LPWAN Technologies for IoT: Real-World Deployment Performance and Practical Comparison

Abstract

Low Power Wide Area Networks (LPWAN) have emerged as essential connectivity solutions for the Internet of Things (IoT), addressing requirements for long range, energy efficient communication that traditional wireless technologies cannot meet. With LPWAN connections projected to grow at 26% compound annual growth rate until 2027, understanding real-world performance is crucial for technology selection. This review examines four leading LPWAN technologies—LoRaWAN, Sigfox, Narrowband IoT (NB-IoT), and LTE-M. This review analyzes 20 peer reviewed studies from 2015–2025 reporting real-world deployment metrics across power consumption, range, data rate, scalability, availability, and security. Across these studies, practical performance diverges from vendor specifications. In the cited rural and urban LoRaWAN deployments LoRaWAN achieves 2+ year battery life and 11 km rural range but suffers collision limitations above 1000 devices per gateway. Sigfox demonstrates exceptional range (280 km record) with minimal power consumption but remains constrained by 12 byte payloads and security vulnerabilities. NB-IoT provides robust performance with 96–100% packet delivery ratios at −127 dBm on the tested commercial networks, and supports tens of thousands devices per cell, though mobility increases energy consumption. In the cited trials LTE-M offers highest throughput and sub 200 ms latency but fails beyond −113 dBm where NB-IoT maintains connectivity. NB-IoT emerges optimal for large scale stationary deployments, while LTE-M suits high throughput mobile applications.

1. Introduction

LPWAN are a foundational element of the modern IoT, addressing the need for long-range, low-power, and low-bandwidth connectivity that traditional cellular and short-range wireless technologies fail to meet. LPWAN technologies are specifically engineered to support battery powered devices that transmit small amounts of data infrequently, making them ideal for a large number of uses, for example: smart cities and buildings [1], agriculture [2], industrial monitoring [3], different emerging edge AI frameworks [4] and usecases [5], and other IoT applications requiring extended range, low cost, and long battery life [6]. LPWAN connections are projected to grow at a 26% compound annual growth rate until 2027, reaching 3 billion connections and representing 10% of all IoT connections worldwide [7].

This review focuses on four leading LPWAN technologies: LoRa, Sigfox, NB-IoT, and LTE-M. These technologies were selected based on their proven existing deployment scale, tooling maturity, and degree of standardization. These criteria reflect the practical engineering realities of building and maintaining scalable IoT systems, where field proven reliability, accessible development infrastructure, and consistent protocol definitions are essential.

In terms of existing deployment scale, LoRa has demonstrably achieved broad commercial use. Examples such as The Things Network, with their LoRaWAN network achieved 1 million connected IoT devices across the world [8], confirm LoRa’s operational maturity. NB-IoT and LTE-M, as part of the 3rd Generation Partnership Project (3GPP) [9], have steadily grown in deployment as operators integrate them into national infrastructure [10,11]. For example, operators such as AT&T, KPN, Orange, Swisscom, Vodafone and Deutsche Telekom have operational NB-IoT and LTE-M networks for a few years now [12,13,14,15]. Sigfox with its Sigfox 0G network has achieved significant global coverage spanning 5 million km² across 70+ national 0G IoT Solution Providers, reaching approximately 1 billion people worldwide currently having 14 million connected devices [16]. Sigfox adoption remains fragmented, with only about 10 million devices deployed worldwide as of 2023, compared to LoRaWAN’s 350 million+ end nodes by June 2024 [17,18].

Tooling maturity differs across LPWAN technologies, but all the main options offer enough ecosystem support. LoRaWAN is especially strong in this area, thanks to its formal certification programs and well maintained regional standards from the LoRa Alliance, which help ensure devices work together as expected [19]. Developers have access to a wide variety of kits, open source stacks like LoRaMAC-Node, LoRa Basic Modem and cloud integration tools [20,21]. NB-IoT and LTE-M are tightly integrated with the established 3GPP ecosystem, which means they benefit from solid backing by major chipmakers and mobile operators [22]. There are plenty of commercial development boards and SDK available, though the quality of integration tools can vary depending on the vendor. The fact that NB-IoT and LTE-M continue to be enhanced in recent 3GPP releases is a good sign of their maturity and stability [22]. Sigfox does provide development kits, platform access, and community support, but its ecosystem is smaller and less widely adopted than LoRaWAN or the 3GPP based options [23]. As a result, developers may find fewer tools and resources for Sigfox projects.

The standardization landscape for LPWAN technologies reveals distinct tiers of maturity. NB-IoT and LTE-M are fully standardized under 3GPP, with global specifications refined across multiple releases. Both LTE-M and NB-IoT were standardized in 3GPP Release 13 and have continued to evolve through subsequent releases to support enhanced features, mobility, and data rates [22,24]. LoRaWAN, though not an international telecom standard, operates under the LoRa Alliance’s structured framework, which defines regional frequency plans, back-end interfaces, link layer, firmware update over-the-air, mandatory certification processes [25]. Sigfox, while proprietary, has established documented technical standards for frequency, security, message sequencing, and device authentication, enabling it to function as a de facto standard in niche markets, albeit without multi-vendor standardization bodies [26].

The exclusion of Wi-SUN, ZETA, and Weightless from mainstream LPWAN comparisons reflects their limited deployment scale and ecosystem maturity, not necessarily technical merits. Wi-SUN demonstrates partial traction in Europe, with deployments such as London’s 12,000 Wi-SUN-enabled streetlights [27]. The Wi-SUN Alliance reports operating in 46 countries with 300 members globally, including European entities [28]. While Wi-SUN’s mesh networking and 50 kbps data rates suit smart grids and street lighting, academic studies highlight Wi-SUN’s scalability challenges, such as large 50+ min formation times for node networks in size of 100+, limiting its viability for large scale IoT deployments [29,30].

The focus on NB-IoT, LTE-M, LoRaWAN, and Sigfox in this report is justified by robust standardization frameworks and proven commercial viability. These technologies have achieved critical mass in terms of research attention, industry support, and market deployment that distinguishes them from alternative LPWAN solutions. Although it must be noted that United States AT&T decided to discontinue NB-IoT services from Q1 2025 [31] and is moving its business IoT customers to an LTE-M plan, while European operators are showing stronger commitment to both NB-IoT and LTE-M.

2. Basics of LPWAN Technologies

LPWAN have emerged as a foundational technology for wireless sensor networks, IoT, and Machine-to-Machine (M2M) communication systems which are often used by resource constrained devices characterized by demand for extended operational longevity, wide area coverage, high scalability, and cost effective deployment. These networks are specifically engineered to address the unique constraints of IoT endpoints, which often operate on limited energy budgets while requiring reliable connectivity across large geographical areas. LPWAN also addresses challenges such as medium access control (MAC) protocol optimization for massive device scalability, spectrum management under region-specific regulatory constraints (e.g., EU duty cycle limitations [32]), link-layer adaptability to various propagation environments [33], coexistence with other wireless systems and security/privacy risks. These constraints necessitate continuous innovation in network architecture to balance energy efficiency and QoS (quality of service), positioning LPWAN as both a enabler and active research domain in modern IoT systems.

When deploying LPWAN technologies for IoT applications it’s important to note that these technologies offer data rates typically ranging from as low as 0.1 kbps to around 1 Mbps, depending on the specific standard. They’re optimized for small payloads, generally between 11 and 1280 bytes. Latency varies significantly: licensed cellular options like LTE-M can deliver sub 60 ms performance, NB-IoT can vary from 1.2 to 100 s, while unlicensed systems such as LoRaWAN and Sigfox experience delays of several seconds [34].

2.1. LoRa and LoRaWAN

LoRa (Long Range) technology was first developed by Semtech Corporation [35] as a physical layer wireless modulation technique based on chirp spread spectrum (CSS) modulation. LoRa provides the foundation for LoRaWAN, which defines the upper layers of the network protocol stack.

LoRa physical layer technology works in unlicensed sub-GHz ISM frequency band—868 (Europe), 915 (North America), 433 MHz (Asia). Base of this technology CSS is frequency modulation in which carrier frequency varies for a defined extent of time. The main characteristic of this type of modulation is a tradeoff between receiver sensitivity and data rate while using fixed 125 kHz, 250 kHz or 500 kHz channels. It uses orthogonal spreading factors (SF) to adjust transmission parameters. A higher SF increases sensitivity, improving the ability to detect weak signals over long distances, but reduces data rate. For example, LoRa device near gateway should use low spreading factor, while distant devices should use higher spreading factors for better sensitivity at the cost of lower data rates.

The LoRaWAN performance metrics as promised by the LoRaWAN Alliance [36] are collected in the Table 1.

2.2. Sigfox

Sigfox is proprietary LPWAN technology owned by UnaBiz [37]. It provides a complete protocol stack that handles both physical layer transmission and upper network layers. Physical layer of this technology uses ultra-narrow band (UNB) modulation that has extremely narrow bandwidth of only 100 Hz per message that reach data rates of 100 to 600 bits per second with payload limited to 12 bytes. End devices are allowed to transmit up to 140 messages per day [38]. Sigfox physical layer technology works in unlicensed sub GHz ISM and SRD frequency bands worldwide—for Europe, Middle East, parts of Africa 868.130 MHz uplink, 869.525 MHz downlink and 865 to 923 MHz in South, North America, Asia and others. Sigfox is not available worldwide [39].

The main characteristic of this modulation is the tradeoff between extremely long range and very low data rates while using a total spectrum allocation of only 192 kHz. Sigfox encodes data using differential binary phase shift keying (DBPSK) for uplink transmissions and Gaussian frequency shift keying (GFSK) for downlink communications [40]. The system uses frequency hopping by transmitting each message three times on different pseudo random frequencies to ensure reliable packet delivery [39]. Sigfox payload is limited to 12 bytes uplink (UL) and 8 bytes downlink (DL).

The Sigfox network protocol defines a simplified communication model without device classes, instead operating on a star network topology where devices are not attached to specific base stations and base stations continuously monitor the spectrum for UNB signals. Base stations relay messages to the Sigfox cloud backend through various connections, and the cloud interfaces with application servers [37].

The Sigfox performance metrics as promised by the UnaBiz are collected in the Table 1.

2.3. NB-IoT

NB-IoT is a cellular LPWAN technology standardized by 3GPP in Release 13 in June 2016 [41]. It provides a complete protocol stack that handles both physical layer transmission and upper network layers. Unlike unlicensed LPWAN solutions, NB-IoT operates within licensed cellular spectrum, typically using a 180 kHz channel inside existing LTE or GSM bands. This approach allows NB-IoT to use existing cellular infrastructure for robust and wide area connectivity. NB-IoT offers about 50 kbps downlink and uplink maximum data rates with payload limited to 1280 bytes [42]. NB-IoT promises battery life of up to 10 years.

NB-IoT is primarily designed for stationary or low mobility applications and does not support seamless handover between cell towers, making it unsuitable for devices that require continuous connectivity while moving. 3GPP Release 14 introduced connection re-establishment procedures for connected mode (as opposed to the power-expensive operation of going into idle mode when moving from one cell to another per Release 13), though full mobility support remains limited compared to LTE-M [43].

Similarly to LoRaWAN, this technology has a trade off between coverage and data rate. NB-IoT uses orthogonal frequency-division multiplexing (OFDM) for downlink communications and single-carrier frequency division multiple access (SC-FDMA) for uplink transmissions. The network can be deployed in three different modes: standalone (using dedicated spectrum), in-band (using resource blocks within a normal LTE carrier), or guard-band (using the unused resource blocks within an LTE carrier’s guard-band). The NB-IoT standard defines power saving features such as power saving mode (PSM) and extended discontinuous reception (eDRX) to extend battery life [42]. NB-IoT implements three extended coverage levels (ECL) (0, 1, and 2) as specified in 3GPP Release 13, where each level determines the number of signal repetitions (up to 2048 in downlink and 128 in uplink) to extend coverage by up to 20 dB beyond standard LTE, achieving a maximum coupling loss (MCL) of 164 dB for reliable communication in challenging environments such as basements and underground locations [44]. ECL dBm thresholds are defined by the telecommunication operator. NB-IoT networks operate on a star network topology where end devices communicate with base stations (eNBs) and base stations connect to a telecommunication operator’s core network which is then connected to specific application servers.

When working with NB-IoT, it is important to understand that many key features, such as power-saving modes (eDRX, PSM) are determined by the mobile network operator rather than the application developer.

The NB-IoT performance metrics as promised by the 3GPP are collected in the Table 1.

2.4. LTE-M

LTE-M also known as LTE Cat-M1 or eMTC, is a cellular LPWAN technology standardized by 3GPP in Release 13 alongside NB-IoT in June 2016 [45]. Comparing the two, LTE-M is distinguished by its higher data rates and full mobility support compared to NB-IoT ultra low power focus. LTE-M provides a complete protocol stack that handles both physical layer transmission and upper network layers through existing LTE infrastructure. The various names—LTE-M, LTE-MTC, eMTC, and Cat-M1/M2—refer to the same family of technologies, with Cat-M1 (3GPP Release 13) and Cat-M2 (3GPP Release 14) indicating the specific versions and capabilities.

LTE-M uses OFDM, SC-FDMA and 16-quadrature amplitude modulation (16-QAM) modulation schemes. LTE-M Cat-M1 uses a 1.08 MHz of bandwidth and supports data rates up to 1 Mbps, operating in licensed LTE spectrum. Same as NB-IoT, LTE-M can be deployed in standalone, in-band or guard-band modes. The standard supports three duplex modes: half-duplex FDD, full-duplex FDD, and TDD operations. The technology also features coverage enhancement through message repetitions (Mode A: mandatory and Mode B: Optional for deeper coverage scenarios), providing up to 15–21 dB additional link budget over standard LTE, and supports advanced power-saving modes like PSM and eDRX, both share the same underlying mechanism, NB-IoT permits much longer eDRX cycles while PSM timer limits are identical, for multi year battery life [42,46]. LTE-M Cat-M2, introduced in 3GPP Release 14, expands bandwidth to 5 MHz and increases peak data rates, making it suitable for more demanding IoT applications. Unlike other LPWAN technologies, LTE-M Cat M2 supports Voice over LTE (VoLTE) capabilities, enabling voice communication directly over the LTE-M network without requiring additional 2 G or 3 G infrastructure [47,48].

The LTE-M performance metrics as promised by the 3GPP are collected in the Table 1.

Table 1. Standards/white-paper targets of LPWAN technologies.

Technology	LoRaWAN	Sigfox	NB-IoT	LTE-M
Range	15–20 km (rural), 2–5 km (urban) [49]	40 km (rural), 10 km (urban) [26]	10 km (rural), 1 km (urban) [50]	10 km (rural), 1 km (urban) [51]
Data Rate	0.3–50 kbps [49]	100–600 bps [26]	50 kbps (DL), 20 (single-tone), 50 (multi-tone) kbps (UL) [52]	1 Mbps (peak) [51]
Bandwidth	125, 250 or 500 kHz [53]	100 Hz [26]	180 kHz [50]	1.4–5 MHz (Cat-M1/M2) [54]
Battery Life	up to 10 years (Class A) [49]	up to 10 years [26]	10 years (200 B UL daily) [55]	10 years (200 B UL daily) [56]
Max Payload	11–242 B [53]	12 B UL/8 B DL [57]	1280 B 1 [58]	1280 B 1 [58]
Carrier Frequency	868/915/433 MHz (ISM) [53]	862–928 MHz (ISM) [26]	Licensed LTE bands [50]	Licensed LTE bands [54]
Latency (end-to-end)	Class A: seconds, Class B: up to 128 s, Class C: near real-time [49]	seconds [26]	1.6–10 s [50]	6.2–14.3 s 2 [56]
Modulation	CSS (LoRa) or FSK [49]	UNB [26]	SC-FDMA UL/OFDM DL [50]	SC-FDMA UL/OFDM DL (+16-QAM) [54]
Security and Privacy	AES-128 [49]	AES-128 [26]	AES-128 [59]	AES-128 [54]

¹ Actual payload may depend on modem, mobile network, and cloud application. Risk of IP fragmentation and thus delivery failure. ² Assuming 164 dB MCL coverage level. Depends on base station antenna configuration.

Table 1. Standards/white-paper targets of LPWAN technologies.

Technology	LoRaWAN	Sigfox	NB-IoT	LTE-M
Range	15–20 km (rural), 2–5 km (urban) [49]	40 km (rural), 10 km (urban) [26]	10 km (rural), 1 km (urban) [50]	10 km (rural), 1 km (urban) [51]
Data Rate	0.3–50 kbps [49]	100–600 bps [26]	50 kbps (DL), 20 (single-tone), 50 (multi-tone) kbps (UL) [52]	1 Mbps (peak) [51]
Bandwidth	125, 250 or 500 kHz [53]	100 Hz [26]	180 kHz [50]	1.4–5 MHz (Cat-M1/M2) [54]
Battery Life	up to 10 years (Class A) [49]	up to 10 years [26]	10 years (200 B UL daily) [55]	10 years (200 B UL daily) [56]
Max Payload	11–242 B [53]	12 B UL/8 B DL [57]	1280 B 1 [58]	1280 B 1 [58]
Carrier Frequency	868/915/433 MHz (ISM) [53]	862–928 MHz (ISM) [26]	Licensed LTE bands [50]	Licensed LTE bands [54]
Latency (end-to-end)	Class A: seconds, Class B: up to 128 s, Class C: near real-time [49]	seconds [26]	1.6–10 s [50]	6.2–14.3 s 2 [56]
Modulation	CSS (LoRa) or FSK [49]	UNB [26]	SC-FDMA UL/OFDM DL [50]	SC-FDMA UL/OFDM DL (+16-QAM) [54]
Security and Privacy	AES-128 [49]	AES-128 [26]	AES-128 [59]	AES-128 [54]

¹ Actual payload may depend on modem, mobile network, and cloud application. Risk of IP fragmentation and thus delivery failure. ² Assuming 164 dB MCL coverage level. Depends on base station antenna configuration.

3. Literature Review

While Section 2 focused mostly on what’s promised by the standards themselves according to their documentation and white papers, this section will examine 20 recent peer-reviewed academic publications about real deployments which use LoRaWAN, SigFox, NB-IoT, or LTE-M to lay groundwork for practical comparison of these technologies.

3.1. Methods

This review targets practical, real world LPWAN deployments published over the past decade (2015–2025). The review scope is deliberately limited to empirical evaluations conducted on live LPWAN deployments, excluding simulation only studies, theoretical analyses, and testbed based research. Literature was identified through Google Scholar searches conducted in July 2025. Search keywords combined LPWAN technologies (LoRaWAN, Sigfox, NB-IoT, LTE-M) with deployment type (real world, practical deployment, field trial) and/or relevant key performance indicators (KPIs)-power consumption, range, data rate, security, scalability, availability, stability. While security was included as a search term and inclusion criterion, most security related literature was found to be based on testbeds, protocol analyses, or simulations rather than real world deployments.

Articles were included if they:

• Report results from operational LPWAN deployment (excluding simulations, testbeds, and protocol analyses) using either or LoRaWAN, Sigfox, NB-IoT, or LTE-M
• Present at least one operational KPI from the following categories:

  –

  Power consumption: battery lifetime or energy consumption measurements from deployed devices

  –

  Range: communication range under operational conditions, including signal strength measurements if reported

  –

  Data rate: throughput achieved in deployed networks

  –

  Scalability: number of devices deployed and packet delivery rate (PDR) achieved at scale

  –

  Availability and stability: geographic, environmental, legal, or commercial limitations of deployments, including frequency regulation constraints and operator infrastructure dependencies. How stable are these available devices: PDR, variance in latency, susceptibility to interference and/or network congestion

  –

  Security: analysis of deployed systems based on operational data

• Were published between 2015 and 2025 in peer reviewed articles or conference proceedings

Titles and abstracts were screened for relevance to the review scope. Full texts of potentially relevant articles were assessed to confirm that they reported results from live deployments meeting the inclusion criteria. A total of 20 articles were selected for detailed analysis. As a review based primarily on Google Scholar, this methodology may not capture all relevant publications and is subject to the limitations of single database searching. However, explicit definition of inclusion criteria and clear delimitation to live deployments aim to maximize reproducibility and clarity regarding the scope of this article.

Table 2. Literature review of LPWAN deployments and KPIs covered.

KPI	LoRaWAN	Sigfox	NB-IoT	LTE-M
Energy eff.	[60,61,62,63,64,65,66]	[62,63,64,67,68,69]	[63,64,65,70,71,72]	[70,71,72]
Range	[60,62,63,65,66,73,74]	[62,63,67]	[63,65,71,75,76]	[71,72]
Data rate	[61,63,65]	[63,67,69]	[63,65,71,72]	[71,72]
Security	[77,78]	[78]	[78,79]	[79]
Scalability	[60,61,66]	[67]
Availability	[61,62,63,65,66,73,74]	[62,64,67,68]	[63,64,65,71,72,75,76]	[71,72]
Stability	[61,63,65,66,73]	[67]	[63,65,70,71,72,75,76]	[70,71,72]

and

Table 3. Study Characteristics and Quality Assessment.

Article	LPWAN	Study Design	Environment	Sample Size
[60]	LoRaWAN	Survey	Various	N/A
[61]	LoRaWAN	Survey	Various	N/A
[74]	LoRaWAN	Field trial	Urban (indoors, outdoors)	Small
[73]	LoRaWAN	Field trial	Urban (indoors, outdoors)	Medium
[66]	LoRaWAN	Field trial	Urban	Large
[77]	LoRaWAN	Laboratory experiment	Laboratory	Small
[62]	LoRaWAN, Sigfox	Field trial	Urban, rural	Medium
[67]	Sigfox	Field trial	Urban, rural	Large
[68]	Sigfox	Field trial	Urban (indoors)	Small
[69]	Sigfox	Laboratory experiment	Laboratory	Small
[70]	NB-IoT, LTE-M	Laboratory experiment	Laboratory	Medium
[71]	NB-IoT, LTE-M	Field trial	Urban (indoors)	Small
[79]	NB-IoT, LTE-M	Laboratory experiment	N/A	Small
[72]	NB-IoT, LTE-M	Laboratory experiment	Laboratory	Small
[75]	NB-IoT	Field trial	Urban (indoors, outdoors, underground)	Medium
[76]	NB-IoT	Field trial	Urban	Small
[63]	LoRa, Sigfox, NB-IoT	Survey	Urban	N/A
[64]	LoRaWAN, Sigfox, NB-IoT	Laboratory experiment, field trial	Laboratory, rural	Medium
[78]	LoRaWAN, Sigfox, NB-IoT	Analytical, proof-of-concept	N/A	Small
[65]	LoRaWAN, NB-IoT	Field trial	Urban	Small

summarizes the reviewed literature.

Sample size was assessed using a multi-dimensional framework considering device count (Small: <10, Medium: 10–100, Large: >100), data point count (Small: <10 k, Medium: 10 k–100 k, Large: >100 k), deployment duration (Small: <1 week, Medium: 1–6 months, Large: >6 months), with the classification determined by the strongest dimension when multiple metrics were available, or marked as N/A for secondary research studies without empirical data collection.

3.2. Power Consumption and Energy Efficiency

LoRaWAN is recognized for energy efficiency in low data rate, uplink heavy IoT use cases, with studies showing the potential for multi year battery life under ideal laboratory conditions [60,63,64]. However, real world deployments often fall short for example Singh et al. [64] observed that practical implementation details can reduce LoRaWAN battery life estimates from years to several months—in their study to 7 months. Energy consumption is highly sensitive to network configuration and conditions: adaptive data rate (ADR) mechanisms can optimize battery life, but only when carefully tailored to the deployment context, otherwise, energy is wasted (Kufakunesu et al. [61]). Field experience also shows that interference, poor radio planning, and backend outages can increase airtime and quickly erode LoRaWAN’s power consumption advantages [62,66,73,74]. Effectively, LoRaWAN’s low power promise is best realized with careful configuration and reliable infrastructure.

Sigfox targets ultra low energy uplinks by using 100 bit/s ultra narrow band bursts and limiting payloads to 12 bytes. In head to head tests it sent 20% more messages than LoRaWAN on the same power budget, showing better energy efficiency even though LoRaWAN achieved higher PDR [62]. Analytical modeling shows a 2400 mAh battery can last 1.5–2.5 years when the device transmits once every 10 min and up to 14.6 years as the reporting interval widens [69]. Field prototypes show that a self-powered smart meter ran 118 days on a 2200 mAh cell (12B payload every 3 h) [68]. Under heavier traffic (5 byte packets every 5 min) Sigfox’s three replica uplink scheme restricts autonomy to roughly 6 months on a 10,000 mAh battery. Less efficient than LoRaWAN and NB-IoT for the same test [64].

NB-IoT’s energy budget is shaped more by idle connected time than by transmit bursts. Vomhoff et al. [70] shows that during idle periods where network reachability is required (connected to one base station), NB-IoT draws less current than LTE-M, extending battery life in continuous monitoring applications. Martinez et al. [65] measured 121 µW mean power for 64 byte hourly transmissions, rising minimally to 143 µW for 512 byte payloads, Martinez et al. [65] predicts 2 to 3 years of battery life (1000 mAH) sending 64 to 512 bytes every 60 min. Singh et al. [64] assumes roughly 7 year battery life with 10,000 mAh cells. Field trials confirm coverage-energy tradeoffs: Ferreira et al. [76] achieved above 96% packet success in challenging environments down to –127 dBm using ECL retransmissions, while Malik et al. [75] observed degraded underground performance beyond 400 m. Field measurements for mobile nodes show that whenever they leave the serving cell: a roaming tracker that crossed an international border needed 43 s to reattach, compared with 8 s when stationary, multiplying radio on time and roughly quintupling energy consumption [80]. 3GPP Release 14 (Cat NB2) addresses this overhead with RRC Connection Re-establishment and full connected/idle mode mobility, letting a UE resume its context in the target cell instead of repeating a full attach and thus cutting signalling and power draw [81,82].

LTE-M’s energy profile favours talkative devices rather than sleepy ones. Vomhoff et al. [70] find that while connected-idle, LTE-M draws noticeably more current than NB-IoT, eroding battery life for sparsely reporting sensors. Boiano’s side by side trials with identical BG96 hardware show LTE-M uplink bursts peak at the same 1.75 W as NB-IoT but average 200 mW higher. Its airtime, however, is 3–5 times shorter, softening the per message cost for bursty uploads [71]. Field measurements by Labdaoui et al. [72] confirm this trade-off: LTE-M is the more energy efficient option when payloads are sent every few minutes, whereas NB-IoT retains an advantage once intervals stretch to hours or days, Labdaoui et al. [72] also estimate that under optimal coverage conditions, the LTE-M node could potentially operate for up to 6.8 years when transmitting once per day (1024 byte payload on 3600 mAH battery).

To enable fair energy efficiency comparison across these four LPWAN technologies despite differences in payload sizes, transmission intervals etc., we calculated energy per message from the estimated battery lifetime in the articles. We excluded studies that were too application specific or lacked sufficient technical detail for normalized comparison. We also assume commonly used 3.6 V battery. For LoRaWAN with a 2400 mAh battery lasting 2 years at 10 min intervals, the energy per message is calculated as: 2400 mAh × 3.6 V × 1000/[(2 × 365 × 24 × 60)/10 min] = 8,640,000 $\mathsf{\mu }$Wh/105,120 which equals 82.2 $\mathsf{\mu }$Wh/msg, mean power of 82.2 $\mathsf{\mu }$Wh/(10/60) = 493.2 $\mathsf{\mu }$W. Same calulcation was done for Sigfox (2400 mAh, 1.5 year battery life with message every 10 min), NB-IoT (1000 mAh, 2 year battery life with message every 60 min) and LTE-M (3600 mAh, 6.8 year battery life with message daily). Normalized results are presented in the

Table 4. Normalized power consumption.

LPWAN	Payload	Interval	Battery	Lifetime	Energy/Message	Mean Power
LoRaWAN	5 B	10 min	2400 mAh	2 years	82.2 $\mathsf{\mu }$Wh	493.2 $\mathsf{\mu }$W
Sigfox	5 B	10 min	2400 mAh	1.5 years	109.6 $\mathsf{\mu }$Wh	657.6 $\mathsf{\mu }$W
NB-IoT	512 B	60 min	1000 mAh	2 years	205.5 $\mathsf{\mu }$Wh	205.5 $\mathsf{\mu }$W
LTE-M	1024 B	24 h	3600 mAh	6.8 years	5221.6 $\mathsf{\mu }$Wh	217.6 $\mathsf{\mu }$W

.

LoRaWAN and Sigfox show lower energy per message but higher mean power, while NB-IoT and LTE-M demonstrates lower mean power despite higher per message energy costs. This illustrates that energy optimization in LPWAN systems reflects fundamental design choices between frequent small payload transmission and infrequent large payload transmissions. Assuming a 3600 mAh 3.6 V battery and the calculated mean power consumption values, the estimated battery life is 3.0 years for LoRaWAN, 2.2 years for Sigfox, 7.2 years for NB-IoT, and 6.8 years for LTE-M.

3.3. Working Range

LoRaWAN’s CSS routinely spans several km in cities and tens of km rurally, which surveys list as a core strength [60]. Controlled 868 MHz trials still achieved above 80% PDR at received signal strength indicator (RSSI) over –110 dBm, achieving 11 km range in rural and 3 km in urban environment [62]. Southampton’s air quality network kept 72.4% PDR across nodes about 2 km from gateways [66], while a 26 day Glasgow study achieved 95.7% PDR outdoors with devices mounted on the rooftops of two buildings 1.9 km and 2.1 km away from the transmitting mote [73]. Careful propagation mapping, as in Girona’s smart parking rollout, lets planners sustain these ranges using low spreading factors and minimal duty cycle overhead [74].

Sigfox demonstrates exceptional long range capabilities using its UNB modulation, achieving practical outdoor range of 10 km in urban environment [62]. The Wild et al. [67] wildlife tracking study documented a impressive transmission distance of 280 km for flying species and 195 km for terrestrial deployments, with flying animals achieving superior connectivity (68.3% PDR) compared to terrestrial species (54.1% PDR).

NB-IoT delivers reliable cellular coverage in diverse environments. Field trials on Quectel BG96 modules showed outdoor connectivity up to 700 m with only 1/20 device outage (RSSI down to −110 dBm, 90% PDR), while indoor performance incurred 3/20 outages and underground links degraded sharply beyond 400 m with 8/20 outages [75]. In harsher settings, NB-IoT sustained above 96% PDR across indoor spaces, two level garages, and a sealed concrete dome up to 1.4 km from the cell, despite RSRP values as low as −127 dBm by using its ECL retransmission mechanism [76]. At extreme low signal (−113 dBm), NB-IoT maintained connections where LTE-M failed [71].

LTE-M provides cellular grade range comparable to standard LTE coverage but with slightly reduced link budget sensitivity compared to NB-IoT. In tests using Quectel BG96, eMTC maintained 0% packet loss at −93 dBm and above, delivering reliable connectivity out to typical cell-edge distances, but it failed to establish links at −113 dBm where NB-IoT still held connections [71]. Labdaoui et al. [72] confirm that, in commercial French networks, LTE-M achieves strong in-building and urban coverage but its effective range shrinks in deep indoor or highly attenuated scenarios compared to NB-IoT’s extended link-budget of up to 20 dB.

3.4. Data Rate

LoRaWAN deployments typically employ SF between SF7 and SF12 trading off data rate for link range. In city scale air quality monitoring, Basford et al. [66] used SF10 (980 bps in EU868), which delivered 72.4% of 135,000 messages. ADR schemes span the full SF7–SF12 range (0.3–5.5 kbps) and can be tuned per deployment to balance throughput and energy [61]. In smart-parking trials, context-specific ADR enabled use of lower SFs (SF8–SF9, 3.1–1.7 kbps) to meet frequent transmission requirements without sacrificing battery life.

Sigfox offers data rates 100 bit/s for uplink and up to 600 bit/s in higher-rate modes, delivering very under 12 byte payloads. Cordero et al. [69] reports that they achieved 100 bps.

NB-IoT can sustain application layer throughputs above 20 kbps under favorable radio conditions and still deliver useful rates deep indoors. Laboratory tests on Orange Belgium’s commercial network recorded peak uplink and downlink throughput of 11 kbps and 17 kbps, respectively, when the modem operated at RSSI levels around −100 dBm [83]. Basu et al. [83] found these figures held steady until RSRP dropped past −115 dBm, after which ECL coverage extension preserved connectivity but reduced user throughput to the low kbps range and extended latency from 300 ms to several seconds.

Labdaoui, Nassim, Fabienne Nouvel, and Stéphane Dutertre conducted a practical throughput measurement of commercial LTE-M modem by u-blox SARA-R422S dual mode modules on live French networks. Their study reports that LTE-M achieves substantially higher data rates than NB-IoT (348 DL and 145 kbps UL) under the same network conditions [72].

3.5. Security and Privacy

LoRaWAN 1.0 secures traffic with the LoRa Alliance’s built in: AES-128 in counter mode for encryption plus AES-CMAC for integrity, using distinct network (NwkSKey) and application (AppSKey) session keys derived during a mutual-authentication join procedure [78].

Sigfox also uses pre shared 128-bit AES keys and a frame based message authentication code, yet its prone to replay and denial of service attacks (DoS) that are hard to mitigate, making it unsuitable for critical use cases [78].

NB-IoT and LTE-M inherit 3GPP’s cellular security/privacy mechanisms: the EPS-AKA authentication protocol and 128-EEA2/AES encryption and integrity, giving strong cryptographic guarantees when correctly deployed. Field tests nonetheless revealed billing abuse, battery drain and device hibernation attacks. Exploiting them demands specialised gear, sophisticated know-how and legitimate network credentials, which limits real world risk [79].

3.6. Scalability

LoRaWAN typically supports about 1000 end devices per gateway before MAC collisions, cross-spreading-factor interference, and the 1% duty-cycle cap push PDR below 90%, a limit observed in survey analyses, long term smart city trials and simulations [60,61,66].

Sigfox constrains each device to 140 uplinks and four downlinks per day [84]. Simulation supported scalability studies indicate that network level PDR begins to decline once device density passes roughly 200 units per km², because the protocol’s three redundant replicas increasingly collide [85].

NB-IoT employs licensed 180 kHz carriers and repetition based ECL. Studies indicate that random-access collisions, rather than user-plane spectrum, become the primary scalability bottleneck, yet simulations predict support for roughly 52,500 devices per cell (3GPP target) and up to 200,000 devices under ideal scheduling assumptions [65,86].

LTE-M trades some density for speed: its 1.4 MHz bandwidth supports roughly 10,000 devices per cell [71,87].

3.7. Network Availability and Stability

Availability describes how often a service can be accessed when and where a device attempts to communicate. Stability refers to the repeatability of that access over time, environmental change, and interference. Important availability metric analyzed in most of the analyzed articles is PDR, for stability variance in latency, server/gateway downtime, susceptibility to interference or network congestion, and security disruptions that can force devices offline.

LoRaWAN’s large link budget lets it reach near perfect indoor PDR (99.95%) and at least 95% outdoors in sparse traffic, yet long term and city scale studies reveal that ALOHA collisions, interference and single server outages can drag average PDR down to 72%, making availability highly context dependent [60,66,73]. Sigfox enjoys predictable scheduling and record 280 km line of sight links, but its PDR swings from 96.7% in clear urban skies to below 19.8% in dense forests, with replay/DoS weaknesses that can further erode stability despite proven 100% PDR in indoor settings [62,67,68,78]. NB-IoT running in licensed spectrum sustains 96–100% PDR even at −127 dBm thanks to ECL, giving it the strongest raw availability. However, live trials also show deep indoor or peak hour congestion can inflate latency from sub second to minutes [65,75,76]. LTE-M trades some link budget for speed: it maintains sub 200 ms latency and perfect PDR at good signal, but drops completely beyond −113 dBm [70,71,72].

3.8. Cost

Costs can be split into two parts—deployment costs and maintenance costs. All of these costs are compiled at the time of writing this article—July 2025. LoRaWAN private networks work in unlicensed spectrum, so upfront cost is dominated by gateways. Eight channel indoor/outdoor units retail from about €120 to €2000, while carrier grade systems can exceed €3000 but cut required site count by 30–50% [88,89].

Sigfox users generally lease capacity from the public operator. No infrastructure is needed beyond the device itself. Where coverage is missing, a plug and play “Access-Station Micro” (typically €600–800) can be installed. The hardware is purchased (or occasionally rented) by the customer or operator, who is then responsible for its upkeep, while the station simply extends the public Sigfox network [90,91].

NB-IoT and (or) LTE-M reuses LTE base stations and licensed spectrum already owned by telecommunication operators, so deployment cost is covered by the operator. Operator needs to explicitly support NB-IoT (or LTE-M), it is not operational by default.

LoRaWAN maintenance costs primarily stem from gateway electricity consumption, periodic firmware updates, network server maintenance (platform plans from €190/month for 500 devices scaling to €3200/month for 49,500 devices [92]), and the time required for technicians to access and service gateway placements. Comparing to other LPWAN technologies—they are rather high. Self hosting LoRa (not LoRaWAN) network servers are reasonable alternative but requires significant engineering work.

Sigfox maintenance is minimal, operators handle infrastructure support, and users pay a yearly subscription (€4.40/device up to 1000 devices excluding VAT). Routine checks and occasional firmware updates (Sigfox does not support upgrading firmware over the air) are the only additional costs [93].

Recurring connectivity fees for NB-IoT and LTE-M are compiled in

Table 5. Recurring connectivity fees for NB-IoT and LTE-M excluding VAT.

Operator	NB-IoT Plan	LTE-M Plan
Vodafone	IoT Easy Connect—1024 MB per month, 10 years, €18 per SIM (€1.8 per year) [94].	Same IoT Easy Connect tariff applies. €1.8 per year [94].
Deutsche Telekom	IoT Business LPWA—6.5 MB per month, 10 years, €13.49 (€1.35 per year) [95].	Under same LPWA pool or IoT Business Classic: from €2.07 per month, 100 MB per month (€24.84 per year) [95].
AT&T	AT&T has stopped selling NB-IoT data plans as of 2024 [31].	$1.50 per month ($18 per year) [96].
Verizon	From $0.75 per device per month, 250 KB data (from $9 per year) [97]	From $1.50 per device per month, 1 MB data (10 MB $6.00, 100 MB $14.00) (from $18 per year) [97]
1NCE	€10 for 10 years (500 MB), €1 per year [98]	Same as for NB-IoT.

.

4. Discussion

Table 6. Field measurements LoRaWAN vs. Sigfox.

	LoRaWAN	Sigfox
Range	Urban: up to 3 km at (80% PDR); Rural: up to 11 km (80% PDR)	Urban: 10 km; Rural: up to 195 km (54.1% PDR). In line-of-sight 280 km (68.3% PDR);
Data rate	SF7–SF12: 0.3–5.5 kbps; SF10: 980 bps.	100 bps uplink; up to 600 bps downlink.
Energy eff.	3 years. 4 82.2 $\mathsf{\mu }$Wh/msg and 493.2 $\mathsf{\mu }$W mean power 1	2.2 years. 4 109.6 $\mathsf{\mu }$Wh/msg and 657.6 $\mathsf{\mu }$W mean power 1
Data rate	SF7–SF12: 0.3–5.5 kbps; SF10: 980 bps.	100 bps uplink; up to 600 bps downlink.
Security	AES-128 CTR & AES-CMAC; distinct NwkSKey & AppSKey for authentification	128-bit AES
Scalability	∼1000 devices/gateway 2	∼200 dev/km2 3; 140 uplinks & 4 downlinks per day per device;
Availability	Global availability. Indoor  99.95%; outdoor under 95% PDR.	Available Europe, Overseas France, Middle East and Africa, Brazil, Canada, Mexico, Puerto Rico, USA, Japan, Latin America, Asia Pacific, South Korea, India, Russia. Urban  96.7%, rural 19.8%; indoor 100% PDR

¹ Based on calculations in Table 4. ² Over 1000 devices, PDR drops under 90% due to ALOHA collisions & duty-cycle limits. ³ 312 devices show average PDR of 56.2%. ⁴ Battery life, assuming 3600 mAh, 3.6 V.

and

Table 7. Field measurements NB-IoT vs. LTE-M.

	NB-IoT	LTE-M
Range	Urban: Up to 1.4 km (>90% PDR). Reachable up to −127 dBm.	Same as NB-IoT. 2 Reachable up to −113 dBm.
Data Rate	11 kbps UL, 17 kbps DL 3	348 kbps DL, 145 kbps UL
Energy eff.	7.2 years. 4 205.5 $\mathsf{\mu }$Wh/msg and 205.5 $\mathsf{\mu }$W mean power 1	6.8 years. 4 5221.6$\mathsf{\mu }$Wh/msg and 217.6 $\mathsf{\mu }$W mean power 1
Security	3GPP EPS-AKA; 128-EEA2/AES encryption & integrity; strong cryptography.	Same as NB-IoT.
Scalability	3GPP target  52,500 devices/cell; up to 200,000 devices under ideal scheduling. 4	Up to 10,000 devices/cell.
Availability	Availability depends on telecommunications operator. 96–100% PDR at –127 dBm, deep indoor and underground coverage & peak hour network congestion increases latency. Better coverage than LTE.	Availability depends on telecommunications operator. 100% PDR at good signal; drops completely beyond −113 dBm & peak hour network congestion increases latency. Same coverage as LTE.

¹ Based on calculations in Table 4. ² Effective range shrinks in highly attenuated scenarios compared to NB-IoT extended link budget of up to 20 dB. ³ Falls to low kbps under bad coverage. ⁴ Battery life, assuming 3600 mAh, 3.6 V.

summarizes the findings. In the cited field trials and laboratory experiments, NB-IoT combines good battery life, deep indoor coverage, high device density support, and robust 3GPP security. Its low per-device subscription cost and lack of private gateway maintenance make it highly cost effective for large scale and long lived deployments. Wherever reliable connectivity, indoors or outdoors, is required, and moderate data rates suffice, NB-IoT should be the technology of choice. Field studies reveal that mobile NB-IoT nodes consume more energy during cell handovers. Although 3GPP Release 14 adds RRC connection re-establishment and full mobility support to lessen this overhead, the studies did not specify which release their networks used, so real world adoption of these energy saving features remains uncertain.

Across the operator networks evaluated in the reviewed studies, LTE-M delivered higher throughput and lower uplink latency than NB-IoT, making it well suited for workloads with frequent or larger bursts—wearable health monitors, real time asset tracking, and responsive smart city nodes (e.g., fill level sensors reporting every few minutes). The same trials reported shorter battery life and reduced deep-indoor resilience relative to NB-IoT.

For organizations requiring complete control of infrastructure and willing to manage gateways, the deployments examined in the LoRaWAN studies showed that privately operated networks can achieve multi-kilometer coverage at low upfront gateway prices (€120–2000 [88,89]). These studies highlighted the importance of careful ADR configuration and reliable infrastructure, especially under higher network loads, to sustain reliable PDR.

Given Sigfox’s limited payload size, low duty cycle, and the fact that the security analyses within the surveyed literature point to replay and DoS weaknesses, reserve Sigfox for cases demanding the longest possible link budget (tens to hundreds of kilometers) and extremely infrequent, small messages, such as wildlife trackers flying across remote regions or very sparse environmental sensors. Avoid Sigfox where security requirements or message volumes exceed minimum.

Sigfox achieves impressive maximum distances in rural/line-of-sight (LoS) conditions but with significantly degraded PDR. LoRaWAN, NB-IoT and LTE-M perform closer to specifications but at the cost of PDR. LTE-M and NB-IoT show the largest gaps between advertised and measured throughput. However, it remains unclear from the analyzed studies whether the tested applications were actually designed to push for maximum data rates. This makes direct comparison with advertised peak rates potentially misleading, the field measurements may represent typical operating conditions rather than maximum achievable performance. Battery life promised is often disconnected from field reality. Actual battery longevity depends on numerous factors-coverage quality, message frequency, payload size, sleep modes. LoRaWAN, Sigfox show particularly severe gaps, with real world deployments achieving 2–4 times shorter lifetimes than promised. NB-IoT, LTE-M perform closer to vendor claims but still fall about 30% short. A notable gap in current research is the evaluation of NB-IoT mobility scenarios in network which use 3GPP Release 14, which introduced connected mode re-establishment procedures to reduce power-expensive idle mode transitions during cell handover. Future LPWAN technology research should explore the reasons behind the gaps between promissed and observed performance, with suggestions on how to better design the LPWAN technologies in order to mitigate this effect.