IoT in products for extreme environments
To ensure future growth and jobs, Danish industry must excel at developing robust products that are able to be part of the Internet of Things (IoT). In this context, it is obviously important that companies can have an overview of wireless technologies such as Wi-Fi, LoRa, RFID and IoT-A. At the same time sensor types, cloud computing and potential data analysis models have to be managed, but all of this will be covered in another article, because often it is important to start at the bottom and look at the overall lines in a design. In this article, we will describe where special attention needs to be paid when developing products for IoT. First, it has to be clear why the IoT concept requires robust and reliable products, and then it must be clarified and determined which steps shall be taken to achieve this. Product robustness does not begin with carrying out tests after the product development is finished. It needs to be included already at the concept phase, where for example, user involvement through the test of interactive mock-ups can be used to write accurate requirements specifications.
Today, there are a lot of technologies and platforms available, that can help you to acquire data from a thermometer to a cloud. For most people it is fairly easy to find out how to carry out data analysis in the cloud on the temperature data from this or other sensors. To a large extent, the necessary technologies for IoT concepts to work already exist. As the title in the latest issue of Elektronik og Data (Electronic and Data Magazine) so concisely put it: “IoT is more than technology”, because IoT is about creating value. For a society like Denmark, savings can be achieved using automation. We have increased our productivity for decades through the automation of factories and production plants, but IoT can provide pervasive automation, i.e. automation that spreads throughout our society, which is not limited to a specific location. A good example of this is the Bigbelly litter bin used in public parks, which automatically sends a message when it needs to be emptied. Thus, optimal routes can be automatically generated for waste collectors with associated savings in time and resources. However, the challenge with automation is that you have to be able to trust the IoT product if the system is to function. If a machine only works half the time in an automated factory, the entire factory will break down or the quality of the manufactured products will be compromised. In the same way, an IoT network will not be able to deliver the desired automation performance, if the individual parts do not work reliably.
One of the challenges we often face, is how to establish contact between the end user and the developing company. It may sound trivial, but input from the users is a central point in modern electronic development. Even when there is no user interface on a small sensor, it can still be in the way or be aesthetically unattractive, so it will be removed, and therefore can no longer supply value. Such design faults can be avoided by incorporating end-users as part of the process, already in the early phases of product development. For example through workshops or testing of interactive mock-ups in the context in which the product will finally be used, so a mapping of the users’ daily routines and interaction with the product can be made. In this way, the development company acquires knowledge about the context that the product will be incorporated in, also known as context validation, where the following question is answered: What service shall the product supply?
More and more companies focus on making a service, rather than making a product. It is still a physical product that comes out of the factory, but it is the service that the product supplies, which creates the value and thus the earnings in the company.
When moving from products to services, a central concept is availability – or in plain talking, uptime. In other words, can you rely on the service to run all of the time? For example, for IoT systems, the service may be limited due to the operating status of the servers, whether the sensor is functioning, whether there is a connection between two devices, whether they are able to re-establish contact automatically, etc. It is therefore important, that the system has status monitoring, if signals have not been sent from a sensor (if a litter bin is empty or full), it may be faulty, and will need to be checked. At the same time, it is essential to ensure that the individual parts are as robust as possible. If a sensor in a litter bin erroneously reports that it needs to be emptied, the waste collectors’ planned route will no longer be optimal. A way to ensure robustness is to evaluate the influence of environmental, mechanical, and electromagnetic factors.
One of the things that especially affects the introduction of IoT, is that the sensors are used in all sorts of feasible and infeasible places. This means that the environment, in which the electronics are exposed to, often varies a lot. This is especially true for temperature and humidity, but also true for sweat, chemicals, salt, or fungal spores. This has an impact on the choice of the design. For example, when replacing traditional electric lightbulbs with LED lights, the thermal design is more important than the humidity design if the product is to be used indoors. On the other hand, if the product is to be used outdoors, because LEDs produce much less heat, the outdoor lamp is cooler and can be affected by humidity, which can lead to short-circuiting and a error in the lamp. It is therefore important that knowledge about mission profiling is acquired, i.e. the profile of the environmental, mechanical, EMC and other effects that the product will be exposed to. The challenge is intrinsic with the benefits of IoT, which delivers pervasive automation, i.e. IoT is ubiquitous. To be truly ubiquitous, it means that many IoT products will have to be designed for indoor and outdoor use, in sunshine and in snow, in laboratories and in pigsties. All very different environments, which require different focus on design. Finally, you have to be aware that when a product becomes mobile, there are often abrupt changes in its environment, for example from a cold car to a warm room, or vice versa. This means that it is also relevant to look at the design for dealing with thermal shock and condensation.
Another thing that changes is that the sensors are used in all sorts of feasible and unfeasible ways. This means the product is exposed to knocks and vibrations, while it is perhaps squeezed or twisted. Because the products are no longer connected to a cable, there is mobility and thus a changed loading profile in purely mechanical terms. One of the things that often happen is that the product has a greater risk of being lost or knocked over the edge of something. Many products today are worn down by being in the same position for 20 years, exposed to the same low-frequency vibrations, but mobile products are often exposed to impulse effects, which also contain waves with higher frequencies and brief but high amplitudes. It is precisely this effect acting on small sensors and oscillators for radio communication that will be more damaging than the normal low frequency vibrations. The fault may also develop from overstress rather than wear, where the product fails the very first time it is dropped rather than after several years of use.
Above all, it is important to understand the operation in the majority of sensors. These normally generate very small electrical signals, and it is not unusual for these signals to be in the microvolt range. This means that the signals need to be amplified via an amplifier circuit before they are processed by an analogue to a digital converter. It means that any electromagnetic noise that is present will also be amplified, and it is precisely here that products have a great deal of problems with noise immunity. The majority of IoT products used today are wireless. This is because it is often cheaper to include a wireless chip, which communicates in a licence-free band, than it is to include 5 m of cable to connect the product to the nearest network. One of the challenges this entails, is that radio signals generate electromagnetic energy, and this can interfere with the sensor’s electronic circuit. Thus the requirement that the sensor should only measure when the radio signal is switched off. The introduction of a wireless system also leads to another major problem with immunity – the radio signal receiver. A radio signal receiver normally works with signals down to -120 dBm in sensitivity. In other words, just one femtowatt (1 x 10-15). Such small signals are obviously weak and highly susceptible to interference. For noise, which lies outside the communication band, filters and good RF design can ensure a degree of immunity, but for noise that lies within the same communication band, there is often not a lot that can be done. Spread spectrum, diversity and error-correction codes can help to a certaindegree, but you cannot come close to the same sensitivity level, if the interference is in the same band, as if there is no interference. Often, a strategy is chosen where you just wait a couple of seconds (or a couple of days in extreme cases) before sending the data. If this is not a possibility, parallel communication frequencies can be used, so for example, both 868 MHz is used as the primary communication and a 433 MHz system or GSM system is used as a backup.
One of the dominating trends in recent years in particular, is low power wide area networks – LPWAN. These networks are often based on very narrow band of communications. Because the narrow communication bands lead to a low bit rate, the bits will also be longer. Since LPWAN systems are often used the same way as base stations, for example for mobile phones, it will be what is known as multi path propagation. This means that the receiver receives a signal, which is reflected by buildings and thus can be delayed and staggered in time in relation to each other. This increases the risk of inter symbol interference, i.e. the previous symbol interferes with the current symbol.
Even though IoT is considered completely new, we should remember that it is based on many known technologies. This also means that there are a lot of supporting standards. For example, you can use the IEC-60068 series of standards for environmental testing. These include IEC 60068-2-14 used for environmental testing of thermal shock, IEC 60068-2-30 used for condensation, IEC 60068-2-31 used for mechanical drop testing, and IEC 60068-2-55 used for bounce testing. Finally, you can also look at immunity by using a reverberation chamber. This can be done, for example through IEC 61000-4-21.
So it is clear that not all technical challenges are new to IoT, since there are already several existing standards that can be used to ensure the products have the correct functionality and thus support pervasive automation.