The Guide to Low-Power Wide Area Networks
Michael Alba posted on April 30, 2018 |
LoRaWAN, LTE-M, Sigfox, NB-IoT… which LPWAN should you choose?

Written by Michael Alba and Juliver Ramirez

The wide range of LPWAN options. (Logo images courtesy of 3GPP, Ingenu, LinkLabs, LoRaWAN, Nwave, Sigfox andWeightless.)

The wide range of LPWAN options. (Logo images courtesy of 3GPP, Ingenu, LinkLabs, LoRaWAN, Nwave, Sigfox and Weightless.)

Low-power wide area networks (LPWANs) are becoming an increasingly prevalent technology for the Internet of Things (IoT). As the name suggests, LPWANs are wireless communication protocols that strive to optimize two important factors for the IoT:

  • Low power: IoT sensors and devices must regularly transmit data, but they may not always be easily accessible. Therefore, it is crucial that the battery life of these devices is prolonged as much as possible.
  • Wide area coverage: If IoT devices are to be of practical use, they must maintain a network connection from anywhere they’re needed. This includes potentially remote industrial or agricultural sites.

LPWANs are distinct from wireless technology such as Zigbee, Bluetooth and other personal area networks (PANs). Though they can be used for IoT applications, PANs are limited in their range and applications.

Not surprisingly, there are several competing LPWAN protocols, meaning IoT engineers face many choices for their IoT applications. Broadly speaking, LPWANs can be grouped into two categories: cellular, meaning they employ licensed cellular frequencies, and non-cellular, meaning they make use of the unlicensed industrial, scientific and medical (ISM) radio bands.

To help understand the differences between these competing LPWANs, we spoke with industry and academic representatives about the different options available, the use cases for each, and how best to decide which LPWAN is right for your particular IoT application.

Plenty of Protocols

Before we dive into the differences between LPWANs, here’s a look at some of the main LPWAN options available:




  • LTE-M (aka Cat-M / Cat-M1 / LTE Cat-M1 / eMTC)
  • NB-IoT (aka Cat-M2)
  • EC-GSM-IoT
  • LoRa
    • LoRaWAN
    • Symphony Link
  • Sigfox
  • Ingenu (RPMA)
  • Weightless
    • Weightless-P
    • Nwave

“LPWAN refers to anything that allows you to communicate at large distances using low-power devices,” said Akshay Gadre, head of the LPWAN research project at Carnegie Mellon University (CMU). “These are all options that allow you to do that particular thing. LPWANs are not defined by these particular protocols; these protocols are defined by the LPWAN requirements.”

And what are those LPWAN requirements?There doesn’t seem to be a formal specification for the technology, but most LPWANs conform to the following guidelines:

  • Low data rate (100bps to 1Mbps)
  • Long range (up to and exceeding 10km)
  • Low power (up to and exceeding 10 years of battery life)

Not all LPWAN protocols are on equal footing, and some have more momentum than others. On the cellular side, one of the most popular options is LTE-M, an LPWAN that works over existing cellular infrastructure. On the non-cellular side, LoRaWAN is the leading contender, operating in the 915MHz ISM band.

Cellular LPWANs

LTE-M (aka Cat-M / Cat-M1 / LTE Cat-M1/ eMTC)

The 3rd Generation Partnership Project (3GPP) cellular standards body has published three different cellular LPWANs, of which LTE-M is the biggest contender. LTE-M is short for LTE-MTC, or Machine Type Communication, and it’s a release of the cellular LTE standard with a specific focus on machine-to-machine communication.

An advantage of LTE-M is that it’s completely compatible with existing cellular networks. Telecom providers don’t need to install any new hardware to use it—they just need to upload new software into their existing base stations. Because of this, telecoms like Verizon and AT&T have already begun rolling out LTE-M services. In Canada, telecom Bell has implemented a pilot project with an Ontario vineyard that utilizes Bell’s LTE-M network to monitor environmental conditions.

According to Chris Moorhead, VP of IoT NORAM at Gemalto, LTE-M is a cellular solution tailored to IoT and Industrial IoT (IIoT) applications. Compared to standard LTE, it offers better sleep, better power budgets, and other feature sets to extend a device’s coverage.

“Right now, I think Cat-M [LTE-M] is the best one to use,” said Moorhead. “A year or two, a year and a half, Narrowband IoT might come into play.”

NB-IoT (a.k.a. Cat-M2)

Narrowband IoT (NB-IoT) is the second LPWAN standard put forth by 3GPP, but it differs from LTE-M in several respects. For one, it has a much lower throughput—250kbps compared to 1Mbps. Another difference is that NB-IoT is based on a direct-sequence spread spectrum (DSSS) modulation scheme, so it’s not related to LTE in the same way as LTE-M. Additionally, it does not necessarily operate within LTE bands. NB-IoT is designed to be used either within LTE guard bands or independently in unused Global System for Mobile (GSM) bands with a 180kHz channel bandwidth.

While some North American telecoms, like T-Mobile, have begun investing in NB-IoT technology this particular LPWAN has been more prominent in Europe. According to Moorhead, there are a number of reasons that NB-IoT has lagged behind LTE-M—including a lack of hardware support.

"There are still only a few modules available,” Moorhead said. “For Narrowband to be actualized on the benefits and the business case, you really need that cost-optimized device. And to get that, you need a native Narrowband chipset, and we haven’t seen many come into the market yet. I think the majority of them will be coming out towards the end of this year and into next year. So, having a true Narrowband chipset is still a limiting factor.”

Despite its lag behind LTE-M, there are some who believe that NB-IoT will fill a particular IoT niche. Asset tracking company BeWhere, which produces Bluetooth Low Energy (BLE) beacons, is working to introduce cellular LPWAN alternatives. Alban Hoxha, CTO of BeWhere, predicts an industrial use case for NB-IoT.

“It’s a more reliable protocol,” Hoxha said. “Exception monitoring will be the holy grail of NB-IoT. It will replace Bluetooth (BLE) to a certain extent, but there will be another area where they will complement each other. Today, the way BLE works is they constantly advertise. The backhaul is a problem, so you always need an edge gateway to move the data to the cloud.” (NB-IoT, in contrast, does not require a gateway. With this standard, data is sent directly to the main server.)

NB-IoT also has a lower power requirement than LTE-M, though this a double-edged sword. While less power is always good when it comes to the IoT, it also means that NB-IoT has worse penetration than LTE-M.


Extended Coverage Global System for Mobile IoT (EC-GSM-IoT) is the third 3GPP LPWAN standard operating within the licensed spectrum. Unlike LTE-M, which operates on the LTE bands, EC-GSM operates on General Packet Radio Service (GPRS) spectrum. To use EC-GSM-IoT on GSM networks, this protocol simply requires a software upgrade of GSM networks, since most mobile hardware companies already support it. Of the three cellular options, EC-GSM-IoT seems to have the least momentum.

Non-Cellular LPWANs


The unlicensed ISM bands played host to the earliest LPWAN, Sigfox, according to Sigfox’s Ajay Rayne. “We can take credit for pretty much establishing the category of LPWANs,” he said. “I would say we practically invented the term.”

Whether the company coined the term or not, Sigfox has been in the LPWAN arena for nearly a decade, and is now one of the leading contenders of the non-cellular options. The company was founded in France in 2009, and now has networks in 47 countries and counting. Its goal is to operate in 60 countries by the end of 2018, and to eventually create a global Sigfox network—a local network around the world, in Rayne’s words.

It’s worth discussing the business model of Sigfox, as it’s a bit different than the other models we’ll discuss shortly. Sigfox’s business is the network—users only pay to send messages. The actual cost depends on the number of messages that are sent. One or two messages a day will cost roughly $1/year. One message every 10 minutes—about 140 messages a day—will cost roughly $1/month.

“Sigfox’s model is very simple,” Rayne explained. “Sigfox only makes money on the subscription. We do not charge any royalties to any of our chip vendors. In fact, we provide our IP at no charge and royalty free. We have several reference designs that are available, including some that are available royalty free. As a designer, you only have to focus on the design side. You don’t need to worry about the intermediate layers of the network—do I need a switch? Do I need a router? Do I need a gateway? Do I need a back-end authentication server? None of that. Because we take care of the network side. All the engineers have to do is focus on their core competency and build the best device possible.”

As simple as that may sound, Sigfox does have one potential drawback: message size. Each message can carry a payload up to 12 bytes, which may not meet the requirements of some users. Rayne openly acknowledges this fact, as he considers Sigfox to be complementary to higher throughput LPWANs like NB-IoT and LTE-M.

“Sigfox is secure, small amounts of data,” Rayne said. “That’s the market we play in, and we believe the majority of IoT devices are going to be in that range of 12 bytes of data. But if it’s anything beyond 12 bytes, or if you want streaming video or audio, we can’t play—and that’s the limitation.”


Of all the non-cellular LPWAN options, LoRaWAN is perhaps the biggest contender. It’s backed by the LoRa Alliance, a consortium of over 500 companies including Cisco, IBM, Alibaba, SK Telecom and more. The support of so many IoT players has given LoRaWAN a comfortable seat at the LPWAN table.

“On the unlicensed side, LoRa is gaining a lot of acceptance as a standard," said Sanjay Khatri, head of IoT Platform Product Marketing at Cisco IoT.

LoRa…LoRaWAN…what’s the difference? LoRa refers to a physical layer (PHY) owned by Semtech Corporation, while LoRaWAN refers to the Media Access Control (MAC) layer specification, which is maintained by the LoRa Alliance. While the LoRaWAN specification is openly available, LoRa is proprietary, and Semtech receives royalties from chip vendors that sell LoRa modules.

“The air interface is owned by Semtech, but there's a whole stack of networking that goes on top of that,” said Khatri.

Unlike Sigfox, engineers using LoRaWAN will need to consider networking components like gateways, access points, servers, and switches. However, LoRa Alliance member companies like Cisco offer gateways and other prebuilt LoRaWAN solutions.

The Cisco Wireless Gateway for LoRaWAN. (Image courtesy of Cisco.)
The Cisco Wireless Gateway for LoRaWAN. (Image courtesy of Cisco.)

With a message payload that is up to 222 bytes, LoRaWAN is suited to a broader range of applications than Sigfox, and can support more sophisticated devices like those requiring over-the-air (OTA) updates. This is one reason that Khatri believes LoRaWAN provides a more sustainable ecosystem than Sigfox.

“In LoRa, you have more headroom in terms of bandwidth and size of messages,” he said. “And the threshold isn't as high for companies to come in and license it.”

However, Khatri also believes LoRaWAN can potentially complement licensed LPWAN technology. “Our view at Cisco is it's not an either/or [licensed and unlicensed]; they're going to be complementary. The last mile you would essentially use LoRa, but then connecting that last mile back into the cloud, there's many different forms of technology,including LTE.”

For its LPWAN research projects, Carnegie Mellon University uses LoRaWAN. These projects include Choir, a system to improve LPWANs in dense urban environments, and OpenChirp, an open LPWAN management framework.

“Different protocols allow you to innovate at different layers, but LoRa is the best for university researchers,” said CMU’s Akshay Gadre. “It allows you to customize all of the layers as you want. LoRa has the complete stacks on the application layer to the physical layer open to you. The only thing that is constrained is the hardware—the silicon is protected by Semtech.”

LoRaWAN isn’t the only LPWAN to use the LoRa PHY layer. Link Labs has developed a competing LoRa-based LPWAN called Symphony Link, which claims to improve upon LoRaWAN with features like guaranteed message receipt,greater capacity, the use of repeaters, and easier firmware upgrades. However, due to the backing of the LoRa Alliance, Symphony Link never stood much of a chance against LoRaWAN, and Link Labs recently shifted its strategy to focus on IoT application development.

Ingenu (RPMA)

Unlike LoRa and Sigfox, which both utilize the 915MHz ISM band, Ingenu utilizes the 2.4GHz ISM band—the same band used by Wi-Fi and Bluetooth. The advantage of this band is that it’s globally available, meaning developers don’t have to consider what regions their products will be used in. In contrast, LoRa and Sigfox use different bands based on location, whether in North America (915MHz) or Europe (869MHz). The 2.4GHz band also has much more bandwidth available than the 915MHz band.

Graphic detailing different spectral allotments and requirements around the globe. (Image courtesy of Ingenu.)
Graphic detailing different spectral allotments and requirements around the globe. (Image courtesy of Ingenu.)

At the heart of the Ingenu LPWAN is a system called Random Phase Multiple Access (RPMA), which is a PHY and MAC layer developed by Ingenu to meet what it considers to be the specific needs of LPWANs: a global band (2.4GHz), wide coverage (a single RPMA access point can cover up to 176 square miles), tremendous capacity (a single RPMA access point can receive 535,117 messages per hour), long battery life (10-20+ years), and robustness to interference. RPMA also addresses what Ingenu calls the LPWA Device Bill of Rights: two-way data, delivery acknowledgment, flexible packet sizes, network responsiveness, authentication and broadcast capability.

Besides authoring the LPWAN standard, Ingenu operates the Machine Network, a publicly available RPMA network that provides over 100,000 square miles of coverage in more than 30 cities in the U.S., as well as coverage in over 30 countries globally.

Ingenu makes strong claims about the superiority of RPMA compared to LoRa, Sigfox and cellular LPWANs, but according to Cisco’s Sanjay Khatri, Ingenu has tailed off a bit in recent years. Just last month, Ingenu announced a shift in its corporate strategy to deliver RPMA via Platform as a Service (PaaS).


The Weightless LPWAN standard began as three standards, named Weightless-N, Weightless-W and Weightless-P. It seems the Weightless Special Interest Group (SIG) has all but abandoned Weightless-N and -W, and now puts forth a single standard, Weightless-P, sometimes simply called Weightless.

One of the key characteristics of Weightless is that it’s designed to operate in a variety of bands—in fact, Weightless is defined for operation in all the unlicensed sub-GHz ISM bands, including the 163, 433, 470, 780, 868, 915 and 923MHz bands.

Weightless is also an open standard, overseen by the nonprofit Weightless SIG, which tends to be better for both innovation and competition than proprietary standards like LoRa. However, despite its name, Weightless seems to be stuck on the ground—a scarcity of available hardware and infrequent updates to the specification give Weightless the appearance of stagnation.

But Weightless is alive and well in one sense: Nwave, a smart parking technology company, uses a proprietary LPWAN for its smart parking meters that originally formed the basis for Weightless-N, the first Weightless standard.

Which LPWAN Is Right for Me?

With so many LPWAN options available, it’s not easy to know which one is best for your needs. If you’re committed to a cellular option, LTE-M is probably your best bet right now—it’s got the most support from telecoms and the most commercial hardware. While NB-IoT may find a niche in the years to come, the hardware and infrastructure still has a way to go before this standard can realize its potential.

As for the non-cellular options, Gemalto’s Chris Moorhead views the playing field much like the VHS/Betamax debate of the 70s and 80s. All of the choices provide the key characteristics of LPWANs to more or lesser degrees. Ultimately, the winner of this debate may just be the LPWAN with the most momentum. Right now, it seems that LoRaWAN—with the backing of the LoRa Alliance—is in this top spot. Sigfox, the next leading contender, is easier to implement from an engineering perspective, but it may be held back by its low message size of 12 bytes. Ingenu and Weightless, both perfectly good standards, may already have been left behind in the LPWAN race.

Ultimately,your choice of LPWAN will come down to the specifics of your IoT application.

“For us, it's really an investigation on what the customer wants to do, what the customer wants to accomplish, what regions the customer wants to deploy in,and what the critical aspects of their deployment are,” said Moorhead. “Is it connectivity everywhere? Is it power management, power budget? Dependent on the size of the data packet? That all comes to play. There's no real easy answer.”

To help you out, here’s a summary of some of the key aspects of the leading LPWANs:

Data rate
Channel Bandwidth
Message Payload
Channel Access Method
Battery Life
Type Cellular
Frequency LTE bands
Data rate 1Mbps
Channel Bandwidth 1.4MHz
Message Payload  
Channel Access Method Downlink: OFDMA, Uplink: SC-FDMA
Range Within any area of LTE coverage
Battery Life More than 10 years
Modulation 16-QAM
Type Cellular
Frequency Subset of LTE bands, standalone on GSM bands
Data rate 250kbps
Channel Bandwidth 180kHz
Message Payload 1600 bytes
Channel Access Method Downlink: OFDMA, Uplink: SC-FDMA
Range 1km (urban), 10km (rural)
Battery Life Up to 10 years
Modulation QPSK, BPSK
Type Cellular
Frequency GSM bands
Data rate 70-240kbps
Channel Bandwidth 200kHz
Message Payload  
Channel Access Method TDMA / FDMA
Range Within existing GSM coverage
Battery Life Up to 10 years
Modulation GMSK, 8PSK
Type Non-cellular
Frequency 915MHz (869MHz, Europe)
Data rate 0.3-50kpbs
Channel Bandwidth 125, 250 and 500kHz
Message Payload 51- 222 bytes
Channel Access Method ALOHA
Range 2-5km (urban), 15km (suburban)
Battery Life Up to 10 years
Modulation CSS
LPWAN Sigfox
Type Non-cellular
Frequency 915MHz (869MHz, Europe)
Data rate 100bps
Channel Bandwidth 100Hz
Message Payload 12bytes
Channel Access Method RFTDMA
Range >40km
Battery Life Up to 10 years
Modulation BPSK
Type Non-cellular
Frequency 2.4GHz
Data rate 624kbps (uplink), 156kbps (downlink)
Channel Bandwidth 1MHz
Message Payload Flexible message size
Channel Access Method RPMA
Range Up to 176 square miles
Battery Life Up to 15 years
Modulation D-BPSK
LPWAN Weightless
Type Non-cellular
Frequency 163, 433, 470, 780, 868, 915 and 923MHz
Data rate 200bps-100kbps
Channel Bandwidth 12.5kHz
Message Payload <48 bytes
Channel Access Method TDMA
Range 2km (urban)
Battery Life 3-8 years
Modulation GMSK, OQPSK


8PSK—8-ary Phase-Shift Keying

16-QAM—16 Quadrature Amplitude Modulation

ALOHA—Additive Links On-line Hawaii Area

BPSK—Binary Phase-Shift Keying

CSS—Chirp Spread Spectrum

D-BPSK—Differential Binary Phase-Shift Keying

GMSK—Gaussian Minimum Shift Keying

OFDMA—Orthogonal Frequency-Division Multiple Access

OQPSK—Offset Quadrature Phase-Shift Keying

QPSK—Quadrature Phase-Shift Keying

RFTDMA—Random Frequency and Time Division Multiple Access

RPMA—Random Phase Multiple Access

SC-FDMA—Single Carrier Frequency Division Multiple Access

TDMA—Time Division Multiple Access

Recommended For You