Close

March 2016: Navigating the IoT wireless WAN connectivity labyrinth

Executive Summary

The Internet of Things (IoT) is creating a storm as it seeks to transform industries, public services and consumer lifestyles by bringing connectivity and intelligence to everyday objects ('things'). While most IoT devices rely on local, personal and proximity based connectivity, such as ZigBee, Z-Wave and RF-ID, nearly 300 million IoT connected devices deployed today use wireless wide area network (WWAN) connectivity that is provided by a network operator, see Exhibit 1. We forecast that this will increase to 1.45 billion connections in 2020 and 4.63 billion by 2025.

Even though IoT has been hyped for many years, it has not been of significant importance to the mobile industry, with only 4.1 percent of wireless connections in 2015 being for IoT devices. Over half of all IoT devices use 2G-GSM/GPRS, even as mobile operators talk about decommissioning their networks, and LTE-Cat 1, which was standardized in 2008 for IoT use, has only been supported by two start-up semi-conductor vendors, namely Altair and Sequans. However, the sentiment towards IoT for many mobile operators has changed over the last 18-24 months, culminating in a slew of partnerships, technology trials and platform strategies, and feverish standardization efforts by the 3GPP.

Exhibit 1: Forecast for WWAN based IoT connections
Source: Tolaga Research 2016

In particular, the 3GPP has standardized software enhancements to 2G-GSM/GPRS and new variants for LTE, including LTE-Cat 0 with Release 12 and LTE-M1 and M2 (aka NB-IoT) which are anticipated with Release 13. As these releases come to market, we do not expect Cat 0 to gain meaningful adoption. Assuming M2 is standardized and comes to market in a timely manner, we believe that it will more widely adopted than M1, with M1 being used for higher bandwidth applications. The 3GPP is also standardizing 5G capabilities for IoT, which it intends to release after the 2020 time-frame. 5G standards incorporate functionality for uncoordinated low latency and busty IoT traffic, and might leverage spread-spectrum radio innovations developed for unlicensed technologies.

Until recently, mobile operators have essentially been the only game in town for WWAN IoT connectivity. However this has changed in recent years with the emergence of low power wireless access (LPWA) solutions. These technologies consist of a variety of proprietary and standards based solutions, and operate in unlicensed radio spectrum. LPWA solutions are offered by companies a variety of companies including, Ingenu, Link Labs, M2COMM, nWave, Qowisio, Semtech (LoRa), Silver Spring, SigFox and Telensa. Each player has its own strategy that is influenced by its incumbent market position, capital resources, technology and target markets.

SigFox, Telensa, nWave and Qowisio operate ultra-narrow-band (UNB) technologies. SigFox has the strongest market momentum amongst competitive UNB solutions. It recently raised USD 151 million Series D funding and has shepherded deployments with its proprietary technology across 12 international markets. Telensa has successfully deployed its proprietary technology for smart-street-lights,recently raised USD 18 million in capital, and is expanding its solution to address smart-city applications. nWave recently standardized its solution with Weightless-N, raised USD 500 thousand in seed funding, published an SDK standard and deployed in London and Copenhagen. Qowisio initially developed its UNB solution for its own products, and since building out private networks, has raised €15 million to deploy a dual mode UNB/LoRa network in France.

Semtech (LoRa), Silver Spring, Link Labs Ingenu, and M2COMM deliver higher bandwidth LPWA solutions relative to UNB, albeit with increased device complexities and in some cases reduced coverage. Semtech is a fab-less semiconductor company that netted USD 5 and 15 million in 2014 and 2015, respectively from its proprietary LoRa technology. LoRa has seen robust market adoption and benefited from the support of the LoRa Alliance and LoRaWAN standards for higher layer functionality.

Silver Spring Networks provides smart-utility and street-light solutions using an IEEE 802.15.4 mesh networking solution that has also been adopted by the WiSUN Alliance. Its revenues in 2015 were USD 489 million and in December 2015 it had 22.9 million IoT devices connected to its network. Link Labs provides a variety of LoRa based solutions, including its Symphony platform, which has LoRa operating in an IEEE 802.15.4 environment. Ingenu comes with its proprietary Random Phase Multiple Access (RPMA) technology and a strong pedigree from Qualcomm and Verizon. Ingenu has raised USD 118 million and has a lofty strategy of building a global 'Machine Network', initially with a focus toward the US market. M2COMM is a Taiwanese fabless semi-conductor vendor, which is adapting its Platanus wireless platform to develop the Weightless-P standard, as a complement to Weightless-N.

As competition heats up for LPWA players, market consolidation is inevitable to the detriment of many of the solutions that are currently being developed and deployed. The fate of LPWA will ultimately be determined by industry heavy-weights including Tier 1 mobile operators, silicon vendors, utility companies, and infrastructure vendors, and end-users in cases where break-through solutions emerge. However, in the interim it is crucial for LPWA players to maintain nimble business models that anticipate changing market dynamics. Notable changes include advancements in LTE and other technologies such as embedded SIMs, solutions that integrate multiple network technologies and innovative pricing models, such as those that enable subscription-less service plans.

As the market for WWAN IoT connectivity solutions consolidates, we do not believe that one “Intel Centrino-like” solution will emerge as victorious in supporting the needs of all IoT use-cases. Instead, several licensed and unlicensed spectrum standards will be needed, with LTE-M1 and especially M2 taking the reigns for licensed spectrum implementations. Unlicensed solutions will be used in cases where licensed spectrum alternatives are too expensive or unavailable, or to support innovative business models, such as neutral-hosts. In addition, industry players will continue to regard unlicensed spectrum technologies as less secure, even though it is the system design as opposed to radio technology that determines the system security. While SigFox and LoRa have an early market lead in supporting unlicensed spectrum technologies, with LoRa currently having greater momentum, there is still the potential for another technology to come to the fore, particularly if it gets strategic support from an industry heavy-weight.

Introduction

The Internet of Things (IoT) is creating a storm as it seeks to transform industries, public services and consumer lifestyles, by bringing connectivity and intelligence to everyday objects ('things'). While most IoT devices rely on local, personal and proximity based connectivity, (e.g ZigBee, Z-Wave and RF-ID), we estimate that 289 million devices are currently deployed with subscriptions for wireless wide area network (WWAN) services. These devices require coverage over large geographical areas and in most cases use licensed spectrum technologies with the support of mobile operators. Today over half of the WWAN connected IoT devices use 2G-GSM/GPRS. In some cases, 2G-CDMA is used where it is deployed and high bandwidth IoT devices typically use 3G-UMTS, or 4G-LTE in some isolated cases.

In recent years low power wireless access (LPWA) solutions that operate in unlicensed radio spectrum have been provided by a slew of companies, including Ingenu, Link Labs, M2COMM, nWave, Qowisio, Semtech (LoRa), Silver Spring, SigFox and Telensa. These companies come with a diverse array of proprietary and standards based technologies and business models that span both horizontal and vertical market strategies. Some of these companies have seen robust growth, with market demand for LPWA solutions being fueled by several drivers including, the need for low cost and energy efficient connectivity modules, and uncertainties in the long term availability of licensed spectrum technologies like 2G-GSM/GPRS.

Despite rhetoric to the contrary, IoT has not been particularly strategic for most mobile operators in the past and only accounts for a small percentage of wireless mobile connections. However, mobile operator sentiment towards IoT has changed, particularly in the last 18-24 months. This change has been driven by a variety of factors, including advancements in IoT solutions, customer demands with compelling use cases, and competition from LPWA alternatives. It has culminated in 3GPP standardization efforts, with a particular emphasis towards reducing the cost, complexity, and power consumption for LTE communication modules. The 3GPP has also standardized advanced features for GSM/GPRS, in an effort to increase its longevity and performance for IoT applications.

While LTE-Cat 1 has been available since 2008 and was designed primarily to support IoT, it has garnered little interest from the mobile industry. Instead it has remained a niche technology supported by start up semiconductor vendors, including Altair and Sequans. In 2015, the 3GPP published its standards for LTE-Cat 0 with 3GPP-Release 12. LTE-Cat 0 is intended to supersede Cat 1. However its future is uncertain, since the 3GPP has plans for additional technology standards with 3GPP-Release 13, namely LTE-M1 and LTE-M2 (formerly known as NB-IoT).

The recent proliferation of WWAN connectivity solutions creates a complex and confusing landscape for the IoT industry to navigate. This report investigates this landscape with an emphasis towards the following:

  • How is the mobile operator support for IoT shaping up?
  • How well do the alternative IoT technologies align with compelling use cases and business models?
  • What business models and technology solutions and what does this mean for LPWA providers?
  • Are unlicensed spectrum systems really less secure than licensed spectrum systems?
  • Where does 5G fit?

Mobile network operators accelerate IoT efforts

Mobile network operators have supported IoT under the guise of M2M for many years. However, the IoT solutions provided by mobile operators have generally been shoehorned into traditional mobile service plans, rather than being tailored to meet the specific needs of IoT. This is not surprising given the small number of IoT connections that the mobile industry is supporting. At the end of 2015, we estimate that there were 289 million IoT devices deployed globally with WWAN subscriptions, effectively making it a niche technology relative to the 7.9 billion mobile connections that industry currently supports.

However for some operators, IoT has already become strategic and accounts for a meaningful percentage of the network connections that they are supporting. Exhibit 2, compares the relative penetration of IoT connections for a sample of mobile operators. Amongst this sample AT&T, Verizon and Rogers having the highest percentage penetration of IoT devices, closely followed by Orange and Bouygues in France, Proximus in Belgium and T-Mobile in Germany.

Exhibit 2: Mobile operator IoT connections as percentage of mobile connections in 2015
Source: Tolaga Research 2016

As mobile operators increase their attention towards the IoT market, many have strategies to target vertical market solutions, such as transportation and logistics, asset tracking, connected cars and smart city applications. These strategies disrupt traditional business models with complicated ecosystems, challenging pricing structures and the need for partnerships to deliver vertical market expertise.

For example, many Tier 1 operators have already forged strategic partnerships with auto makers, and utility companies, logistics companies and municipalities. In addition, some multi-national operators are developing unified platforms across their International footprints, to drive service scale and consistency; albeit with systems integration challenges in cases where the platforms operate with different business and operational support systems (BOSS). Even with these efforts, the mobile industry continues to have a variety of challenges that it needs to address, which include the following:

  • Enabling cost structures to address the low unit revenue opportunities associated with many IoT services. This has been a perennial challenge for mobile operators and depends on highly efficient service distribution and operational models that are developed specifically for IoT. We expect that many of the operational models needed to bring suitable cost structures for IoT will be developed in emerging markets like India, which have already demonstrated their skills in profiting from mobile services with low unit revenues.
  • Simplified supply chains and scalable distribution capabilities, with less complicated radio architectures, embedded SIM capabilities (which will be studied in an upcoming Tolaga Report) and global coverage with a limited number of device SKUs. In the future, mobile operators also might opt for subscription-less service offers to eliminate the need for SIM functionality and billing relationships with IoT end users. However this approach is particularly disruptive to existing business models, since it eliminates recurring revenues that are coveted by operators and their investors.
  • Company cultures that are comfortable with subordinated ecosystem roles. While the mobile industry has made significant strides in this regard, further progress is needed. We believe that the necessary company cultures can only be achieved once company employees and their executives are incentivized to make it happen.

With the complexities associated with bringing IoT solutions to market, there are clear benefits for those players that can capitalize on their market scale. Today virtually half of the WWAN IoT device connections are provided by five mobile operators, namely AT&T, China Mobile, China Unicom, Verizon and Vodafone, see Exhibit 3. However, since the IoT market is still nascent, we believe that other operators, particularly those who have operations in emerging markets like India, Brazil and Africa, will gaining meaningful share of the IoT market over the next 24 to 36 months.

Exhibit 3: Currently six mobile operators provide virtually half of the global WWAN connections
Source: Tolaga Research 2016

Historically mobile operators have been cumbersome in their response to disruptive market forces until they are confronted with formidable competition. As a result, competition from LPWA players is an important catalyst for mobile operators to adapt to IoT demands. As they adapt, we believe that mobile operators will be the primary providers of licensed spectrum and LPWA solutions for the IoT market for the foreseeable future. As this occurs, we expect that the number of IoT devices with WWAN connectivity will increase from 289 to 1455 million between 2015 and 2020 and to 4630 million in 2025.

Licensed Spectrum Technology Plays IoT Catch Up

Until recently, there weren't many viable WWAN connectivity options for IoT, with 2G-GSM/GPRS being used primarily for narrow-band and 3G-UMTS for broadband connectivity. Today over half of the WWAN connected IoT devices use 2G and more than a third use 3G, even as mobile operators make plans to sunset their 2G and 3G networks to make way for 4G. For example, AT&T has announced that it will cease to operate its GSM/GPRS network after January 1, 2017. Verizon has announced that it will sunset both its 2G and 3G CDMA networks by 2021. All mobile operators in Singapore plan to sunset their 2G networks by April 2017. Telenor has announced plans to sunset its 3G network in the 2020 time-frame, five years before it plans to sunset its GSM.

Exhibit 4 identifies current and emerging licensed spectrum IoT connectivity solutions, and includes the connection speeds, target use cases and estimated availability. With 2G and 3G technologies being sunsetted by many operators, the 3GPP is releasing LTE variants that are targeted towards different IoT use-cases. The 3GPP has also made software advancements to 2G-GSM/GPRS to improve its performance for IoT applications, and is in the early stages of standardizing 5G-IoT solutions, which are targeted for the 2020 timeframe. As the 3GPP pursues its standardization efforts, the suitable technology positioning and migration strategies for the mobile industry are complicated by a variety of uncertainties. In particular:

  • The relative success of LTE-M1 and M2 depends on their market timing, and the IoT use cases that gain the greatest market traction. Currently we believe that market demands favor LTE-M2 relative to M1, so long as it is standardized and productized in a timely manner.
  • With growing support expected for LTE-M1 and M2, we don't expect LTE-Cat 0 to see meaningful market adoption.
  • The longevity of 2G-GSM/GPRS is complicated by several factors, including its large installed base and the standardization of EC-GSM/GPRS. Furthermore, since the IoT use cases supported by 2G align with the LTE-M2, there will be a lag in technology availability and in many cases network coverage. Accounting for these factors, we currently believe that the lion's share of GSM/GPRS networks will be decommissioned within the next five to seven years.
  • While it is still early days for 5G, a great deal of progress has already been made in researching solutions for IoT. Some proposed IoT solutions use spread spectrum technologies for bursty and low latency applications and might leverage technology know-how and intellectual property developed for LPWA. This will be discussed in more detail in an upcoming Tolaga report.

Exhibit 4: Positioning of licensed and unlicensed spectrum WWAN IoT connectivity solutions
Source: Tolaga Research 2016

In addition to bandwidth performance, IoT technology solutions must also address a variety of other factors depending on the use cases they are supporting. These include the following:

  • Device Costs, which can be a game changer for many IoT use cases with challenging economics.
  • Device Battery Life, which is particularly relevant to IoT use cases that require devices to be deployed in remote locations for protracted time periods without external power resources. For these use cases, the IoT industry has generally been striving for devices capable of being battery operated for a ten year period.
  • Network Coverage, which is particularly important for IoT use cases with high mobility requirements. Notable examples include connected cars, and goods tracking for transportation and logistics.
  • Reliability and Quality of Service (QoS), which is needed for IoT use cases that depend on real-time functionality, such as drive-less vehicles and those associated with industrial processes.
  • Security, which while important for all IoT use cases is more important for some, such as those that incorporate critical operational functions and sensitive data, or are connected to mission critical systems that might be vulnerable to horizontal attack surfaces.
  • Bi-directional connectivity, which we believe is necessary for most IoT use cases for a variety of reasons, including the need to receive acknowledgment messages for successful communications, periodic software and firmware updates and in some cases for other command and control functions sent to IoT devices.
  • Positioning,which is needed for IoT use cases where the location of the 'thing' needs to be known, such as asset tracking where mobile IoT devices must be accurately located.

Exhibit 5 assesses the relative importance of these factors for several use cases, including goods tracking for transportation and logistics, wearables for attached fitness devices and connected cars.

Based on the assessment in Exhibit 5, transportation and logistics require low cost devices, with long battery life, accurate positioning, coverage and security. Wearable fitness devices have less stringent design requirements, but must be cost effective, with accurate positioning, reliable coverage and relatively long battery life. Connected cars have stringent requirements for all the factors listed above, except battery life, since on-board devices have external power sources. In addition connected car applications are generally less cost sensitive than many of other typical IoT use cases.

Exhibit 5: IoT performance priorities vary depending on the use cases
Source: Tolaga Research 2016

An Uncertain Future for GSM/GPRS and a 3G Misfit

Even though 2G and 3G technologies are widely used today for IoT solutions, we believe that both technologies will cease to exist in many markets within the next five to seven years, and will be superseded by 4G-LTE and emerging 5G technologies for cellular services.

While some mobile operators are forging ahead with plans to retire their GSM/GPRS networks, others are looking for opportunities to extend its life specifically for IoT services. To support these efforts, the 3GPP is incorporating enhancements to GSM/GPRS under the guise of EC-EGPRS in Release 13, which is anticipated in the first quarter of 2016. These include features to:

  • Enhance Coverage, by enabling multiple transmissions (aka “Blind Physical Layer Repetitions” BPLR) to increase the probability of signal reception. Players like Ericsson who trialed EC-GSM with Orange in France in November 2015, demonstrated that BPLR can increase the GSM coverage by 20dB. However, we believe this to be overly optimistic, particularly for static devices.
  • Improved Power Efficiencies, with the objective of achieving 10 year battery life. This is achieved with a variety of techniques such as changes to idle mode behavior, enabling devices to “camp” on base stations for protracted time periods, without the need for neighbor cell scanning. It also includes extended discontinuous transmission (eDRX), which extends the time period required for devices to respond to network triggered reporting.
  • Ability to Support Massive IoT deployments, by increasing the granularity for which channel resources can be allocated, and simplifying core networks to eliminate legacy functions that are not needed for IoT operations, such as regular location updates for devices that are deployed in fixed locations (e.g. smart meters).
  • Reduced Device Complexity, by eliminating functionality that is not needed for IoT operations, and leveraging technology advancements, such as System-on- Chip (SoC) architectures.
  • Security Enhancements, that use network based security techniques to address the volatile and unsupervised environments in which IoT devices are commonly deployed.

Since EC-GSM/GPRS is implemented with software upgrades to existing networks, it is well suited to existing GSM/GPRS installations, or in cases where alternative technologies, such as LTE-M1 are not available or lack coverage.

The future of 3G for IoT services is more tenuous than GSM/GPRS. 3G IoT solutions are prohibitively expensive and will remain so as suitable LTE based solutions become readily available over the next 12-24 months. Furthermore, the longevity of 3G networks is being truncated by 4G in many markets, particularly with the maturation of Voice-Over-LTE (VoLTE).

Adapting 4G-LTE Design Priorities for IoT

The primary justification for 4G-LTE has been to squeeze greater capacity from scarce radio network resources and to enable enhanced broadband services. While 4G-LTE is well suited for broadband applications, it has a granular resource allocation scheme and with the right architecture can achieve efficient transceiver design schemes for IoT applications. When the 3GPP published its initial LTE specifications in 2008, it included the LTE-Cat 1 variant that was intended for IoT applications. LTE-Cat 1 enables maximum downlink and uplink throughput speeds of 10 and 5Mbps, respectively but has not been widely supported by the mobile industry.

In March 2015 the 3GPP published standards for LTE-Cat 0 with 3GPP-Release 12 and has plans to publish guidelines for LTE-M1 with Release 13. For both LTE-Cat 0 and M1 enable peak downlink and uplink rates of 1Mbps. Their modems are being simplified using a variety of techniques including half duplex, which eliminates the need for devices to simultaneously transmit and receive. Additional changes are made with LTE-M1 for further simplification by reducing the channel bandwidth from 20 to 1.4MHz, while still maintaining the same peak data rates as LTE-Cat 0.

In addition to LTE-M1, the 3GPP is standardizing LTE-M2 (a.k.a. NB-IoT), which is intended to achieve lower cost, improved energy consumption and throughput relative to LTE-M1. The objective of LTE-M2, has been described by 3GPP as enabling “cellular system support for ultra low complexity and low throughput Internet of Things”, with “low delay sensitivity, ultra low device cost, low device power consumption and an optimized network architecture”. In other words, it was the 3GPP's response to LPWA solutions.

When LTE-M2 was originally proposed it created a firestorm of proposals by companies including Ericsson, Huawei, Nokia and Qualcomm, who were each vying for their Intellectual Property (IP) to be included in the standards. After considerable debate, an approach was agreed upon in September 2015, with the ambitious objective to incorporate LTE-M2 in 3GPP- Release 13, which will be completed after March 2016. While the specific characteristics of LTE-M2 were still being finalized when this report was written, its key characteristics include the following:

  • Radio frequency channels with 180kHz bandwidth for both uplink and down-link connections.
  • OFDMA modulation for down-link connections with 15kHz sub-carrier spacing, which is similar to that for conventional LTE solutions. Note that 3.75kHz sub-carrier spacing was originally considered, to enable faster synchronization and lower latency connections. However, it was subsequently discarded in favor of 15kHz, to better align with conventional LTE.
  • Two options are being considered for uplink connections, namely a FDMA/GMSK scheme similar to that used for GSM, and SC-FDMA, which is used for LTE. OFDMA is not used for uplink connections because its waveform has a high Peak-to-Average-Power-Ratio (PAPR). In particular, since power amplitudes of the narrow-band sub-carriers in an OFDMA waveform are independently and identically distributed, the power distribution of the combined sub-carriers are approximately Gaussian distributed (according to the Central Limit Theorem) with a large Peak-to-Average- Power-Ratios (PAPR). To avoid signal distortion, power amplifiers and digital-to-analog (D/A) and analog-to-digital (A/D) converters used for OFDMA require large dynamic ranges that cannot be easily supported in IoT devices.
  • From a radio spectrum allocation perspective, three different modes of operation are being considered for LTE-M2, which include the option of stand-alone operations in one or more GSM carriers, operations in the LTE guard-band, or in-band operations in a normal LTE carrier.

The key characteristics of the various LTE technologies that are intended for IoT were recently published by Nokia and are compared in Exhibit 6 with LTE-Cat 4, which was standardized by the 3GPP in 2008. LTE-M1 and LTE-M2 are estimated to have 20 percent and less than 15 percent of the modem complexity of LTE-Cat 4 as a consequence of the innovations that are described above.

As is depicted in Exhibit 4 above, we believe that LTE-Cat. 0 will not gain market momentum since it is being eclipsed by LTE-M1 and M2. LTE Cat 1 is positioned to address IoT use cases that require higher bandwidths. The market prospects for LTE Cat 1, M1 and M2 depend on market timing. However, we believe that LTE-M2 is well positioned to support the largest proportion of IoT use cases, potentially limiting the market prospects for both LTE Cat 1 and M1.

Exhibit 5: Comparing LTE Implementations for IoT
Source: Nokia 2015

LPWA Technologies Fill a Void for WWAN IoT Connectivity

The slow response of mobile operators towards the IoT market has created opportunities for new entrant players to deliver low power wireless access (LPWA) solutions that operate in unlicensed spectrum. A slew of companies and technologies have emerged to target LPWA opportunities. Notable examples include Ingenu, Link Labs, M2COMM, nWave, Qowisio, Semtech (LoRa), Silver Spring, SigFox and Telensa. Each of these players have unique technology architectures and market positioning, which impact the business models they are pursuing.

The underlying principle for LPWA solutions is to leverage the improved receiver sensitivity achievable with narrow-band signals. In particular, the minimum signal power at a receiver is determined by its power level relative to that of the noise and interference power that is also received. When the bandwidth of the desired signal is reduced, so to is the in-band noise and interference, which effectively increases the sensitivity of the receiver. Since receivers with higher sensitivities can discern signals at much lower power levels, the effective coverage of a system is increased as the signal bandwidth is reduced.

Ultra-Narrow-Band (UNB) Technologies Test the Limits of Simplicity

Technology companies including nWave, Qowisio, Telensa and SigFox have ultra-narrow-band (UNB) technologies, with data speeds up to 100bps, which are being targeted for sensor networks, precision agriculture, and smart lighting, parking and metering applications. nWave provides UNB radio solutions based on the Weightless-N standard and with the support of systems integrators is targeting smart metering, tracking, precision agriculture and smart-parking applications. Qowisio, Telensa and SigFox have proprietary UNB solutions. Telensa and SigFox have been making efforts to standardize UNB through ETSI, and Qowisio provides dual mode UNB/LoRa network capabilities to broaden the range of IoT use-cases it can address. UNB solutions have proliferated in a fragmented market environment, however we expect this to change as the IoT market matures and large scale competition comes from licensed spectrum solutions.

SigFox raises market attention with lofty plans for a global UNB network

SigFox is a start up player that has a widely publicized ambition to deploy a global UNB IoT network. SigFox was founded in 2009 and has raised four rounds of funding from venture capital and private equity. In its first three funding rounds, it raised €27 million, with Intel Capital and Telefonica Ventures leading the Series B and Series C rounds. It raised USD 115 million from its Series D round from ten investors including Telefonica, SK Telecom Ventures, and NTT DoCoMo.

SigFox's design philosophy has been to simplify its connectivity modules at the expense of increased base station complexity. Its radio modulation scheme uses Binary Phase Shift Keying (BPSK), which is a basic scheme that encodes the transmitted information with the phase changed in a carrier signal. When a SigFox radio transmits a burst of data, it does so at a rate of 100 bits/second with 100Hz channels, to send 12 bytes per message using meager transmission powers ranging between 10 and 13 dBm. Each device can transmit up to 140 messages per day depending on its service plan. To increase service reliability, the device transmits the same information at three different frequencies, with a semi-random frequency hopping scheme. If bi-directional communication is implemented, the device remains in listen mode for a period of approximately 20 seconds for downlink messages. At other times the device remains inactive, which greatly improves its power efficiencies. The downlink messages use a Gaussian Frequency Shift Keying (GFSK) modulation scheme with data speeds of 500 bits/second with 600Hz channels. Since the signals transmitted from SigFox devices are un-synchronized and employ a random access scheme, SigFox base stations listen for and receive signals in parallel within a 200kHz spectrum range. Multiple base stations can receive the same signals, which are resolved in back-end systems to improve system reliability. Comparable random access schemes that are being considered as part of the 5G standardization efforts, will be discussed in an upcoming Tolaga report.

SigFox does not build its own base station or device hardware. Instead it has established an ecosystem of silicon, module, device and infrastructure manufacturers to develop solutions that support its technology.

Rather than target network deployments towards vertical market demands, SigFox's preferred business model is to establish partnerships with incumbent and new entrant network operators to roll out nationwide and wide area network deployments, similar to those of mobile operators. SigFox has networks either deployed or currently under construction in more than a dozen countries. Many of its current deployments are in Western Europe, and it has recently self financed a deployment in San Francisco. Once deployed, both SigFox and its network partners promote the solution to local players looking for IoT connectivity services. These services cost in the order of €1-2 per annum per connection depending on the service level that is provided. Approximately 60 percent of the revenues are paid to the network operator and the remaining 40 percent is paid to SigFox. SigFox's revenues in 2015 are estimated to be €5 million.

In its favor, SigFox has a compelling UNB technology and has gained tremendous market recognition for its ambitious plan to build a global IoT network. It benefits from having built an ecosystem of technology partners, rather than building device and network hardware itself, and has valuable Intellectual Property that can be applied to UNB standards and possibly 5G in the future (e.g. US Patent US 2013/0142191 A1 and US 2014/0321451 A1). However, we do believe that SigFox has challenges that need to be addressed. For example:

  • SigFox presents a relatively high risk proposition for players that invest in its technology. These players must trust that SigFox can maintain a strong market position and continue to support its technology as competition continues to gain steam with both unlicensed and licensed technologies.
  • As the LTE-M2 becomes more readily available it presents a competitive threat for SigFox. This is particularly the case if SigFox is unable to achieve adequate market scale and a competitive cost advantage.
  • While UNB offers compelling economics and operational efficiencies, it is hindered by several limitations, most notably its limited service bandwidth, and its inability to support basic networking protocols and device management functions. Other players like Qowisio have launched dual mode UNB/LoRa capabilities so that they can support devices with higher bandwidth requirements. This is in sharp contrast to SigFox, which has positioned LoRa as a staunch rival, potentially to its detriment.
  • SigFox's long term success depends on its ability to drive its solution into standards. It recognizes that standardization is necessary for the longevity of UNB technologies. We expect that this process will prove difficult, particularly as it requires collaboration with competitors like Telensa and others to drive standardization efforts.

Telensa Seeks to Cherry Pick Smart-City Opportunities

Telensa was spun out of Plextek in 2010 with a focus towards delivering smart street lighting infrastructure solutions using its proprietary UNB radio technology. In January 2016, Telensa raised USD 18 million, which included equity from the Environmental Technologies Fund and debt from the Silicon Valley Bank. Since gaining a market foot-hold with street-light deployments, Telensa has applied its solution to other smart city applications such as smart-parking. Telensa describes its approach to network deployments as being initially targeted towards vertical applications, such as local street-light deployments and then extended to encompass other applications that can leverage the same network.

For street-light services in the UK, Telensa charges service fees in the order of ₤ 2 for each connected light. Its connectivity modules cost approximately ₤10 each and its base station hardware costs approximately ₤ 1000 for each site, which compare favorably with competing licensed spectrum cellular solutions. While Telensa has focused on subscription service opportunities targeted toward the public and commercial sectors, it believes that there are subscription-less opportunities in consumer markets, where network connectivity costs are bundled into the upfront purchase price of the consumer product. A notable example might be consumer electronics and white-ware products. We believe that there are potentially tremendous opportunities in this area, but suspect that many of the use-cases have greater bandwidth requirements than achievable with UNB solutions.

In contrast to SigFox, Telensa designs and builds its own radio hardware technology, with vertically integrated solutions that are targeted towards the specific markets that it is addressing. Its radio solution provides bi-directional connectivity using 2-Frequency Shift Keying (2-FSK) modulation with 500 and 60 bits/second on the uplink and downlink connections respectively. Telensa has evolved its platform from proprietary technology which was originally developed by Plextek, for which a variety of patents were filed, including US 8130681 B2 and US 8184748 B2.

Telensa has been successful in expanding its market presence by capitalizing on the positive business cases that can be achieved from well designed smart-city applications. It has deployed regional networks in eight countries and is well positioned to expand its deployments into other markets and leverage its deployed networks for other services. However, there are several areas that limit Telensa's opportunities, which include the following:

  • By having a proprietary and vertically integrated solution, Telensa's ability to respond to market opportunities is primarily determined by its own resources. We believe that it is crucial for Telensa to license its technology to other players and broaden the scope of its ecosystem. In addition, we believe that it would be beneficial for Telensa to work with strategic partners to deploy and operate its technology outside its own network footprint.
  • Telensa's strategy to target its network deployments towards vertically integrated solutions for smart-city applications is likely to stifle its ability to scale. We believe that market scale is crucial for Telensa's long term success, particularly if and when the market matures, and consolidates towards standardized solutions.
  • While Telensa has only recently raised USD 18 million in capital, we expect that it will need additional capital within the next 24 months to continue to expand at a competitive rate relative to the overall market.
  • While Telensa's UNB technology has been effective for the use cases it is addressing, it is greatly limited in terms of bandwidth performance and might need the base station support of a higher bandwidth technology like 802.15.4 or LoRa, in the future.

nWave Accelerates its Standardization Efforts with Weightless-N

nWave Technologies was founded in 2010 and has a UNB IoT solution, which was proprietary prior to it contributing its intellectual property to the Weightless Special Interest Group (SIG) for the standardization of Weightless-N in October 2014. At the time, Weightless SIG had been focused solely on the standardization of its Weightless-W standard, which provides IoT connectivity using TV white- space spectrum. However, delays and complexities with TV white-space spectrum necessitated that Weightless broaden its remit, culminating in the alliance with nWave to develop Weightless-N for UNB connectivity, and more recently an alliance with M2COMM to develop the Weightless-P standard.

In October 2015, nWave raised USD 500 thousand US Dollars in seed funding from Momenta Partners and Cisco and published its initial SDK for Weightless-N, which uses a Differential Binary Phased Shift Keying (DBPSK) modulation scheme. Similar to other UNB technologies, Weightless-N also incorporates frequency hopping for interference mitigation.

Since nWave has aligned itself with Weightless SIG, it has been targeting smart-city opportunities for Weightless-N, with the key value proposition that it is standards based. To date much of this activity has been focused towards Western European cities and has deployments in London and Copenhagen. In addition, nWave has also capitalized on its standards based approach to ink partnerships with several players including Arkessa, which is a managed IoT service provider and PlatOne an application platform provider who is developing smart parking solutions in conjunction with nWave.

While nWave should be commended for pursuing UNB standardization with Weightless-N, we believe that its market momentum will continue to be hindered by companies like SigFox who have much greater penetration, and capital resources, and a vastly superior ecosystem. nWave is also vulnerable, should SigFox and Telensa be successful in their efforts to standardize with ETSI. To increase its chances of success, we do not believe that nWave is wise in building out networks in Western European markets where players like SigFox and Telensa are strong, and with smart-city solutions that are already highly competitive. A market strategy that might prove more viable for nWave would be to focus its attention towards greenfield markets outside of Western Europe, with solutions that are targeted towards use cases that are not being addressed by traditional UNB providers. For example, nWave might focus on consumer electronics and white-ware, with subscription-less service models that bundle connectivity services in the purchase prices of the connected products. To enable this strategy, nWave will require a significant capital injection beyond the USD 500 thousand seed funding that it raised in October 2015.

Qowisio Disrupts UNB Business Models and Technology Debates

Qowisio was founded in 2009 and has deployed 18 private networks for customers in 29 countries spanning Eastern Europe, the Middle East and Africa. In June 2015 it raised € 10 million to fund a public LPWA network deployment in France. To date Qowisio has primarily focused on solutions for building automation and energy management, and its revenues in 2015 were estimated to be € 15 million. Qowisio is departing from the subscription based business models being pursued by companies like SigFox and Telesna, and is bundling the connectivity as part of the cost of the connected items. In addition, rather than fueling technology debates comparing UNB with LoRa, Qowisio is deploying a dual mode UNB/LoRa network in France.

Qowisio must contend with competitors like SigFox. SigFox is better capitalized, with a superior network footprint and a more mature ecosystem. Qowisio's technical and commercial strategies are well positioned relative to the future direction of the industry, however if successful, these strategies can be easily replicated by Qowisio's competitors. For this reason, we believe that Qowisio's continued focus on vertical market solutions, including those associated with building automation and energy management with private networks, are less risky and likely to be more lucrative than the public network strategy that it is currently pursuing.

Balancing Bandwidth with Performance for LPWA

While UNB technologies achieve tremendous network coverage with cost effective and power efficient devices, limited connection bandwidths constrain the UNB use cases for IoT. A variety of companies including Semtech (LoRa), Silver Spring, Link Labs, Ingenu, and M2COMM are providing connectivity solutions that are capable of achieving higher throughput when radio conditions allow, albeit with increased device complexity. These solutions use a variety of proprietary and standards based spread-spectrum radio technologies and networking innovations, which enable up to several hundreds of kbps throughput. Companies providing these solutions have varied business models that primarily depend on their position in the value chain, capital resources, capabilities of their LPWA technology and target markets.

Semtech Fuels LoRa with Silicon Technology, Licensing Schemes and the LoRa Alliance

Semtech is a publicly listed company which develops analog and mixed signal semiconductor products. It has had annual revenues ranging between USD 550 and 600 million dollars over the last three years for solutions targeted towards consumer electronics, communication and data center infrastructure. Semtech entered the LPWA market through its acquisition of the LoRa technology from Cycleo in 2012 for USD 5 million. Semtech has established an open development environment under the guise of LoRaWAN, which is administered by the LoRa Alliance. The LoRa Alliance is supported by a variety of companies including IBM and Cisco and LoRaWAN provides higher layer protocols for systems that use Semtech's LoRa radio technology. While the LoRa Alliance intends for LoRa to be implemented with LoRaWAN, LoRa can also be integrated into other environments. For example, Link Lab's Symphony platform has LoRa radio technology, but uses the IEEE 802.15.4 for higher layer protocols.

The LoRa radio technology uses a frequency modulated (FM) chirp spread-spectrum scheme for a cost effective radio design with relatively sensitive receivers for improved coverage and capacity. LoRa radio channels range in bandwidth between 125 and 500kHz. Key innovations associated with LoRa's chirp spread-spectrum technology are described in the US Patent – US 7791415 D2. LoRa is agnostic to the higher layer protocols that are used, and can be implemented with mesh or star network architectures and networking protocols such as 6LoWPAN.

LoRa incorporates an adaptive data rate (ADR) scheme so that data rates can be varied between 0.3 and 100kbps depending on link reliability. It supports three device classes that are intended for different operating conditions and performance requirements.

All LoRa devices have bi-directional connectivity, however techniques for enabling this connectivity varies amongst the device classes. Class A devices have the lowest power footprint, with the device initiating both uplink and downlink communications. For Class A devices, an uplink transmission is followed by two short downlink receive windows. The device determines the transmission slot that is used for uplink transmission and for improved reliability includes a small random time variation using an ALOHA-type protocol. A similar scheme has been adopted by SigFox for its devices that have bi-directional communications. Class B devices enable more opportunities than Class A devices for downlink connectivity by including an extra receive window that can be scheduled with a predetermined regularity. Class C devices can receive downlink signals whenever they are not active with uplink communications. However, to achieve this Class C devices must remain in listen mode even when they are not transmitting. For this reason, they are less power efficient than Class A and B devices, but have the lowest latency for downlink communications.

While LoRa is a strategic endeavor for Semtech, it currently only contributes a single digit percentage to its overall company revenues, with the lion's share of company revenues being diversified amongst consumer electronics, communications, industrial and enterprise computing markets. Semtech's earnings from LoRa were USD 5 million in 2014 and expected to be USD 15 million 2015. In January 2016, Semtech forecasted that its earnings from LoRa would increase to USD 30 million in 2016 and to USD 100 million by 2018.

Semtech and the LoRa Alliance are benefiting from the robust ecosystem that they have established to support the industry's adoption of LoRa. Industry adoption is also being fueled by positive market sentiment towards LoRa, the LoRaWAN ecosystem and IoT use cases that LoRa can address. However there are several challenges that we believe will confront Semtech as it endeavors to promote LoRa in the maturing IoT market. In particular:

  • We believe that LoRa will benefit greatly with the support of an increased number of semiconductor providers. Semtech initially pursued a strategy for which it was the sole supplier of LoRa radio semiconductors, but has more recently licensed its technology to MicroChip and ST Microelectronics. This licensing is in response to LoRa customer demands, and has aided in LoRa's recent market success. To continue this success in the future, we believe that Semtech will need to extend its licensing agreements to include additional players, and ultimately towards enabling open standards.
  • In contrast to SigFox, LoRa radio base stations can be purchased by individual players and deployed independently, potentially creating interference and coordination challenges between networks. These challenges are likely to increase as LoRa networks and devices are more widely adopted.
  • LoRa is vulnerable to competition providing UNB solutions, such as SigFox. This is particularly the case if the industry gravitates towards ultra-low-cost devices for unlicensed spectrum operations.
  • While LoRa is being deployed by mobile operators, is vulnerable to future competition for LTE-M1 and M2 as the technologies are standardized by the 3GPP and their capabilities deployed in LTE networks. We believe that the most appropriate avenue for LoRa to compete is through carefully crafted business models. A notable business model might involve networks with standalone LoRa or integrated LTE/LoRa capabilities, and neutral host and subscription-less service offers. These business models will be investigated in a subsequent Tolaga report.

Ingenu pursues lofty plans for its “Machine Network”

Ingenu (previously OnRamp) was established in 2008 with a proprietary technology (US Patent US 7782926 B2 ) it refers to as Random Phase Multiple Access, (RPMA), which is essentially derived from Direct Sequence CDMA that was pioneered by Qualcomm for 3G. Ingenu has a strong pedigree emanating from its roots in Qualcomm, where RPMA was originally hatched. The company also has the endorsement and advisory support of industry titans including Andrew Viterbi (Qualcomm co-founder), and Ivan Seidenberg, the former CEO, and Dick Lynch the former CTO of Verizon. To date Ingenu has raised USD 118 million in funding, of which USD 5.5 million is from debt financing. Notable investors include ConocoPhillips, Enbridge, GE Ventures, NRG Energy and Third Wave ventures.

Ingenu has gained some market traction in the oil and gas industry, and with smart-metering applications. It has also established a similar network deployment strategy to that of SigFox, but with a technology that more closely resembles LoRa. Ingenu has announced plans to deploy what it refers to as the “Machine Network”, which has the lofty objective of delivering global IoT connectivity using Ingenu technology, with an initial focus towards the US market.

Ingenu has patented an innovative dynamic direct sequence spread-spectrum (D-DSS) technique, which randomizes timing to enable multiple devices to use the same spreading code, and dynamically varies its processing gain to optimize the coverage range and bandwidth of the signal transmitted.

Exhibit 7 illustrates a conventional direct sequence spread-spectrum modulation scheme. At the transmitter, narrow band signals are modulated with a spreading code, which spreads the energy of the narrow-band signal over a wide bandwidth range. The spreading code has unique characteristics to improve the signal transmission and reception processes. At the receiver, the same code used to spread the transmitted signal is used for the de-spreading process. As a consequence the de-spreading process effectively spreads incoherent interference.

Exhibit 7: Illustration of Direct Sequence Spread Spectrum
Source:Tolaga Research 2016

The extent of signal spreading in a spread-spectrum system is measured in terms of the processing gain, which represents the ratio of the bandwidth of a spread-spectrum signal relative to the narrow-band signal that it is transporting. The achievable processing gain is determined by the difference between the bandwidth of the narrow-band signal being transmitted, and the channel bandwidth of the spread-spectrum system. If the bandwidth of the narrow-band signal is reduced, a spread spectrum system can have a higher processing gain in addition to improved receiver sensitivity. Higher processing gains enable systems to transmit narrow-band signals with higher power levels without compromising unlicensed spectrum regulations. Ingenu's RPMA technology adapts its processing gain to dynamically balance channel throughput to optimize throughput and coverage demands.

While Ingenu lacks the market presence and momentum of its competitors, it is relatively well capitalized and brings an innovative and compelling solution that builds on many years of industry experience with DS-CDMA technology. However, it is confronted with a variety of challenges, including:

  • The need to diversify execution risk amongst ecosystem partners. We believe that the market momentum of both SigFox and Semtech could not have been achieved without the technology ecosystems that the respective companies have developed for their LPWA technologies. To reap these benefits, Ingenu needs to follow suit.
  • The RPMA radio technology is relatively complicated and depending on how it is implemented, has the potential to be more costly and power hungry relative to competing technologies. This might put the technology at a competitive disadvantage for many IoT use cases.
  • Ingenu's Machine Network is a high risk proposition for players that invest in its technology. These players must trust that Ingenu can maintain a strong market position and support its technology in the face of expensive network deployment demands and competition from other unlicensed and licensed technologies.

Silver Spring Networks drives IoT connectivity with LPWA mesh

Silver Spring Networks was founded in 2002 and was initially focused on building networks for smart-grids and more recently for gas and water metering and smart-street-lights. In 2015 it earned USD 489 million in revenues, and in December 2015 it was reported to have 22.9 million devices connected to its network.

LPWA network technologies like SigFox, LoRa and Ingenu have star-architectures, where base stations are deployed to provide geographical coverage for IoT device connectivity. In contrast, Silver Spring's network uses a mesh architecture, based on the IEEE 802.15.4g standard, where IoT devices communicate amongst themselves to enable network connectivity. This standard has also been adopted by the Wi- SUN Alliance, which is seeing growing acceptance in the utilities sector and has a particular emphasis towards smart-grid use-cases.

Mesh and star network topologies are compared in Exhibit 8 for the hypothetical case where five IoT devices require network connectivity. Since the mesh-network architecture creates connections amongst the IoT devices themselves, it eliminates the need to deploy base stations throughout the network coverage area, as is required for traditional star network architectures. Silver Spring is using the 22.9 million connected IoT devices to create its mesh network. Exhibit 8 also depicts a hybrid star/mesh architecture, where mesh capable IoT devices, (e.g. smart-street-lights and smart-meters) provide a network backbone to which conventional IoT devices can connect using star configurations.

In December 2015, Silver Spring announced its Star-Fish network initiative, for which it plans to leverage existing smart-grid and smart-street-light networks to deliver other services using a hybrid mesh/star network architecture. With its Gen5 network Silver Spring touts throughput speeds of up to 2.4Mbps, with which it plans to enable services outside of the utilities industry, such as transportation, security, building automation and other smart-city applications.

Exhibit 8: Mesh and Star Network Topologies
Source:Tolaga Research 2016

While mesh network architectures are compelling, they have several limitations, which include the following:

  • The communication and networking requirements for each IoT device tends to be more complicated, power consuming and expensive than those associated with star network architectures. These limitations are of less concern for smart-utility applications, where power sources are readily available and device costs can be subsidized against the business cases that are enabled (e.g. 3-5 year payback for smart-street-lights). This is not the case for many other IoT use-cases.
  • To operate reliably, mesh-networks depend on its devices having relatively static radio propagation conditions. Stationary items such as street-lights and smart-meters are well suited to meshing. However, this is typically not the case for devices that are in motion, such as connected cars, or devices that are vulnerable to changes in radio propagation characteristics, such as smart-parking sensors, when they are shadowed by parking vehicles.
  • Security demands tend to be greater for mesh-networks, since each IoT device in the network is not merely responsible for transporting its own data, but also that of other devices. This not only poses challenges in protecting the integrity of the data being transported, but also requires that the network itself is protected from a variety of attack vectors, such as Sybil, sink-hole, selective forwarding, and worm-hole attacks. This is discussed in Tolaga's September 2015 Report, entitled Navigating the Turbulent Seas of IoT Security.

Silver Spring Networks is well positioned to expand its market opportunities for smart-grid and smart-street-light applications and to capitalize on its StarFish platform to deliver smart city solutions. However, as it pursues this strategy, it must contend with a variety of challenges, which include the following:

  • By having a concentrated customer base with large and highly regulated utility companies, Silver Spring's strategies are constrained by the pace of innovation and strategies their customers.
  • As Silver Spring touts its StarFish Gen5 offering, it is promoting 2.4Mbps network speeds. While we don't doubt that StarFish achieves relatively high bandwidths, we believe that Silver Spring it is at risk of under-whelming the market and falling into a similar trap that hindered Municipality broadband initiatives in the past.
  • Silver Spring will see increased competition from mobile operators in the future as licensed spectrum solutions such as LTE-M1 and M2 mature. We believe that utility and smart-city applications will be more aggressively targeted by mobile operators to enable subscription based recurring revenue streams. We also expect that mobile operators will seek street-light and utility infrastructure for broader small-cell network deployments.

M2COMM Leverages Platanus for Weightless-P Standardization

M2COMM is a Taiwanese based fab-less semi-conductor and module provider that was established in 2012 to develop connectivity solutions for the IoT industry. Its proprietary radio solution, Platanus, is targeted primarily for local area network implementations with connectivity speeds of up to 500kbps. It has developed wireless solutions for retail, factory, warehouse, logistics, digital signage and smart-parking applications.

In July 2015, M2COMM announced plans to develop the Weightless-P standard, which is intended to complement Weightless-N. Weightless-P offers higher bandwidth capabilities with an architecture that leverages technical innovations associated with the Platanus platform. Rather than using 200Hz channels, as is the case of Weightless-N, Weightless-P has 12.5kHz channels, with specifications that more closely align those of LoRa. In particular, Weightless-P has adaptive data rates (ADR) between 200bps and 100kbps and a maximum transmission power of 17dBm, and includes channel hopping capabilities to avoid interference. In addition, Weighless-P uses Gaussian Minimum Shift Keying (GMSK) and offset Quadrature Phase Shift Keying (O-QPSK) modulation schemes.

While Weightless-P comes with standards based specifications, we do not believe that it is sufficiently differentiated from technical or commercial perspective to compete with LoRa. The technical specifications of Weightless-P are similar to LoRa, but it lacks a robust ecosystem and has yet to gain market traction. We believe that the success of Weightless-P depends on its ability to disrupt the business models being adopted by LoRa and deficiencies in the LoRa technology. Notable examples, might include support for subscription-less business models that are targeted towards consumer durable products, management schemes that mitigate inter-network interference, and optimized Weightless-N/Weightless-P architectures.

LPWA has awakened a sleeping giant

In the past, LPWA advocates have benefited from inaction from the mobile industry to gain a foot-hold in the IoT market. However, now that the mobile industry has started to take the IoT opportunity more seriously, the competitive landscape is set to change, particularly in light of the 3GPP standardization efforts and mobile operators' increase use of unlicensed spectrum technologies. We believe that this will have a tremendous impact on the LPWA landscape.

Even though the IoT industry is still nascent, it has already started to narrow its focus towards specific LPWA technologies. We believe that this focus and the success of particular LPWA solutions will be determined by industry heavy weights, most notably Tier 1 mobile operators and infrastructure, semi-conductor and consumer electronics vendors, large utility companies and systems integrators.

However, unlike Intel's famous Centrino initiative which resulted in Wi-Fi becoming the de-facto standard for unlicensed wireless broadband, we believe that multiple LPWA technologies will prevail to support the various IoT use cases. For example, the utility industry is currently gravitating towards the Wi-SUN standards and IEEE 802.15.4 for smart-grid applications and several large mobile operators, including Orange, SK Telecom, KPN Mobile, Swisscom and Tata have commenced LoRa deployment initiatives. SigFox has the greatest momentum amongst UNB LPWA players, but is somewhat hamstrung by its business model and by the limitations of its technology.

Many industry commentators compare the fate of LPWA to that of WiMAX, and while there are parallels, we believe that this comparison is an over-simplification of the reality. When WiMAX went head-to-head with the mobile industry, it was driving an agenda (fueled primarily by Intel) to enable the mobile Internet. WiMAX could not succeed without the widespread availability of licensed spectrum and the support of mobile operators and their ecosystem players. The business models touted for WiMAX were disruptive to these players, but fueled LTE standardization efforts under their stewardship.

In essence WiMAX was constrained by spectrum regulation and captive to the mobile industry that it was seeking to disrupt. Neither of these conditions apply to LPWA, since the technologies operate in unlicensed as opposed to licensed radio spectrum and is not dependent upon the mobile operators for its deployment. Industry verticals, such as the utilities industry are making technology decisions independent of the mobile industry, as is illustrated by the WiSUN Alliance and other efforts that include both unlicensed and licensed spectrum technologies. An increasing number of smart-city applications are using unlicensed spectrum technologies and even mobile operators themselves are increasingly seeking unlicensed technologies to complement their licensed spectrum deployments. Interestingly, WiMAX has more recently been chosen as a technology standard for the aviation industry.

Since LPWA solutions operate in unlicensed spectrum, they are prone to performance and reliability issues attributable to interference from other systems operating in the same band. As licensed spectrum solutions such as LTE-M1 and M2 become more readily available, the value proposition for unlicensed spectrum systems will become more challenging.

Conventional business models that pit LPWA network technology costs against those of cellular technologies will no longer be adequate. Instead business models that position LPWA to address use cases that are either complementary or disruptive to competing cellular solutions are needed. For example, these use cases might target neutral host business models to simplify the supply chain, manufacturing and enablement of the connected devices, or they might focus on delivering regional services in areas where LTE coverage is unavailable (e.g. to support precision agriculture applications in rural areas).

Security is commonly identified as an area where LPWA systems are challenged relative to licensed spectrum solutions. However, we do not believe that LPWA technology is intrinsically less secure, but rather it is the manner in which the LPWA systems are implemented that potentially creates security vulnerabilities. In particular:

  • Commonly the security licensed versus unlicensed spectrum solutions is compared in terms of their respective radio technologies (i.e. the Physical Layer). To initiate an attack an adversary might scan the spectrum band to find the radio channels that are used and then attempt to jam them with the aim of disrupting connectivity.
    Radio scanners and wireless transmission modules are more readily available for unlicensed spectrum bands, however this does not preclude an attacker from being capable of targeting licensed spectrum in the same manner.
    Furthermore, unlicensed spectrum systems commonly use spread-spectrum techniques that tend to be more resilient to signal jamming than the OFDMA techniques used for LTE, except in cases where LPWA solutions challenging link-budgets.
  • In many cases, attackers interrogate networks to identify and target architectural weaknesses. These weaknesses can exist irrespective of whether the network uses licensed or unlicensed radio spectrum. Examples, might include the poor management of security credentials, inadequate or non-existent encryption, and network nodes and end-points that can be compromised because they have not been sufficiently hardened.
    Mobile network infrastructure vendors and mobile network operators have been subject to increased scrutiny in recent years. This has resulted in significant investments in techniques to increase the security of their infrastructure. However mobile operators tend to be higher value targets than smaller public and private networks.

Although LPWA has been developed in unlicensed spectrum, efforts are being made as part of 5G to enable solutions similar to LPWA that are capable of operating in licensed spectrum bands. These efforts underline a primary 5G objective to bring a unified radio network environment to support the wide range of current and emerging wireless solutions including narrow-band IoT. While it is anticipated that OFDMA will be the workhorse for 5G, many researchers and contributors to the 5G standards are calling for non-orthogonal spread-spectrum solutions to be used either in-band or within the guard-bands of 5G solutions. A notable example is Qualcomm's proposed Resource Spread Multiple Access (RSMA) solution, which will be discussed in an upcoming Tolaga report.

We believe that as ETSI/3GPP look to standardize 5G it might leverage Intellectual Property developed by LPWA providers like SigFox, Semtech, Ingenu and others. As this standardization occurs, proprietary solutions developed by these companies will be eclipsed and ultimately superseded by 5G or standardization efforts with an organization like IEEE that is specifically targeted towards unlicensed spectrum applications.

Conclusions and Recommendations

Today, the landscape for WWAN IoT connectivity is fragmented and complicated, with many players looking to grab a piece of the action using unlicensed spectrum technology and the mobile industry awakening to the IoT opportunity. As the IoT industry matures, consolidation is inevitable, as are technical and commercial casualties. Even solutions that have a solid market position today are vulnerable to changing industry dynamics that favor standards based solutions, capable of reliably and economically supporting a sufficiently wide variety of IoT use-cases and business models. These dynamics will be dictated primarily by industry heavy-weights that have the market scale and resources to commercialize IoT technologies and business models.

Even though IoT currently represents a small revenue opportunity, it is becoming strategic for mobile operators. This has culminated in feverish standardization efforts by the 3GPP to create a somewhat confusing technology landscape, which is complicated by mobile operators' plans to decommission their 2G and 3G networks and competing LTE variants. However, as the dust settles, we expect that LTE-M2 (aka NB-IoT) will be the dominant licensed spectrum technology for IoT, assuming it meets its price/performance objectives, and its standardization is completed and market adoption well underway within the next 24-36 months. Even with LTE-M2 being dominant, we still see significant demand for LTE-M1 and other LTE variants to support higher bandwidth applications. In addition, the success of LTE depends on mobile operators' ability to evolve their business models and service plans to better align with IoT demands. Notable examples include service plans with extremely low (or free of) subscription fees, service distribution and operational models that eliminate redundant functions that are needed for traditional mobile services, and embedded SIM technology to ease friction in the supply chains and distribution channels.

Over the next 12-24 months, LPWA providers will be confronted with a dramatically changing competitive landscape. This change will be fueled by a variety of players including:

  • Mobile operators who will deploy solutions that operate in both licensed and unlicensed spectrum.
  • Large industry players, such as silicon vendors, utility companies, and infrastructure vendors who nominate their preferred technology solutions for the vertical markets they target, and;
  • End users, in cases where break-though solutions emerge.

To remain competitive, LPWA providers cannot rely solely on their technology performance for differentiation, but instead their strategies must focus on business model opportunities to drive market scale and lock-in. We also expect that LPWA technology standardization is inevitable and as this occurs, market scale will be the primary driver for success. While some LPWA provdiers are benefiting from the comfort of subscription models, which drive recurring revenues, we believe that greater scale will come from business models which bundle the connectivity costs in upfront equipment prices.

As the market for WWAN IoT connectivity solutions consolidates, we do not believe that one solution will emerge as victorious in supporting the needs of IoT. Rather than anticipating an IoT equivalent of Intel's Centrino, we believe that several licensed and unlicensed spectrum standards will be needed to support the IoT industry, with LTE-M1 and M2 taking the reigns for licensed spectrum implementations. While SigFox and LoRa have an early market lead in supporting unlicensed spectrum technologies, with LoRa being ahead, there is still the potential for another technology to come to the fore, particularly if it gets strategic support from industry heavy-weights.

;