So you want to design and deploy networks of wireless IoT devices? Sounds simple in theory; you just connect them to the network. But what network, and how? By their very nature, IoT devices are small, have low power capabilities, and will most likely be sited in the most challenging environments for radio propagation, writes IoT Now contributor, Guy Daniels
Amongst the most important considerations are the network and the antenna. In other words, what type of wireless network will be used for connectivity, and consequently what antenna – or multiple antennas – are needed and will be compatible with the device’s design and physical properties? Do you use an external antenna, or embed one into the module itself?
First, the network. At the far end we have the cellular networks. Although there could be upwards of 20 billion connected devices by 2020, cellular networks will support only a subset of this – as little as 2 per cent according to some analysts. Of these, 2G (GPRS) devices remain the most prominent, although 4G (LTE) should overtake GPRS by 2019, with 3G largely bypassed. But the cellular operators have plans to dramatically increase their share of the market, such as the forthcoming LTE-M standard, which optimises LTE for M2M and IoT use cases.
The alternatives to cellular for remotely located devices are satellites and Low Power Wide Area (LPWA) networking technologies. We then have WiFi, and the various flavours of the 802.11 standard. Whilst the ac variant is becoming more popular in the home, the IEEE recognises that it is not the best for wide area device connectivity, and so they are working on the new “ah” variant.
On a more local scale, there is a plethora of short-range wireless standards from which to choose, including Bluetooth, ZigBee and Z-Wave. These options typically use a central hub or gateway to connect to a router for onward transmission over Wi-Fi, fixed lines or cellular.
Just to complicate matters, the Bluetooth SIG is developing an evolution of the standard that promises support for longer distances, higher speeds and mesh networking, taking it out of the “room-based” environment and even out of the house itself.
The list of connectivity options is huge – and of course, each option comes with its own antenna requirements.
It’s the antenna’s job to pluck the necessary information from the full radio spectrum, and it does this by locating specific wavelengths. Wireless technologies use a set of internationally agreed frequencies, which are inversely proportional to wavelengths. So, if your device is communicating over Wi-Fi 802.11g (the most common type for Wi-Fi IoT), it will be using the 2.4GHz spectrum band. It therefore needs an antenna to pick up signals with wavelengths of about 12cm to pass on to the radio unit. Usually, the wavelength corresponds to the physical size of the antenna – so in this case a typical half-dipole antenna must be 6cm long. Go shorter than a quarter of the wavelength and signal strength and bandwidth fall fast.
An antenna is perhaps one of the most critical components of the product; if it is designed well, the full performance of the radio device will potentially be realised. However a poorly designed antenna will greatly reduce the maximum range of the radio and additionally reduce the usable data-rates at medium ranges.
The three main factors of IoT antenna design are: size, power and reliability. These three factors are often interconnected: it is crucial that devices in remote locations operate reliably – the cost of physically servicing such devices is prohibitively expensive – and being out in the field generally means low-power consumption (they could be required to be operational for ten years). And, of course, these devices are usually very small. Often, IoT modules are no larger than a coin, which places immense design pressures on manufacturers.
Types of Antenna
The most common design for small IoT devices, where physical space is at a premium, is prefabricated chip antennas. They are relatively easy to design and are also cost effective when producing high manufacturing volumes. However, they are not overly efficient and are somewhat restricted in bandwidth adoption. Another factor is that care is needed in the overall design of the module and related circuitry if maximum performance is to be obtained.
Chip antennas can be as small as 2mm square to over 20mm in length, and can be manufactured on a ceramic substrate or a small piece of multi-layer PCB (Printed Circuit Board). However, the chip itself is only half of the antenna system – the PCB forms the other half and acts as a ground for the antenna. As such the chip’s maximum performance is only going to be realised if the PCB and layout are precisely as stipulated on the antenna manufacturer’s datasheet – a detail often overlooked in practice.
For manufacturers of more complex IoT devices, especially those that contain multiple sensors and are destined for large-scale manufacturing where the lowest unit costs are required, PCB antennas could be the best option. The PCB manufacturing process has been well-honed in recent years, and there is a sizeable market in designers, simulation software, reference architectures and support. In fact, there are many free reference designs available. PCB antennas can be planar monopole or have a meandered conductor design, where the antenna is traced in a pattern on the circuit board.
The most popular types of printed antennas include patch antennas, inverted-F antennas (IFA), or planar inverted-F antennas (PIFA). They take up less space than a dipole antenna because they use the ground plane of the circuit board to help them radiate. Variants include flexible printed circuit (FPC) and stamped metal antennas.
The next step in performance would be to source a proprietary antenna design, either in-house or outsourced to specialists. This becomes a more viable option if the exact use case of the device is known from the outset, to enable to antenna to be designed for maximum efficiency.
Many OEMs don’t have the resources to support an RF engineering team, or are able to fulfil the high production volumes needed for custom antenna designs, which is why many of the leading antenna companies are expanding their portfolios to support specialist IoT requirements. By using a “standard” antenna from an established vendor, an IoT OEM can be assured of a pre-tuned, quality product that can be easily integrated onto a PCB assembly, and which are typically multi-band enabled via a single antenna. These relatively low-cost, repeatable designs will feature many standard commercial off the shelf (COTS) components. In addition, these firms can also provide systems-oriented field application engineers to help with faster prototyping and time-to-market.
Where budgets are tightest, wired antennas could be an option. They are extremely low cost and design-wise very flexible, but require decent design skills if they are to be optimised to fit into a module. They do need more testing and simulation regarding electromagnetic performance, and may even lead to a rethink about the physical housing of a device, if added bulk is to be avoided. Another reason to use wired antennas is if multiple wireless technology transceivers are required for the IoT device. In such a case, several different antennas can be attached to the module board via micro-coaxial cable, enabling the various antennas to be situated apart from one another for better performance.
However, where budgets are more flexible and confined space is not a problem, then the maximum performance is still to be found in a whip antenna. Yes, the antenna is external, requires a physical connection with the module PCB, and is expensive to buy, but it will give the best performance. Also, if the module housing is constructed of metal, then an external antenna is going to be essential.
And you wonder why IoT antenna design is generally regarded as an exercise in balancing trade-offs?
Another design factor is the choice of single-ended or differential antennas. In single-ended signalling, the transmitter generates a single voltage that the receiver compares with a reference voltage. However, significant EM interference can be generated by a single-ended system and as such this design is not always feasible for optimal circuit board construction. Differential antennas use two complementary signals, and are either a pair of wires twisted together or a pair of traces on a circuit board. These are usually more suited to low power applications because of their treatment of multipath issues – as waves propagate, they bounce and deflect off objects, interfere with other signals, start to decay quicker and create multipath signal noise. Yet differential systems require more space.
For many IoT applications, physical space is limited, hence higher frequency systems are preferred – the higher the frequency of the radio system used, the lower the wavelength and hence a more compact design of module and antenna is possible. Lower the frequency and you need to allocate more space in your device. Yet lower frequency systems can communicate over far greater distance. We’re back to trade-offs again. We are going to need cm-Wave and mm-Wave frequencies (of 60GHz and above) to enable more efficient and sophisticated antennas in tiny form factors.
A final note of caution on antenna design: there are numerous other factors to consider and be aware of when settling on an antenna design for your IoT module. Space is always limited on a circuit board, and so antennas need to coexist with other electronic components, whose proximity can cause EM and interference issues. PCB antennas also need precise etching in pre-determined patterns; get this wrong and you’ll never have an optimised connection.
But, just because you want to have multi-radio connectivity doesn’t always mean you need separate antennas. A single antenna is possible that is tightly integrated into the device, but it’s a highly complex feat of EM engineering that requires extensive signal simulation to ensure that the required frequencies and bandwidths are covered whilst also maintaining the required isolation between bands and achieving the best all-round performance possible. Look no further than your high-end smartphone for such a device, which will have at least five different GSM and LTE antenna ports, plus Bluetooth and possibly two WiFi frequencies (and we no longer have to pull out a plastic whip antenna; it’s all integrated into the handset, invisible to users).
IoT brings similarities, but also differences. Take the connected car sector as an example. In terms of M2M/IoT there’s obviously vehicle tracking (inherited from the telematics sector) and which will soon be mandated by law in certain regions as part of emergency positioning programmes, such as eCall in the EU, but there’s an increasing number of Vehicle to Vehicle (V2V) functions coming on to the market now.
However, these connected IoT units need to coexist with a plethora of other radio systems. On a high-end, in fully equipped car, it would not be uncommon to also have: AM/FM, DAB radio, satellite radio, 2G, 3G and 4G cellular, in-car WiFi, Bluetooth, GPS, radio for remote keyless entry, a variety of on-board sensors, and even digital terrestrial TV.
And so the IoT antennas have to share limited space on the outside of the car – which, being essentially a metal box, is a mobile Faraday cage – hence the typical ‘shark fin’ antenna that sit on top of most vehicles.
Where space is at a premium, such as in small sensor modules, there is a need to combine antennas to support different frequencies. Ethertronics, for example, combines an active antenna, RF systems and chip technology into a turnkey plug-and-play module, which dynamically senses and optimises the antenna system, without any external control. Operating from 700MHz to 3GHz, it can cover cellular, Bluetooth, WiFi and ZigBee.
“With the continued expansion of the M2M and IoT markets, we’re seeing increased demand for advanced plug-and-play solutions that need to meet higher expectations for the users and designers,” said Olivier Pajona, chief scientist for Ethertronics. “EtherModule 2.0 is a testament to our continued dedication to customers by developing highly advanced solutions to help them differentiate their products and stand out in a competitive market by providing maximum performance and reliability.”
So is antenna design the domain of the Black Arts? Not exactly, but it is certainly a highly skilled process that requires time, money and expertise, and is a critical, yet often overlooked, aspect of the IoT value chain.
Having determined your choice of connectivity, then gone ahead and designed the optimum antenna, there is still another important consideration for maximising IoT radio effectiveness. This time, though, it’s out of the module designer’s hands, as it’s a network issue.
For those looking to deploy their own wide area IoT networks, using LPWA technologies, the first stage is to run extensive simulations. Siradel is one firm that is focused on this sector, with its S_IoT dedicated LPWA simulation software. Place your devices within a 3D visualisation of the actual environment and then see how signal propagation maps out. Get it right in software and you minimise the requirements for actual field site surveys and reduce deployment costs.
Cellular operators have their own in-house expertise when it comes to deploying radio networks, but even they need to rethink their approach when faced with millions of new connected devices expected in the coming years.
Germany-based Core Network Dynamics (CND) believes it has a solution to solve what will soon become a traffic crisis for IoT on cellular networks. The company was formed in 2013 to commercialise OpenEPC (Enhanced Packet Core) technology developed by the Fraunhofer FOKUS research institute. Using Network Function Virtualisation (NFV) – a rapidly emerging field of network design for telecoms operators – CND believes its OpenEPC platform can be used to isolate traffic between devices, by using virtualisation techniques to create private networks – effectively, networks within networks. According to the company, a complete mobile network infrastructure in software can now be run on commodity hardware as small as a Raspberry Pi.