Contents
The device layer is the most tangible and approachable layer of the IoT project stack. After determining the business goal of the project, considering certain elements of hardware will be the first step. We have broken this down into the following hardware elements:
- Sensor
- Communication module
- Processor
- Antenna
- Housing
- Power Source
Sensor
In the most abstract sense, a sensor is what allows the world of IoT to perceive the world. They take data from the environment, and send it wherever is needed - whether for local processing on the device, or to be handled on a remote server.
This area of the device layer is especially exciting as we get to see both regular iteration and invention. A wide array of sensor types is already available, and new ones are being invented regularly, opening up entirely new use cases for IoT.
Sensors are very flexible in the device layer, whether addressing a single metric, or multiple. Some devices are highly modular, allowing separate sensors to be plugged in, while other devices will have sensors hardwired in.
Communication module
The Communication module, or transceiver is the element of hardware that facilitates communication (big surprise, right?) between different components or services. It can be wireless, or wired, and either specialised for a specific technology, or covering multiple. Typically we have found that it makes sense to specialise for a single communication technology. Combining multiple radio technologies might make sense for some very rare projects, but usually for a single use case, a single radio technology will be all that is needed.
Processor
Which processor makes sense for an IoT device depends on the business requirements, and the environmental restrictions. Generally it will come down to what information needs to be taken from the environment, whether it's simple metrics, or if something more complex like video or audio is required.
As of 2024, we're reaching the stage of development in the IoT industry where specialised processors for different functions are increasingly available, meaning that efficiency and security can be greatly improved over older devices. The SoC (system on chip) approach to hardware development allows various components to be integrated into a single chip, making more efficient and compact hardware a possibility.
For example, specific radio communication standards are beginning to be integrated into processors. Hardware enforced security is also becoming increasingly common. eSIMs are an area we will be keeping a close eye on.
An insightful comment on this topic
An insightful comment about this topic
Antenna
Antennas are a highly underestimated part of planning an efficient IoT device. Just buy a compatible one and throw it on, right? Well, not really. Following experts like Harald Naumann makes it quickly clear that there's a lot of innovation and value in careful consideration of the antenna.
As always this is a matter of balancing cost and effectiveness, but ultimately if the device does not have good signal, it will compromise the entire project. Fortunately, cost is not usually the limiting factor as optimal antennas can still be incredibly cheap. The cost usually will come at the hardware selection/design stage, meaning that it's worth investing a bit of time or resources to make sure the appropriate type of antennae is used to ensure signal and robustness.
Housing
Housing tends to be fairly expansive for IoT devices at the moment, allowing plenty of space for customisation to meet the needs of different projects. Given the remote nature of many projects, space is often a variable with plenty of flexibility. Of course, it's still good to keep devices as small as possible, but priority is usually placed on being robust, flexible, and affordable.
Power source
For low-power IoT projects it's quite common to rely on battery power, as it enables far more easy placement of devices or gateways. This couples gracefully with the low data requirements. There are circumstances where it might make sense for a mains power source to be involved, and renewable power sources are emerging as an increasingly available option.
Battery technology is evolving fast, pushed forward by the requirements of the wide array of industries making use of them. Li-ion batteries are the most common for long-term projects, despite the typically higher cost. The goal of many IoT projects is to aim for as lasting deployment as possible - while this may cost a little more in initial investment, it easily pays off in the long term by reducing maintenance costs.
It's quite common to find LPWA IoT devices with a predicted 5-10 years of lifespan on a single battery, for example the SolidusTech LPWAN miniUNI Truebner SMT100 soil moisture sensor making use of the Saft LS 14500 Li-ion battery to keep the device going for as estimated 5+ years.
With renewable sources of energy becoming more cost efficient, solar or wind are becoming increasingly appealing as a supplement for battery power. Yet given the reliability requirements of most IoT devices, as well as the added complexity and cost of using rechargeable batteries, it's not yet common to find these as part of a project. We're really keen to see how these options become part of future IoT projects, though!
Summary
Depending on the project, the device layer may have the convenience of having the elements above precombined and calibrated, or it may need careful selection of each element. Either way, it's important to consider the best practices discussed here. Of course, we'd encourage anyone wanting to plan or scale a business grade project to join our community to discuss details directly.
Selecting sensors that make your project durable
This comment was contributed by M2MC.eu
At M2MC have worked with a variety of sensors, covering every sector IoT is active in. The most important consideration we have found is that sensors should always be certified to work with the specific type of solution you're working on. When you're establishing projects that are expected to be deployed for 5-10 years, a huge impact on the cost is whether you need to perform maintenance or not.
Sensors tend to be vulnerable to damage or obfuscation over time, so it can be a false economy to use what appears to be cheaper at the start of the project, ending up costing more in maintenance or replacement.