Ultimate Guide to Internet of Things (IoT) Connectivity

Two bluetooth devices showing the range between them

Overview

The last few years have seen a revolution in connectivity options available for Internet of Things (IoT) devices and applications. As more and more systems become intelligent, the need to obtain real time data and control system to improve efficiencies, reduce energy consumption and automate older system has only increased.

As an example, IoT systems intelligently monitor the production and storage of food to avoid spoilage from temperature control failure. Another application is the monitoring of saline bags in ambulances.

These cold-chain monitoring approaches can be applied to food, chemicals and many other products that require tight temperature control. We’ve created products for all of these applications, and have a lot to share about how to go about choosing the right wireless IoT connectivity.

This guide takes you through the technologies, challenges and features of IoT connectivity so you can make a better decision.

Short to Medium range IoT Connectivity

Wireless can be categorized in a lot of ways, but categorizing it by its range makes a lot of sense because range is the major limitation and tradeoff in every wireless system. Range competes with power consumption and throughput, the other two critical wireless parameters. In most IoT devices, the more range, the better, but high throughput is not usually a requirement. Hence looking at range simplifies our approach.

Achieving good range while keeping power consumption constrained and throughput high can be challenging and involves constant tradeoffs. The first set of wireless IoT technologies we’ll look at are mostly short range, usually 200m and less. We’ll then discuss other technologies that provide better range, but have increased costs and complexity.

Bluetooth Low Energy

Bluetooth Logo

Bluetooth Low Energy is an evolution of the Bluetooth technology most of us know and love. We've used Bluetooth Classic in Headsets, Speakers and Smartphones to stream music and have phone calls. Bluetooth Low Energy, also called BLE, is an update to Bluetooth specification that allows extremely low power data transfer while not providing audio support.

BLE Technology Summary
Bluetooth LE
Range 30m (typical) to 1km (long range)
Frequency Bands ISM 2.4GHz - Worldwide
Throughput 1Mbps, 2Mbps with Bluetooth 5
Power Consumption Extremely low - microampere average
Networking Star (aprox. 3-20 devices), Mesh (with BLE Mesh)
Smartphone Connectivity Yes
Market Penetration Tens of thousands of interoperable certified products. Billions of BLE devices worldwide

Although it was originally designed for transferring small amounts of data, BLE has evolved thorough various iterations to support more capabilities such as speed, range and mesh networking. This has allowed BLE to compete with many other technologies and work better for wirelessly connecting IoT systems.

BLE forms a star network, with a Central device connected to many Peripheral devices. As opposed to Classic Bluetooth, BLE can form star networks with many devices connected at the same time, far beyond the 7-8 devices in a piconet. For example, gateways we've developed can connect 16 sensors simultaneously without running mesh. Even more devices are possible

With the latest BLE mesh specification, BLE competes better against Zigbee and other mesh solution and allows a very large number of devices to be connected simultaneously.

Bluetooth Low Energy is a complete protocol that defines everything from the low-level hardware to the data transfer at the application level, with the ability to customize the services being provided. This flexibility has allowed IoT product designers to create very low power devices in a variety of applications, from consumer products to sensors, medical devices, and industrial machinery using standard devices and software.

Multiple BLE peripheral devices connected to a single Central device

Aside from its very low power profile, BLE’s main advantage compared to most other wireless solutions is its ability to communicate out of the box with billions of existing Smartphones, Tablets, Gateways and smart devices. Because of Bluetooth’s ubiquity, it reduces the system cost. For example, an Android tablet can act as a gateway without needing to create new hardware.

BLE can be integrated in IoT products to provide connectivity at a low cost and low power while making use of the security and connection features in a wide variety of applications where the data throughput requirements are below 2Mbps, which covers a significant number of applications. Chipsets can cost $2 or less in volume, and in some cases can be less than $1.

What range can you get with BLE? Although Bluetooth’s range was originally designed to be around 30m, recent updates to the Bluetooth specification allow you to achieve 1km line of sight communications. ALthough we've included BLE in a category of short range wireless, BLE is no longer the extremely short-range technology it once was with the right design.

Should you choose BLE as your Internet of Things connectivity of choice? That depends on what you’re building and what you’re looking to achieve. If connectivity to Smartphones is needed, then BLE should probably be the top of the list because it's low power and can be customized for any number of applications. Some of the advantages of using BLE for IoT include:

  • Easy integration with Smartphones and Tablets which allows easy device control and internet connectivity
  • Decreased engineering effort due to using an established specification designed for robustness
  • Large BLE vendor chipset selection and support with highly integrated features
  • Very low power wireless enabling long operational lifetime
  • New specifications such as Bluetooth 5.1 enable improved location and tracking applications
  • Uses 2.4GHz band for worldwide compatibility
  • Mature Well-known technology with broad support and billions of units

Some disadvantages of BLE:

  • Range limitations in some cases depending on design
  • Because it uses the 2.4GHz band, it can suffer from interference from other devices including WiFi
  • Some higher costs due Bluetooth SIG listing costs ($2.5k to $8k in fees)
  • Mesh Support is early and may require integrating with other solutions
  • 1Mbps to 2Mbps throughput may not be sufficient in some applications and power consumption will require low data throughput

We have a few articles, including one on selecting BLE chipsets. There are many companies that can provide the BLE chipset for your IoT product, which is a big advantage. The right device depends highly on your end application (IoT Sensor, IoT Controller, etc.).

One of BLE’s recent improvements has been the development of BLE Mesh capability which allows devices to communicate among themselves, not just in a star network configuration. In fact, the lack of Mesh capability in BLE was one of the reasons some IoT developers avoided it as a connectivity option – range and reliability are critical factors and mesh networking can be critical to avoid these issues.

BLE Chipset Vendors

More Information:

Zigbee

Zigbee Logo

Zigbee is another protocol that typically operates in the 2.4GHz band. Zigbee’s advantage has always been its Mesh networking capability and low power. Mesh networking allows Zigbee to increase its effective range since the devices can forward data from other devices farther away. In addition, Zigbee is very low power and devices using Zigbee can run for very long time.

Zigbee Technology Summary
Zigbee
Range 30m to 100m
Frequency Bands Typically ISM 2.4GHz, 915Mhz/868MHz (rare)
Throughput 250 kbit/s
Power Consumption Extremely low - tens of microamperes average
Networking Star (aprox. 3-20 devices), Mesh (with BLE Mesh)
Smartphone Connectivity Yes

Zigbee has been on a roller coaster for some time, with its popularity as an IoT connectivity solution changing over the years. For a long time Zigbee adoption was hampered by its own specification limiting the use cases along with other issues. Since the release of Zigbee 3.0, Zigbee has recovered some of its popularity for applications beyond just lightbulbs and switches. With Zigbee 3.0, Zigbee has shed some of its inflexibility to allow for more custom communications that is more suitable for many IoT devices.

These days, integrating Zigbee in a product is easier than before because Bluetooth LE vendors such as Nordic, TI, NXP and others offer Zigbee support in their BLE System on Chip (SoC). This means that you can choose a single device, evaluate Zigbee, BLE or Thread, and then choose the right protocol or protocol combination that works for you.

Whether to use Zigbee depends. If you’re looking to integrate into a Zigbee ecosystem, then it’s clear choice. If you're planning to deploy many devices in proximity, then the mesh capability will help tremendously. Otherwise it can also work with BLE as well and run for a long time.

Zigbee also is well established protocol with an ecosystem for development and certification that enables creating products that can work well.

Thread

Thread Alliance Logo

Thread is another wireless protocol created by Nest, the consumer products company, as an IPv6 low power mesh solution. Originally closed source, Thread was opened up with the OpenThread implementation. The Thread protocol is now promulgated by the Thread Group, an alliance of various manufacturers and Nest Labs (acquired by Google).

Thread is built on proven technologies such as IPv6, 6LoWPAN, IEEE 802.15.4 and adds features that are needed to establish links and route packets.

Thread integrates well for consumer products, such as sensors that augment Nest's existing systems, but can be adapted to many other applications. Its security features are helpful

Like Zigbee, many manufacturers provide support for Thread in their devices, allowing you to test and evaluate.

More Information:

Proprietary Radios

Before Bluetooth LE appeared, proprietary radios ruled the waves. Custom wireless protocols in the ISM bands (868MHz/915Mhz/2.4GHz and others) were and remain extremely popular. These radios typically run proprietary protocols that are proprietary to each company and each company.

For some applications a proprietary radio for IoT connectivity makes a lot of sense: using the 433MHz/868MHz/915MHz bands are a great way to achieve greater range and lower interference because the 2.4GHz band is busy with Wi-Fi, Microwave ovens and many other devices.

Using lower frequencies for RF transmission allows for greater range purely because of physics. The lower frequency propagates longer. For example, a radio running in the 915MHz band has 8dB more link budget than the same radio with the same output power at 2.4GHz. Even more can be achieved by going to lower frequencies

You may be wondering why not just use 915MHz or the other Sub-1GHz bands? This is due to a few issues. First, it’s hard to build a worldwide compatible device. The US allows 915MHz, while Europe allows 868MHz band. These bands are far enough apart that designing RF that allows either band means performance is degraded in both bands. You can use 433MHz but can't support multiple bands due to the antenna. Most designs support a single band due to performance and cost considerations.

The advantage of using 2.4GHz is that you can deploy a product worldwide with one design. Even better, the antenna can be very small which is a necessity in many designs.

One of the most challenging aspects of using proprietary radios in IoT connectivity is that they require you to do a lot of work developing and testing the wireless radio and wireless protocols. So, with great power comes great responsibility. This is something that BLE and similar protocols help you avoid – operating in the worldwide 2.4GHz band and with a common protocol, it’s much easier to connect devices together.

Until recently BLE was also limited in output power and that meant Proprietary Radios achieved a lot of range even beyond the physical reasons.

Proprietary Radios can achieve quite a lot of range, usually around 2km, with more is possible in some cases. This long range is possible because of the 20dBm output power you can achieve, the benefit of the improved propagation and large efficient antennas. Because most IoT product developers don’t want to reinvent the wheel, it’s common to reuse the wireless protocols provided by the chipset vendors. Some of the most popular Sub-1GHz proprietary radio vendors for Internet of Things connectivity are Silicon Labs and Texas Instruments.

In the past, Sub-1GHz IoT radios were available as radio chips only – you had to add an external processor. But in the last few years, because of technology changes, these devices have been changing similar to what happened with BLE. Vendors now offer a complete package – a radio with processing and peripherals, which allows you to create complete IoT devices with a single chip, lowering complexity, cost and size.

There’s a few other things to know about this. One reason why 2.4GHz is popular is that the higher frequency means smaller antennas. Smaller antennas allow IoT products to be smaller. 915MHz and lower have longer wavelengths, which require larger antennas to be efficient.

The following frequency bands are available:

  • 169MHz - US and Europe
  • 315MHz - US
  • 433MHz - US and Europe
  • 868MHz - Europe
  • 915MHz - US FCC
  • 2.4GHz - Most of the World

One has to be very careful with developing in these radio bands. While using a protocol such as BLE helps keep you compliant with regulations, custom radios require you to be careful in how you design your system. Some bands have very specific limitations as far as transmission duration, periodic data transmissions, etc. These limitations make these bands less appealing in some applications.

It's important to define the product and understand it well before selecting a band to make sure you can work within the limitations imposed by the regulations.

Multi Band and Multi-Radio IoT

The large number of protocols and the similarity among them have pushed chipset vendors to offer devices that offer multiple IoT connectivity protocols in the same packages For example, Silicon Labs offers devices supporting BLE, Zigbee, Thread and Sub-1GHz in the same device. Nordic Semiconductor offers BLE, Zigbee and Thread as well. There are design decisions required due to the time slotting to allow multiple protocols, but it is possible to bridge and take advantage of multiple technologies at the same time.

Using these radios can be very advantageous in cases where you need to connect to a legacy network or protocols while bringing them to a more modern system. Multi Band Radio options are great, but they do come with extra costs. The chip is more expensive because of the size needed to support multiple bands. Typically, there are multiple outputs on the chip for each band because the RF circuit that's optimal for each band is separate.

Multi-radio chipsets are less expensive, since they mostly rely on different software stacks, but chipset vendors do charge more given the effort needed to support the different protocols.

Z-Wave

Z-Wave Logo

Z-Wave is a proprietary networking protocol developed by various companies and controlled by Sigma Designs. Recently Silicon Labs acquired the company and the protocol. It has gained a large foothold in consumer and some IoT products because of it’s low power and integration, with particular focus in home management and security systems.

Z-Wave Technology Summary
Bluetooth LE
Range Up to 100m, 10m typical
Frequency Bands ISM 868MHz/915MHz (region dependent)
Throughput 40kbps, 100kbps (Z-Wave Plus)
Power Consumption Low - microampere average
Networking Mesh with up to 232 devices in a network
Smartphone Connectivity Z-Wave Gateway Required
Market Penetration Over 2400 interoperable certified products. 100m Z-Wave products worldwide

Using Z-Wave would require integrating into an ecosystem that supports it and in general we have not developed with it enough to recommend it. IoT developers tend to look for connectivity that is open and not proprietary, which means Z-Wave does have an advantage of being a full solution with security and other things already developed as well as quite low power.

Over the last two years, Z-Wave has evolved in several aspects including functionality and security. In fact, the Z-Wave protocol had weak and broken security features allowing hackers to control it, but updates to the specification improved the security features.

Z-Wave has an advantage in integration of devices. Many security and home devices use Z-Wave already and allow you to control all your home from a single application. This makes it much easier for users to control everything instead of having multiple apps. Over 700 manufacturers support Z-Wave, so for certain IoT products, the network effect of integrating to this ecosystem can be a significant boost.

More Information:

Wi-Fi

Z-Wave Logo

Wi-Fi, or 802.11 WLAN, is the familiar technology we use to connect computers to the internet. It connects everything from washing machines to sensors. As a IoT wireless solution, Wi-Fi allows a device to connect directly to a wireless router and to the internet without much intermediaries, which means you don’t need to include an actual gateway since most locations already have one.

Wi-Fi is also fast, with embedded Wi-Fi solution optimized for Internet of Things enabling 10-20Mbps, plenty to send almost any amount sensor data and even video. The downside of this is that Wi-Fi is relatively power hungry. Wi-Fi transmission current consumption is usually around ~120mA, with some short peaks appearing at 800mA. Because of this, Wi-Fi uses larger batteries compared to BLE and can’t use low cost.

It is capable of operating at a relatively low power. Wi-Fi SoCs can go to sleep, often maintaining their connection to the Access Point (AP) or wireless router, while consuming tens or hundreds of uA. Remaining connected to the AP is important because it allows the device to wakeup periodically and receive data from the internet like a server. If Wi-Fi is forced to disconnect, it will not be able to receive data without re-connecting, which takes time and can dd to power consumption (except in cases of long sleep).

Embedded Wi-Fi devices have improved in recent years to better meet IoT use cases, with lower power, smaller components and. Wi-Fi is more expensive that Bluetooth Low Energy or the slower device solutions. Modules are common in Wi-Fi systems because the cost of Wi-Fi and other certifications are significant, so much so that Apple, which sells millions of iPhones itself, used Murata Wi-Fi modules in its products.

As far as vendors of Wi-Fi devices, a few companies stand out. The top tier vendors have been for a long time Qualcomm, Broadcom, MediaTek, Texas Instruments, and Marvell with smaller companies such as Gainspan and Redpine Signals filling some niche applications. In recent years, Cypress Semiconductor acquired Broadcom’s IoT Wireless business unit, and Cypress now has a leading position in Wi-Fi IoT connectivity solutions.

It’s important that we distinguish between two levels of Wi-Fi solutions: Embedded and Processor.

Processor Wi-Fi solutions are higher-end, higher throughput Wi-Fi solutions that require a high speed Linux/Windows processor and TCP/IP stack to run. These will usually support 802.11ac, 802.11ax and other advanced protocols and typically support multiple antennas. You will find these devices in Smartphones, Computers, etc. Throughput of tens of Mbps to hundreds of Mbps are the goal. This type of solutions can be commonly seen in IoT Gateways, where multiple radios with Wi-Fi or Cellular backhaul are used.

Embedded Wi-Fi takes Wi-Fi and optimizes it for running at lower power, sacrificing throughput for power savings. Typically they will be limited to 802.11n, where 65Mbps is more than enough to support the data transfer, and because MIMO requires multiple antennas which increase the cost and complexity of the device. TI’s CC3220 is a good example of a fully integrated embedded Wi-Fi device which includes, as are many others.

Wi-Fi has a high speed data transfer, up to 10Mbps compared to 2Mbps and below for all other short range wireless. This gives Wi-Fi an advantage because it means the radio is on for a shorter period of time. This reduces interference and power consumption, just as talking fast means less interference with other people talking. On the other hand, it sacrifices range, because higher data rates can’t be sustained over long distances (as many of us have discovered).

Despite the number of years 802.11 Wi-Fi radios have been available, the compatibility and interoperability of Wi-Fi can vary, and it’s highly recommended to go with a well-known manufacturer. We’ve sometimes found issues with Wi-Fi radios refusing to work well with commonly Access Points. Because there are optional parts to the 802.11 and Wi-Fi specification, and because each manufacturer can implement these slightly differently, issues can happen.

One question IoT developers ask about Wi-Fi is 5GHz support. The 5GHz band has an advantage in that it has much less interference than 2.4GHz. And the higher frequency also means smaller antennas and more bandwidth for higher throughput, but it also means lower range. We tend to avoid 5GHz in many IoT devices for a few reasons:

  • Either a higher bandwidth antenna is needed that supports both bands, or separate antennas are used, increasing cost
  • 5GHz requires more certifications than 2.4GHz alone. 5GHz Wi-Fi operates in the same band as Radar, for example Radar used in airports, which can interfere. Wi-Fi chipsets then have to be able to avoid interfering. This functionality has to be certified.
  • More configuration and development work
  • 5GHz is shorter range and intended more for video streaming and high throughput, which IoT doesn't require
  • 5GHz chipsets are typically more expensive and cost more to design, test and validate

Using 5GHz is best left for applications where router is not too far and power isn’t an issue. It does help avoid interference with Bluetooth (which is why smartphones preferably should use 5GHz Access Points if they are also going to stream Bluetooth audio). Most embedded Wi-Fi radios for IoT connectivity don’t support 5GHz for the reasons above.

Wi-Fi is a star network topology, especially because 802.11ah has failed to gain traction. It is typically the system backhaul because it provides easy internet access.

Wi-Fi can be used for the edge IoT nodes as well, if properly designed and optimized, but this can be problematic when having to support a large number of devices.

As for Wi-Fi Options, there are quite a few. Here are some devices you should consider:

  • Texas Instruments - CC322x Series of Embedded Wi-Fi Microcontrollers
  • Qualcomm - QCA400x (QCA4002 and QCA4004) Wi-Fi Network Processor, QCA4020 and QCA4024 Wi-Fi Embedded
  • Cypress Semi - CYW43364 Wi-Fi Network Processor, CYW43907 Wi-Fi MCU
  • GainSpan (Telit)
  • RedPine Signals - RS14100 Wireless MCU SoCs
  • MediaTek - MT7681

Several manufacturers offer Combo Bluetooth + Wi-Fi options. Bluetooth is extremely suited for helping with provisioning the Wi-Fi device to the Access Point with little user

Low-power wide-area network (LPWAN) for Long Range IoT Connectivity

We've talked about short and medium range connectivity, but very often a sensor system itself needs either back haul connectivity or is perhaps spread over very long distances. This is where long range wireless can help IoT bridge the gap and distances. There are several technologies available for IoT long range, and we'll cover the most important ones.

LoRa

LoRa is a relatively recent addition as an Internet of Things connectivity option. LoRa is a different kind of wireless physical protocol, one which sacrifices speed for extreme range by spreading the data being sent. In practical terms LoRa can achieve 100km of range in real world usage. There's even reports that 700km has been achieved, but this is an extreme that benefited from line of sight and antennas placed high above ground. in real world use you can expect 100km or less, usually a few dozen kms.

LoRa is a proprietary protocol promoted by Semtech, a US based company which acquired the IP through several acquisitions. LoRa operates in the 915MHz and 868MHz bands.

LoRa sacrifices speed for range significantly. IT spreads the data, making a transmission take up to a second or more, compared to milliseconds for other protocols. This is somewhat of a problem - a LoRa radio remains on for quite a long time which can interfere with other devices. Because of this, the amount of transmissions (and messages) for LoRa has to be limited. But for the applications LoRa is made for, which are slow sensing and control, it works well. Designers have control over the spreading factor, so you can control how fast or slow the transmission should occur.

Lora is suitable for very low bandwidth, non-real time applications where there is tolerance for missed packets. Real time applications require a different protocol and architecture.

As a protocol LoRa specifies only the lower physical and link layers. LoRa IoT connectivity device integrators will have to build other layers on top of LoRa. One popular option designed specifically for wide is LoRaWAN, which allows companies to create a cellular tower capable of accepting LoRa devices from many

The primary LoRa vendor, Semtech, currently offers a few options for connectivity including the SX1276 and SX1272. The SX1272 is the latest generation LoRA radio that improves power consumption. Both of these are Radio only options and require an external microcontroller to and run the LoRa and LoRaWAN stack.

Because of how slow LoRa transmits, a radio must listen for a significant amount of time. Doing so for a large number of channels with a single receiver is difficult. Semtech offers LoRa gateway chipsets such as the SX1301 and SX1308 which include multiple parallel correlators that allow you to receive data from tens of thousands of devices. These devices have been used to build gateways that are installed in cellular towers to provide wide area LoRa support. You can buy a LoRa device and provision it to LoRaWAN compatible network without having to create your own gateway. This capability is unique to Cellular and LoRa deployments.

More Information:

Cellular

Cellular IoT Connectivity has existed for many years, typically under the Machine to Machine (M2M) moniker. The greatest advantage of Cellular connectivity in IoT applications is mobility and the coverage of cellular towers which enable communications without a gateway.

Cellular connectivity has been typically relegated to mostly high-end devices, primarily because of the high cost of modules. As opposed to most other connectivity options, most Cellular connectivity added to IoT devices is done via modules because getting a discrete design certified costs hundreds of thousands of dollars and significant time investment which few companies could afford. Most IoT integrators rely on modules from existing companies which can be integrated. But these modules could cost anywhere from $30 to $90.

The other downside of Cellular connectivity is the subscription cost. Just as a a smartphone requires a monthly subscription, so do devices. While providers offer plans covering multiple devices to OEMs, this is still more expensive than using a gateway using BLE, WI-Fi or other subscription-free protocols that can connect over Internet.

One advantage of Cellular that we’ve seen during deployments is the fact that compared to Wi-Fi, gateways don’t need to depend on the Wi-Fi of the site in which you’re deploying. We’ve seen companies run into issues impacting enterprise networks when deploying their system, sometimes leading to costly network issues and outages. Cellular sidesteps this issue completely and gives you complete control. There's often no need to wait for the user to provision the system.

Cellular IoT can be an advantage when you have a gateway supporting many low throughput devices, so that the subscription and hardware cost is amortized among many devices. Another good advantage is being able to deploy in places where there’s no internet.

A big disadvantage of Cellular IoT connectivity is the amount of power needed. These devices typically run off of mains powered power supplies or large lithium ion batteries (the same as used in cell phones).

As an example, Cellular devices can reach peaks of 2A for microseconds, with peaks in the 100-800mA very common. This requires a large battery and significant optimization to work. Because of this , Cellular is most common in gateways that are AC powered or in sensors that themselves are AC powered. Compare this with average current of 30-50uA achievable with Bluetooth Low Energy.

Cellular connectivity is split among several options. The highest end solutions include Cat 3, which is a high end cellular radio that you may find in Mobile phones and other applications. Their cost usually prevents them. The lowest standard category of radios is Category 1 or Cat 1 radios that provide a nominal 10Mbps download. This is good for some IoT applications, but may be overkill for some applications.

The issues we discussed with IoT cellular solutions such as cost and complexity haven’t been ignored by the cell providers. The cost and difficulty in deploying Internet of Things Cellular systems has led companies to band together and create two new standards for cellular connectivity, Cat M1 and NB-IoT. Both of these allow for lower throughput which is more than sufficient for many IoT applications, while reducing costs in development and BoM.

Cellular IoT Summary and Comparison
Cat 1 Cat M1 Cat NB-IoT
Frequency Bands Worldwide US Worldwide
Throughput 10Mbps/5Mbps (Up/Down) 1Mbps, 375Mbps 1Mbps, 375Mbps
Power Consumption High - 2A peaks Medium - 2A peaks Medium - 2A peaks
Cost Medium - 60% of Cat 3 Low 50% Cat 3 Low 40% Cat 3
End Use Gateways, Vehicle, Digital Signs Low Gateways, Sensors, Machine Control, Utilities, Asset Tracking Sensors, Machine Control, Utilities, Asset Tracking

LTE Cat M1 for IoT Connectivity

LTE Cat M1 is a new standard already deployed that takes the current LTE Cat M radios and simplifies them. By reducing the throughput (which small Internet of things devices don’t need), the radio and hardware costs go down significantly.

Cat M1 was released by Verizon in the US with the support of several companies making modules and hardware. As one of the first IoT low power cellular options, it has an advantage. Part of this is because Cat M is compatible with the current LTE network, so it made it easy for carriers to add support.

NB-IoT for IoT Connectivity

NB-IoT is another IoT wireless connectivity option that has gained traction primarily outside the US and is about to be added in the US. It is different than Cat M1 in that it does not use LTE radios and operates in a different band. Because of this, NB-IoT can be even lower cost than Cat M1 radios. One limitation of NB-IoT is the lack of mobility. NB-IoT is intended primarily for fixed device. Both LTE-Cat M1 and NB-IoT offer solutions that are much cheaper than traditional cellular, with the intended cost of chipsets in volumes at around $5 and $15 modules. This makes them a good candidate for sensors and systems.

The other good news about these two technologies is the fact that the certification path is much easier. Cellular providers can’t support tens of thousands of companies trying to certify, so they’ve simplified the process significantly if using an existing certified modules. This helps lower the development cost and efforts. Cell providers are incentivized to enable IoT wireless connectivity and they’ve been working hard to bring these technologies to the real world.

Choosing an IoT Connectivity Option

Some decisions are going to be straightforward. Power consumption is a big limitation for battery powered devices. So let’s look at options in this sense.

IoT Options for Coin Cell Batteries

Because they’re small and low cost, coin cell batteries are popular in IoT products. But their limited capacity and output power limit the connectivity options: BLE, Zigbee, Thread, Proprietary Radios and LoRa as the only options for these batteries. Some IoT devices can run for months or years on a single battery if designed and optimized properly.

Coin Cell batteries are limited in their output current, especially their peaks. A wireless radio consumes the most power when it is transmitting, and this causes peaks in power consumption. For many of the radios mentioned above, these peaks are between 5mA to 15mA. But, this depends greatly on the output power that's required. The higher the output power, the higher the peaks.

Current peaks above the rated current cause a non-linear decrease in the actual battery capacity. So the system lifetime is impacted beyond just the straight linear extrapolation of current draw and time. These peaks can cause an additional 10-30% decrease in capacity.

IoT Options for Lithium or Alkaline Batteries

Alkaline and Lithium batteries have capacities of 60mA all the way to 20Ah. The advantage of these batteries are the higher current they can provide which are required for cellular and Wi-Fi radios. If the battery chemistry allows for Lithium or Alkaline, then the previous options are still possible. The time the system runs depends on many factor, but will typically require either battery replacement or charging after a while. These options will also allow BLE, Zigbee, Thread, etc to run for a very long time, often close to the battery lifetime.

Throughput

Some IoT device are small sensors which send a few bytes per minute. Others stream a significant amount of data. The right wireless IoT technology depends on the end use case.

One of the challenges in designing an IoT system is that the throughput depends significantly on range. Some solutions such as Wi-Fi, Cat 1, Cat M1 provide significant range and throughput, but consume a lot of power as well.

IoT Connectivity Options Throughput Summary
Technology Throughput
BLE 250kbps, 1Mbps, 2Mbps
Wi-Fi 10Mbps+
Proprietary Radios Up to 2Mbps
Z-Wave 40kbps, 100kbps
Zigbee 250kbps
LoRa 250bit/s to 11kbit/s
Cat 1 10Mbps
Cat M1 1Mbps, 375kbps
NB-IoT 250 kbit/s (Down), 250kbps/20kbps (Up)

Latency

Latency is another significant factor, and it also depends on where the data is needed. For example, users looking at a temperature backend can tolerate significant delay in the data received which can be buffered and therefore optimized better for low power consumption. Applications where machine has to make critical split-second decisions require must smaller latencies and better reliability. Some technologies such as BLE have controllable latency down to about 7.25ms. Wi-Fi can transfer more data, but have more latency due to the stack and data.

Open vs Closed IoT protocols and ecosystems

Some IoT products and developers opt for open protocols whose specifications are freely available, while others don’t have that limitation. In some cases, the line isn't clear on what's open or closed since some organization provide the specs freely but charge for certification.

What’s the right choice for you depends. Some proprietary protocols provide a more turn key solution for some use cases, which can speed up development. As an example, Z-Wave for Security devices for the home. Sometimes closes protocols result in few vendors providing alternatives or a second source, which can be important. One example of this is LoRa, which is owned by Semtech (Semtech has both control of the branding and the IP behind the technology). Recently other companies such as microchip have begun creating their own solution licensed from Semtech, in part because companies will refuse to adopt a technology with only one vendor due to the risk. So multi-sourcing.

Most organizations supporting protocols provide a certification process, which allows improved compatibility but also adds to the development cost. For example, Bluetooth certification helps ensure Bluetooth devices from multiple vendors can talk without significant issues. Thread and others do the same.

Proprietary radios, since they're developed by each organization, typically result in closed ecosystems. Despite some advantages in this approach, open ecosystems allow for networks of devices from multiple vendors that can take advantage of networking effects and create greater value than a single vendor can achieve on their own. For example, Nest created Thread, but Nest benefits from other vendors certifying their Thread compatible products so that the Nest thermostat can use data from other non-Nest sensors, increasing the value of the ecosystem.

Cost

Let’s face it, cost is a significant factor in every IoT device. It’s simply impossible to deploy IoT systems without additional cost, and adding wireless connectivity to an IoT device increases the cost further. This cost comes not just from the chips, but several factors:

  • Wireless Technologies that are faster and higher power typically cost more. Complex radios such as cellular require physically larger dies (the silicon chip inside the package), which increases cost to manufacture. That’s why BLE which is simple is cheaper than Wi-Fi, and why NB-IoT is cheaper than the older LTE Cat 1 and higher radios which are 40-60% more complex
  • Certification costs including the radio certification costs to comply with regulatory agencies (FCC, Industry Canada, European CE, etc.). The more complex the radio, the more bands supported, the more expensive. For example, certifying Wi-Fi at 2.4GHz and 5GHz is more expensive than just 2.4GHz Wi-Fi. Zigbee is also 2.4GHz but much simpler because less modulation schemes means less testing needed. Modules help alleviate this cost, and make sense for low quantity designs
  • Organizations charge certification fees - For Example, Wi-Fi certification is quite expensive compared to Bluetooth SIG certification. Modules may help with this. Bluetooth is the exception – the Bluetooth SIG changed the rules a few years back because certified modules which didn’t require paying them fees were preventing them from funding their operations. Now every Bluetooth design requires paying. Proprietary radios benefit from the fact that you control everything and they require no certification from an organization (except for FCC, IC, CE, etc).
  • Higher throughput and power radios require batteries and more expensive power solution. While a CR2032 battery used in a small sensor can cost 30 cents, a Lithium polymer battery for Cellular may cost $5.
  • More complex protocols require more complex and expensive equipment for debugging. Sometimes vendors provide low cost debug equipment, but this often is insufficient. Well defined and standardized protocols tend to be easier to debug than others, although very simple proprietary protocols can be even easier. Protocols such as Bluetooth have equipment that can allow debug and verification without custom work. Other more obscure protocols do not have such a thing.
  • The cost of development should not be underestimated. Unless your IoT product ships hundreds of thousands or millions of units, the development cost will amortize less over each unit. Using a slightly more expensive protocol that is well understood can save time and cost.
  • The deployment and support cost should also be taken into consideration. A slightly larger and more expensive battery can allow a device to run longer and avoid battery replacement, which is a manual and expensive process. Taking into account the total unit cost over the deployment lifetime is critical in IoT applications.

Range

If there’s one thing that customers always bring to our attention in their designs it’s range. Wireless depends on the physical propagation of waves, and elements such as metal and objects impact it. Dealing with this starts with the selection of the right technology, but there are many decisions after it: for example, do you add an external amplifier to increase range, resulting in higher power consumption and cost? Or do you leverage mesh networking?

It may not matter too much for an IoT product that has to be in the proximity of a user’s smartphone, but for a sensor deployed in a factory monitoring temperature which isn’t easily accessible, it matters a lot.

Testing and Certification

Another aspect of picking among all the Internet of Things connectivity options is testing. Every wireless design is unique and has to be tested. We’ve heard stories of products that got barely any testing and worked. These are the exception, not the norm. Maybe the range isn’t that critical, or you’re pumping so much power (and wasting power) that you’re compensating.

The reality is that wireless RF circuits and antennas are delicate designs that need attention to detail, testing and optimization. Our experience bear this as we've had to help many companies re-design wireless devices that ended up under-performing in real life.. Sometimes the wireless limitations are not obvious except when the system has certain limits.

Testing of IoT wireless connectivity should include several aspects
  • Regulatory Certification Testing
  • Radio Transmitter Testing – Sensitivity, Output Power, Frequency
  • Protocol Testing – For example, compliance with Bluetooth specifications
  • Interference Testing – Sometimes changes need to be made to better deal with real-life interference. Some protocols are better at this than others
  • Range Testing - full testing of the system end to end helps make sure there are no integration issues
  • Temperature and Environmental Testing – although manufacturers state their radios usually operate at -40 to 85 degrees, the reality is much more complex. The more challenging the use case, the more vibration. Did you know LoRa radios, because they’re so slow in transmitting, can drift if they’re vibration, or because of the heating of the LoRa chip itself? Yes, that actually happens.

Well known radios supporting known protocols can make testing much easier. In fact, it’s easier to compare your system’s performance to reference designs or specifications, making sure your design is doing things right. For example, sensitivity testing on BLE, which is well defined, can tell us whether the radio is getting interference from other radios or other sources. Establishing your own specifications may be more difficult.

From the previous points you may think that we tend to recommend only established protocols such as Bluetooth, Wi-Fi, Thread, Zigbee, etc. Although that’s true, our background in custom wireless protocols before all of these existed means we’ve spent a lot of time developing custom solution. The truth is, there is usually little reason to re-invent the wheel, especially when engineers have worked through the issues to build a solution. But there are times when proprietary radios for IoT connectivity make a lot of sense. This is true for applications where a combination of long range, throughput and other features are needed and nothing fits exactly.

Conclusion

Choosing the right IoT wireless connectivity solution depends on a variety of factors and product requirements. We've gone over the most popular and viable technologies available today that we've designed with at Argenox. Each technology has its advantages and disadvantages that need to be considered to design the right product.