These days we see frequent, sometimes incredulous, forecasts of billions of IoT devices by 2020 and warnings of the disruptive change that will ripple across organizations and markets, but there is little mention of the underlying technologies involved and why one is better than another. Which ever numbers you believe - Cisco's 50 Billion (above) or MIT's more conservative 25 Billion (below) - they all need a means of connecting to something, and that "means" will predominantly be wireless. In this article we look into the wireless access technologies being considered for different use cases by IoT solution developers today, and why.
Not all IoT devices will be connected wirelessly. Some will use Ethernet, others Power Line Communication, but the majority will rely on some form of wireless access network. Here the options are plentiful and include several new ones you may not have heard of such as 6LoWPAN, LoRa, wirelessHART 802.11ah, Z-Wave and others. They range from RFID to WiMax and everything in-between, with Wi-Fi, 4G/LTE, Zigbee and Bluetooth the frontrunners today.
There’s good reason for so many options – IoT applies to virtually every industry, and the range of application requirement vary greatly from case to case. So, to choose the right tech, you need to break down the networking requirements in each use case.
The selection of the right Wireless access technology for your project is a central design challenge, which comes down to balancing several inter-related requirements of the IoT application: Power consumption, range and proximity, number of devices, message size and time-sensitivity, and of course device and network costs. Let’s take a look.
How sensors, devices and machines are powered has implications on the applicable access technologies, the initial cost of the device, and the total cost of ownership over its operational lifetime. Battery operated devices must conserve energy to minimize battery replacement or recharge cycles, which in turn limits the amount of data they can transmit or receive. Energy harvesting technology is still expensive and immature, so there are few battery powered devices which augment the power source with harvested energy, though it is likely this will increase as IoT penetrates the industrial farming sector, more Smart City applications emerge and new energy harvesting techniques evolve.
Personal Area Networks (wearables and consumer electronics etc.), Smart Home (home automation and appliances etc.), Smart Factory (process control, building automation etc.) and Smart City (street lighting, parking, waste management etc.) applications each involve varying distances between the devices and the network communicating with them. Effective range is the result of many factors including the transmission frequency, modulation scheme, and transmission power.
In general, the more dense the modulation, and the higher the frequency, and the shorter the range becomes. Likewise the ability to penetrate building materials and other obstacles is similarly affected. Depending on the use case, the ability to penetrate obstacles, versus the requirement for line-of-sight is an important consideration.
Take your infrared TV remote control for example, which requires line-of-sight within a range of about 10 meters. RF-based alternatives such as Bluetooth, Zigbee or even Wi-Fi offer great potential to revolutionize home entertainment systems, because they do not require line-of-sight, have equal or better range, and can transmit much more data. So instead of needing to point the remote at the target appliance, you could be controlling it and other devices, from another room. Plus, with bi-directional communication and more bandwidth, you could be browsing the program guide or playlists on the remote itself which now might be your smartphone.
Many familiar wireless access technologies assume a relatively small number of edge devices in the network. Bluetooth has historically been used for one-to-one connectivity, such as connecting a headset to a smartphone, and nowadays wearables to or one to very few devices such as synching multiple mobile devices in near proximity. Bluetooth classic operates in a point-to-point or point-to-multipoint master-slave relationship, where one master can communicate with up to 7 slaves. BLE allows more devices, practically about 10 to 20.
Wi-Fi access points are typically capable of handling 30-40 active mobile users and can potentially support up to 4096 connections, but not tens of thousands. But other technologies were designed for much larger numbers of connected devices. For example a 4G/LTE six sector base station can provide coverage for 10-15 thousand connected devices, though it may only support up to 1000 concurrent voice calls or 50-100 data users. While Zigbee supports up to 65,000 nodes. Depending on the IoT application in mind, the numbers of connected devices is an important factor driving the technology selection.
Many IoT use cases require very little bandwidth, and / or require infrequent communication. 100Kbps is ample for sensor and meter applications. For this reason there exist a large number of technologies that could be suitable. However when designing the communications framework, it is important to consider how the application might evolve in future, in order to ensure the solution can scale.
The time sensitivity of collected data is a very important system design criterion. If the information exchange needs to occur in real-time or near real-time, this has an overriding influence on the preferred technology and the likely power requirements, even if the message size is small.
If the data transmissions are frequent, the device may need to remain connected at all times. Many applications, such as process-control or wearables fall into this category. Whereas a smart dumpster only needs to report that it is full, once, and it doesn’t even need do so immediately when it becomes full – a few hours delay may be fine. In such cases, it is therefore feasible to turn off communication most of the time, and only wake it up on a schedule for a brief transmission and then turn it off again. How about a vending machine, reporting it is nearly out of a certain type of beverage? Once, within an hour of reaching that state may be adequate, depending on the restocking lead-time, and remaining inventory of the machine, but that exact moment is not necessary. In contrast a self-service coffee maker in a high-volume cafeteria might need to report it is running low on ingredients rather promptly.
As you can see, each use case is different, but the immediacy of communication affects whether or not the device’s communications can be shut down or put into a sleep state, versus staying connected all the time. When the data transfer is not time-sensitive, message size becomes less of an issue, as it can be stored and forwarded periodically even at low data rates.
The highly probable displacement of infra-red by Wi-Fi and Bluetooth, is a great example of the disruption that IoT can and will cause in many industries. Recently some Wi-Fi vendors have incorporated Bluetooth and Zigbee into their access points which may enable them to compete differently in a variety of retail, smart home, smart energy and smart city applications, perhaps creating new niches, or expanding their addressable market. Whichever the case, the IoT market opportunity is so big and so disruptive that solution vendors in every field need to realize they are about to become exposed to new competitors from unexpected places, with new ways of solving previously unsolved problems, as well as new ways to solve old problems.
To survive through 2020 and beyond, almost every technology solution provider and many non-tech solution providers, need to fully embrace IoT technologies and new paradigms - either to protect markets they already play in, to compete in a disruptive way in existing markets they don’t play in, or to pioneer totally new markets.
But for many, this means getting into networking and wireless communications technologies that are unfamiliar territory. It also means they must build more intelligent devices which rely on embedded software, and this opens up a whole set of new challenges, such as how to upgrade devices with new software versions, how to gather data from devices and analyze it in the cloud and many more.
Vendor solutions will likely need to be a lot more open than before as well, to leverage existing frameworks, platforms and services, rather than creating everything oneself, and missing the window to market. There will be many build-buy-rent tradeoffs to be made.
With a decade of experience building embedded systems for networking and telecom products, embedUR can help you make the right technology choices, develop your IoT solution, and get to market on time, and on budget, with the least possible risk.