technologies of 3G and LTE have a good range and throughput, and an existing infrastructure. There are, however, places where they don’t work and roaming can be problematic, and expensive, though hardware and data costs are coming down. Like Wifi, it uses MIMO technology. Among the main disadvantages are that it is a power hungry technology and the approval process is complex.
The main infrastructure for setting up an IoT network is already available; there are companies renting out servers with applications for managing data. From a hardware point of view, there are lots of companies, such as Digi, making modules to provide connectivity to sensors and routers. Digi will provide the complete system including the software with the idea of reducing the effort by the user and getting to market more quickly. The Digi range starts with a suite of tools for connecting devices supporting any type of local network. The company provides an infrastructure that is accessible anytime and from anywhere, is scalable according to needs, and provides protection from cyber threats. There is the capability for integrating any application and supporting any type of user device.
The key benefits of this include no coding or custom integration, and no infrastructure to operate or maintain. It is reliable, scalable and secure, with easy application integration and open APIs. And there is out-of-the-box compatibility with Digi hardware.
For those who want to start from scratch, it may take a long time to set up. But there is no need, as users can now pick and choose from a suite of products and services to create IoT networks that suit their applications.