There are many different wireless machine-to-machine (M2M) technologies behind the Internet of Things (IoT). One of these technologies is the cellular network. In 1979, Nippon (a Japanese company) created the first commercial automated cellular network. In the 36 years since then, cellular networks have evolved from simple wireless voice communication to having complex broadband capabilities.
Today, cellular networks are designed primarily for mobile devices, and are thus tailored toward voice and video calls, and other high data-rate applications. They have been optimized for these broadband usage patterns. Because of its widespread coverage, IoT developers are turning to cellular to support their apps. But is this a good idea? Is cellular suitable for most M2M connected devices? Is there a better alternative? We will examine these questions (and more) within this article.
Elements Of Cellular IoT
Perhaps the most commonly cited benefit of using cellular IoT is ubiquitous coverage. Cell towers are already up everywhere, and the major networks cover roughly 99% of the U.S. population. So, if you want to deploy your sensor network at the drop of a hat, cellular networks are readily available.
Low power, wide-area (LPWA) networks, which were designed for the low data rate IoT applications, must be installed and deployed in order to be operable. But the consideration many engineers and IoT developers forget is that these networks typically don’t need to cover more than a few square miles of distance. If you need a network to cover an office building, a school campus, or a farm, you’ll be able to effectively do so with a LPWA network using only a couple of gateways. So, if you’re willing to install your own gateways or wait for an LPWA public network to be installed, the broad geographical advantage of cellular IoT becomes less critical.
Cellular networks gained significant deployment in the IoT space when 2G networks were created. The reason application developers and network operators liked 2G was the store and forward architecture. The radios could wake up, send their data, receive meant for them, and go back to sleep. This saved a great deal of battery life, because the radio didn’t have to be on all the time waiting for messages.
When 3G and 4G were created, store and forward architecture was removed and replaced with an always-on approach. This means that the radio is always listening for a message that tells the device when it is getting a new piece of information. Because of this near-constant communication with the network, the battery life takes a hit.
With an LPWA network, sensors go to sleep and wake up when they need to send data packets. Since the radio is asleep most of the time this allows batteries to last between five and 10 years (depending on the technology). Cellular providers have recognized that battery life is a concern for those in the IoT space, and they understand that cellular is insufficient for very small bits of data. In response to these concerns, they’ve proposed a new protocol called LTE-M. In an interesting twist, it’s actually designed to bring back some of the elements of 2G cellular transmission, like store and forward.
It’s important to note that even if these capabilities return, cellular-connected M2M devices still require a great deal more power and energy than LPWA networks. 4G-LTE data-only power requirements are ten times more for transmit (~120mA vs. ~12mA) and 1000 times more for standby (~2 milliamps vs. ~2 microamps) than the new emerging LPWAN network systems. This is a challenge that cellular providers are going to have to face.
LTE-M is set to be proposed late next year, and it will then need to be ratified and adopted. We will likely see the first iterations for LTE-M in late 2017 or 2018. So, if you need a solution that is more battery efficient now (or in the next three or so years), you should probably look into low power, wide-area networks.
When it comes to the latest iterations of cellular—4G and LTE—there are 44 bands that operators can broadcast on. So when a device tries to make a connection, there is plenty of potential bandwidth.
But remember, LTE networks are designed for broadband communications. They’re optimized for high quality voice and data, including video capabilities, like watching Netflix or making FaceTime calls. We like to describe this as a huge 8-lane highway completely filled with dump trucks that are designed to move around big data. So, what happens when you don’t need big data moved around—you just need to send small data packets? Well, instead of those dump trucks being filled to the brim, they have only a shovel-full of data. Clearly, this would be a very inefficient transport scheme.
It’s important to remember that many M2M applications are sending less than a couple hundred bytes every day. An alarm panel, for example, sends only 40 bytes each day. Even if you assumed that a device was sending 30 bytes an hour, that’s only 22 kilobytes a month. This is substantially less than cellular companies’ data plans, which are meant for gigabytes of data.
So, following our analogy, what you really need for sensor-based network and many of the new IoT applications isn’t an 8-lane highway filled with massive data loads—you just need a bike lane! That is what LPWA networks provide: an efficient way to move around small amounts of data. Remember, when you think of the Internet of Things as a whole, you aren’t talking about a couple hundred radios with sensors—you’re talking about tens of billions of devices. And if all of those devices showed up on an LTE network tomorrow running TCP/IP, it would completely crush the network.
Cost Of Cellular IoT
An LTE module purchased in very high volume will cost between $25 and $30. An LPWA module purchased in very high volume will cost about $10. As LPWA networks mature—within the next three years—the price per module should dip between $5 and $10. That $10 difference can be a lot of money when it’s multiplied by the thousands (or millions) of modules that may be required.
The LPWA networks that are being deployed today (mostly in Europe) are charging in the range of a few dollars a year. Compare that to cellular data fees today—in the range of a few dollars a month—and you’ll see a big difference when you start multiplying these numbers by the millions of devices deployed. Remember, if you’re deploying your own infrastructure, as many organizations do for their LPWA applications, there isn’t a monthly recurring cost.
The other thing to keep in mind with some cellular systems is that there are other costs, like SIM fees, charges for activating and deactivating, and more. These types of changes can rack up if users aren’t careful. Additionally, some networks will charge users whether they are transmitting data or not. LPWA networks may do the same thing, but the cost is so much lower that it may be a nominal fee.
What type of network is best?
In the end, you need to determine what kind of data you’re sending and how much data you’re sending, so you can select the right tool for the job. Following the analogy above, if your applications use a dump truck of data, then cellular IoT may be the direction you want to go. But if you’re using only small amounts of data—a shovel-full in said dump truck—you’d be better off taking the bike lane and using an LPWA network.
It’s also important to note that cellular and LPWA aren’t the only types of M2M networks available. In fact, we’ve written a white paper about the seven most popular M2M networks, and we suggest you take a look to better understand which network option best suits your organization.