Sunteți pe pagina 1din 6

An Evaluation Of Low Power Wide Area Network

Technologies For The Internet Of Things


Keith E. Nolan, Wael Guibene, Mark Y. Kelly
Intel Labs Europe, Intel, Leixlip, Dublin, Ireland Email: {keith.nolan,wael.guibene,mark.y.kelly}@intel.com

Abstract—We explore the state of the art in solutions for low reports stating a compound annual growth rate (CAGR)of well
power wide area (LPWA) networks and technologies serving over 20%. However it is also a fragmented market, with a
the Internet of Things (IoT) and Connectivity for Everything diverse range of applications and products. Cisco [3], Analysys
markets. These networks are forecast to capture up to 55%
market share using battery-powered devices operating up to 10 Mason [4], Market Publishers [5], Forrester, McKinsey [6],
years and link distances measured in tens of kilometers. In this BMP Consulting Australia [7], Frost & Sullivan [8], Erics-
paper, we survey two LPWA technologies; ultra-narrow band son [9], [10] ABI Research [11] and Strategy Analytics [12],
solutions by SigFox and the LoRa technology by Semtech. Both have all produced forward-looking reports for this market
technologies operate in the licence-exempt industrial, scientific, sector with a generally unified forecast of rapid growth and
& medical (ISM) bands (EU 868 MHz / US 915 MHz). We survey
both solutions in terms of physical layer (PHY) and associated an ever increasing diversity of IoT applications and services.
medium access control (MAC) capabilities from an end-to-end Analysys Mason forecast that the fastest-growing M2M sectors
system viewpoint. We then proceed to explore coverage ranges in will be utilities and smart grids, healthcare, and security [4]
eastern Ireland. We present results indicating a potential coverage
area of 3, 800 km2 and from a real-world experimental test case B. Characterising IoT
involving the use of SigFox’s technology operating over a 25 km
The IoT has become a term that encompasses a wide range
test link between a 25 mW LPWA client test and a basestation.
Finally, we provide example results demonstrating a received of new applications and services built on objects that can
SNR consistently exceeding 20 dB over this test link distance. compute and communicate. As a result, it has become a term
Index Terms—internet of things, sensors, wireless communica- that has a degree of fuzziness; it cannot be defined in terms
tions of a single technology, communication protocol, form factor,
I. I NTRODUCTION application or service.
Atzori et al. surmise that the reason of today apparent
The Internet of Things (IoT) is considered to be the third
fuzziness around this term is a consequence of the name
wave of information technology after internet and mobile
’Internet of Things’ itself, which syntactically is composed
communications. The Internet of Things concept was formally
of two terms. The first one pushes towards a network oriented
proposed in 2005 by the International Telecommunication
vision of IoT, while the second one moves the focus on generic
Union (ITU) where they published ’ITU Internet Report 2005:
objects to be integrated into a common framework [13].
Internet of Things’ [1]. It was noted in this report that all
Perara et al. outline five characteristics of the IoT [14].
objects can exchange information via the networks actively.
These are:
The IoT is included by the US National Intelligence Council
• Intelligence - data generation and transformation into
(NIC) in the list of six Disruptive Civil Technologies with
potential impacts on US national power [2]. NIC foresees that knowledge
• Architecture - heterogeneous event and time-driven in-
by 2025 Internet nodes may reside in everyday things e.g. food
packages, furniture, paper documents, and more. It highlights frastructure
• Complex system - granular system of objects each with
future opportunities that will arise, starting from the idea that
popular demand combined with technology advances could different levels of resources
• Size considerations - it is predicted that there will be 50-
drive widespread diffusion of an Internet of Things that could,
like the present Internet, contribute invaluably to economic 100 billion devices connected to the internet by 2020 [15]
• Time considerations - handle billions of parallel and
development. The possible threats deriving from a widespread
adoption of such a technology are also stressed. Indeed, it is simultaneous events
• Space considerations - objects can be deployed over wide
emphasized that to the extent that everyday objects become
information security risks, the IoT could distribute those risks geographical areas
• Everything-as-a-service - efficient, scalable, versatile ap-
far more widely than the Internet has to date.
plications and services
A. Background
IoT is underpinned by an infrastructure comprising poten- C. Wireless Connectivity
tially high volumes of connected devices. Machine commu- From a technical perspective, IoT now leverages a wide
nications is a very large market that is growing rapidly with range of wired and wireless communications technologies and

978-1-5090-0304-4/16/$31.00 ©2016 IEEE 439


protocols combined with application/service requirements that ISM band. Our studies center on SigFox’s ultra-narrow band
range from low bandwidth & high latency to high band- (UNB) technology and LoRa developed by Semtech.
width/high frequency acquisition. In addition, IoT emcom-
A. LoRa based LPWAN
passes nodes that operate over short range e.g. centimeters,
to long range e.g. 30+ km. LoRa is a scheme proprietary to Semtech. It is a chirp spread
Wireline technology examples include Ethernet, optical fi- spectrum (CSS) type modulation, allowing multiple data rates
bre, Building Automation and Control Networks (BACnet), as both bandwidth and spreading factors are configurable.
Local Operating Network (LONWorks), X10 protocol for LoRa products offer a low power consumption ICs with a
home automation, consumer electronics bus (CEBus). Cellular 10 µA consumption (25 mA TX, 10 mA RX duty-cycled)
communications technologies include General Packet Radio The medium access control (MAC) layer (LoRaWAN) lay-
Service (GPRS), Enhanced Data Rates for GSM Evolution ing above this proprietary modulation is open source and
(EDGE), High Speed Packet Access (HSPA), Long Term is defined by the LoRa Alliance consortium. The so-called
Evolution (LTE), LTE+. LoRaWAN is a LPWAN offering a fully bidirectional sym-
Mesh networking technologies include IEEE 802.15.4 Zig- metrical link between the endpoint and the gateway. The
bee, ZWave, Flexnet, Wireless Highway Addressable Remote overall network architecture is described in Fig. 1 where
Transducer Protocol (HART). endpoints communicate with basestations called gateways or
Low-power wide-area technologies include SigFox, LoRA, concentrators. Data are then conveyed to a network server
Adaptrum, WEIGHTLESS, IEEE 802.22, and IEEE 802.15.4g before connecting to end-user applications and servers.
Powerline carrier technologies include PRIME and G3, and
wireless local/metropolitan area network technologies include
WiFi and WiMAX.
Sub-1 GHz spectrum segments have been proven popular
for new wireless communications services since the transition
from analogue to digital television and subsequent band clear-
ance. Unused available spectrum segments in the UHF/VHF
bands, known as ’television white space’ (TVWS) have been
the subject of over a decade of research. Early theoretical
work involving TVWS has created the foundation for the
development of real-world tests and trials. These are needed to
better understand how the techniques and systems developed
over the past decade could play a useful role in deployed
systems [16], [17]. Somewhat similar trials being carried out
involving TVWS and in some cases, M2M, are currently
underway wordwide. Fig. 1. LoRaWAN E2E Network Architecture
IoT deployments can leverage a combination of commu-
nications protocols e.g. Message Queuing Telemetry Trans- Fig. 1) depicts a LoRa wireless AES-128 encrypted link
port (MQTT) [18]–[20], Constrained Application Protocol from the endpoint to the gateway (GW) and the network server.
(CoAP) [21], [22] and Thread [23]. The GW is then connected to the internet via a 3G/4G/Ethernet
In summary, IoT infrastructure now comprises: link to the core network and packets are routed in the cloud
• A plethora of communications technologies and protocols to Network Servers and Application/Client Servers.
• A myriad of potential applications and services LoRaWAN is primarily intended for IoT devices operating
• Volumes of connected IoT devices on a scale unlike up to ten years on battery power alone in regional, national
what we have experienced during previous waves of or global deployments [25]. The intention of LoRaWAN is
information technology. to provide secure bi-directional communication, mobility and
Vision Mobile predict that the top IoT platforms in 2020 GPS-free localization services. LoRaWAN supports payload
will be cross-vertical. Sector-specific platforms will be niche sizes for bidirectional transmissions ranging from 19 bytes
or in decline [24]. Therefore to progress to a cross-vertical IoT to 250 bytes in compliance with ETSI 169/433/868 MHz
approach, a platform that can support the development and and FCC 915 MHz ISM band regulations. The LoRaWAN
deployment of new IoT applications and services and reduce protocol overhead is 12 bytes. Efforts via the LoRa Alliance
the commercialisation complexity yet support the five key IoT are underway in porting the MAC layer to China’s 470 MHz
characteristics [14] is required. ISM band. LoRaWAN defines three types of classes operating
in the network and two network types. The first network type
II. S URVEY OF T WO LPWA T ECHNOLOGIES is the ”LoRaWAN Private Network”, which can be operated
Our focus now shifts to the use of LPWA in an IoT context. by private individuals (e.g for asset tracking, private wireless
We explore the use of sub-1 GHz low power wide area wireless sensor networks, and home automation), or companies in
communications technologies; specifically, the 868/915 MHz an industrial context for I-IoT (Industrial IoT) and Industry

440
4.0. The second type of network is the ”LoRaWAN Public SigFox network offer different engagement models based on
Network” where mobile network operators own the network the volume of devices and number of messages transmitted
infrastructure and offer a network as a server/infrastructure per day:
as a service market offering to their clients. On these two • Platinum : 101 to 140 messages + 4 downlink
type of networks, three classes of devices can co-exist. ”Class • Gold : 51 to 100 messages + 2 downlink
A” is a device-initiated scheme, where the device transmits • Silver : 3 to 50 messages + 1 downlink
its payload when it needs to and then offers two reception • One : 1 to 2 messages + no downlink
windows. The device receive window length ranges from five In ITU Region 1, operation in the 868 MHz band is subject
to eight symbols to minimise energy usage. The second class, to duty cycle limitations; in this case, the maximum duty cycle
”Class B”, is where the network offers the possibility to ping is 10%, which accounts for the maximum of 140 messages/day
the device at regular slots and provides support for fallback for Sigfox’s Platinum offering. The downlink mode only
to ”Class A” in case of battery constraint operation models. works as an acknowledgement to an uplink message. The
The third class of device is denoted ”Class C”. In this class, Sigfox device transmits a message using three frequencies.
the device acts much like in ”Class A” except that when not Following message transmission on the first frequency, the
transmitting it is in continuous receive mode. device switches to receive mode in order to attempt to receive
B. SigFox based LPWAN a downlink message from the base station. The maximum
downlink receive time-window is 25 seconds. A new uplink
SigFox created an ultra-narrowband IoT communications message cannot be dispatched within six seconds following
system designed to support IoT deployments over long ranges receipt of a downlink message.
e.g. in excess of 20 km between a client device and a base We present a table listing the features of SigFox and LoRa in
station [26]. This company has adopted an operator model; Table I extracted from specification datasheets: [27]–[29]. Key
essentially creating a cellular network for IoT devices. Ultra- differences include that LoRa was designed from the outset
narrow band operation is achieved using channel bandwidths for both uplink/downlink data whereas SigFox was designed
lower than 1 kHz that transport data payloads of 12 bytes predominately for downlink only and an uplink option is
uplink and 8 bytes dowlink with a protocol overhead of available by leveraging a downlink acknowledgement message
26 bytes. SigFox have targeted licence-exempt spectrum for to end nodes. This limited uplink functionality supports the
their product, namely the 915 MHz band in the U.S. and transmission of up to four 8-byte messages to each device per
868 MHz band in Europe. SigFox’s model is a cloud-based day. LoRA offers a wider variety of payload sizes e.g. from 19
approach where data are passed to the backend server and to 250 bytes whereas SigFox’s uplink payload size is limited to
customer portal directly; users must then implement callbacks 12 bytes. In addition, SigFox’s technology is proprietary and
to route the received data to their own systems. LoRaWAN is being further developed via the open, non-profit
LoRa Alliance initiative.

TABLE I
L O R A AND S IG F OX F EATURE TABLE

LPWA Features LoRa (Semtech SX1272) SigFox (AXSEM)


Symmetrical Technology Y Y
Uplink Data Data
Downlink Data + ACK ACK
Payload size 19-250 bytes 12 bytes
Protocol Overhead 12 bytes 26 bytes
TX power 13dBm 14dBm
TX consumption 28mA 45mA
RX consumption 10.5 mA 10 mA
Fig. 2. SigFox E2E Network Architecture Encryption AES-128 E2E AES-128
Open Standard Y N
SigFox relies on several module makers for its endpoints. In Technology CSS/FSK/OOK/GMSK GFSK/BPSK
Sensitivity (dBm) -137 -129
this paper, figures are given for the AXEM AX-Sigfox SoC,
for example for the AX-SigFox SoC, the power consumption
given in the datasheet is: 45 mA TX current and 10 mA RX
current. The SigFox network operates different uplink and III. E XPERIMENTAL A NALYSES
downlink modulation schemes. For downlink, SigFox use a We conducted an experimental evaluation of outdoor range
GFSK scheme operating at 500 bps on a 600 Hz spectrum test estimation for LPWA using the parameters detailed in
segment and BPSK using 100 bps on a 100 Hz channel Table II.
for uplink. Each uplink message is transmitted three times in Fig. 3, we present a coverage map model for the example
on different frequencies to help ensure it will be delivered. case where one LPWA basestation is located on a high site

441
TABLE II
LPWA COVERAGE ESTIMATION

Parameter Value
TX power 14 dBm
Frequency 868.1 MHz
RX gateway antenna gain 6 dBd
RX antenna polarisation Vertical
RX antenna polar pattern omnidirectional
TX device height above ground 2.5 m
RX sensitivity -138 dB to -123 dB
RX antenna height above sea level 470 m
Terrain model buildings/trees/ground cover incl.

on the eastern seaboard of Ireland. The location, Three Rock


mountain is 450 m above sea level and overlooks the city of
Dublin in Rep. of Ireland. The total coverage area depicted in
this figure is approximately 3, 800km2 . The distance from the
eastern coast of Ireland to the western fringes of the combined Fig. 3. Estimated coverage area using one LPWA basestation on the eastern
seaboard of Ireland [extracted from an overlay image using Google Earth]
coverage area is approximately 90 km. In Fig. 4, we depict the
estimated land area where at least 90% coverage is available
using the single LPWA site; this corresponds to 1380 km2 or
approximately 36% of the total area shown in Fig 3. We can
conclude from this study that potentially wide swathes of land
can be covered by a small number of LPWA basestations.
In addition to our coverage estimation study, our tests
also involved dispatching LPWA signals from a client device
situated in Leixlip, Co. Kildare, Ireland to a basestation 25 km
away and located 757 m above sea level as depicted in Fig. 6.
The experimental setup is depicted in Fig. 5 where a Sigfox
endpoint dispatched payloads to a basestation located 770 m
above sea level at a point-to-point distance of 25 km. Data
from this endpoint are retrieved using Sigfox’s backend web
server. The estimated coverage area, determined using the
same parameters as detailed in Table II with a modification
to account for the antenna height difference (770 m) is shown
in Fig. 7. Fig. 4. Estimated core coverage area where appromimately ¿90% coverage
The received client device transmissions are shown via is available [extracted from an overlay image using Google Earth]
the power spectral density plot presented in Fig. 8. The
received LPWA signals are displayed on the right hand side
operate in the ISM bands (EU 868 MHz / US 915 MHz). Both
of the power spectral density waterfall plot in this case. The
solutions were surveyed in terms of PHY and associated MAC
minimum noise floor was approximately -140 dBm in this
capabilities from an end-to-end system viewpoint. We then
experiment.
proceeded to explore a coverage estimation test case where we
A table of example results regarding the performance of the
found that one LPWA basestation installed 470 m above sea
link is presented in Table III. From this table, we note that the
level on Three Rock mountain in Dublin can potentially serve
received SNR exceeded 20 dB and that the variations in both
a core coverage area of 1380 km2 within a total estimated
SNR and received signal strength indication (RSSI) were low.
area of over 3, 800 km2 . We also conducted a real-world
IV. C ONCLUSIONS range evaluation involving the use of SigFox’s technology
demonstrating successful 25 km test link operation between
We explored the state of the art in industrial solutions a 14 dBm LPWA client test and a basestation. We provided
for LPWA networks and technologies. These networks are example results demonstrating a received SNR consistently
forecasted to capture up to 55% in share of market (SOM) exceeding 20 dB over this test link distance.
with ten year battery powered devices and ranges of tens of
kilometers. In this paper we survey two players in the LPWA ACKNOWLEDGEMENTS
market; ultra-narrow band solutions by SigFox and the LoRa We wish to thank 2RN for the help and guidance regarding
spread spectrum technology by Semtech. Both techniques the establishment of the test network used for the experimental

442
Fig. 8. Power spectral density and waterfall plot at a high site basestation showing received LPWA signals originating from our test location 25 km from the
basestation

TABLE III
E XAMPLE STATISTICS REGARDING THE RECEIVED LPWA FRAMES OVER A 25 KM LINK DISTANCE

Time Delay (s) Basestation RSSI (dBm) SNR (dB) Freq (MHz)
09-07-15 13:05:50 1.4 -114.50 23.99 868.2080
09-07-15 13:05:35 1.4 -115.50 23.68 868.2119
09-07-15 13:05:21 1.4 -114.50 21.91 868.1949
09-07-15 13:05:06 1.4 -114.50 23.57 868.1985

Fig. 5. Experimental test set-up. A Sigfox endpoint dispatches payloads to a


basestation located 770 m above sea level at a distance of 25 km. Data from
this endpoint are retrieved using Sigfox’s backend web server

Fig. 7. Estimated coverage area for the high site (Kippure mountain) chosen
for the real-world range evaluation [extracted from an overlay image using
Google Earth]

analyses and development of the coverage range estimation


data

R EFERENCES

[1] ITU, “The internet of things,” Tech. Rep., 2005.


[2] T. N. I. COUNCIL, N. NIC, and S. C. B. Intelligence, “Disruptive
Fig. 6. Map showing distance between a LPWA test site at Leixlip, Ireland
Civil Technologies Six Technologies With Potential Impacts
and a basestation on a high site in Co. Wicklow over a 25 km point to point
on US Interests Out to 2025,” in National Intelligence
distance
Council, no. April, 2008, p. 48. [Online]. Available:
http://scholar.google.com/scholar?hl=en\&btnG=Search\&q=intitle:
Disruptive+Civil+Technologies+Six+Technologies+With+Potential+
Impacts+on+\{US\}+Interests+Out+to+2025\#1

443
[3] T. Cisco, “Cisco Visual Networking Index : Global Internet of Things, 2010. [Online]. Available: http://link.springer.
Mobile Data Traffic Forecast Update , 2010 2015,” com/chapter/10.1007/978-1-4419-1674-7\ 38$\backslash$npapers3:
Growth Lakeland, vol. 2011, no. 4, pp. 2010–2015, 2011. //publication/uuid/6B27B463-9681-4606-8D82-FFF4BED83FEC
[Online]. Available: http://www.cisco.com/en/US/solutions/collateral/ [16] P. Pawelczak, K. Nolan, L. Doyle, S. Oh, and D. Cabric, “Cognitive
ns341/ns525/ns537/ns705/ns827/white\ paper\ c11-520862.html radio: Ten years of experimentation and development,” IEEE Commu-
[4] A. Mason, “Imagine an M2M world with 2.1 billion connected things nications Magazine, vol. 49, no. 3, pp. 90–100, 2011.
?” Analysys Mason, Tech. Rep., 2011. [Online]. Available: http://www. [17] Keith, “The Evolution of TV White Spaces,” Connected World
analysysmason.com/about-us/news/insight/M2M\ forecast\ Jan2011/ Magazine, vol. May/June 2, pp. –, 2012. [Online]. Avail-
[5] M. Publishers, “Smart Grid Utility Data Market,” Market Publishers, able: http://connectedworldmag.com/10\ 2\ magazinearticle.aspx?id=
Tech. Rep., 2010. [Online]. Available: http://pdf.marketpublishers.com/ MAZ0120327101251680
sbi/smart\ grid\ utility\ data\ market.pdf [18] A. Stanford-Clark and A. Nipper, “MQ Telemetry Transport,” 2013.
[6] McKinsey&Company, “McKinsey on Smart Grid,” [Online]. Available: http://mqtt.org/
McKinsey, Tech. Rep., 2010. [Online]. Avail- [19] N. O’Leary, “Paho - Open Source messaging for M2M,” 2014. [Online].
able: http://www.mckinsey.com/Client\ Service/Electric\ Power\ Available: http://www.eclipse.org/paho/
and\ Natural\ Gas/Latest\ thinking/McKinsey\ on\ Smart\ Grid [20] U. Hunkeler, H. L. Truong, and A. Stanford-Clark, “MQTT-S
[7] B. M. P. T. Consultants, “Smart grid architecture-telecommunications A publish/subscribe protocol for Wireless Sensor Networks,” 2008
technologies for enabling looking-forward and viable smart grid appli- 3rd International Conference on Communication Systems Software
cations,” BMP Telecommunications Consultants, Tech. Rep., 2010. and Middleware and Workshops (COMSWARE ’08), pp. 791–
[8] F. . Sullivan, “Machine to machine communications - This is going to 798, 2008. [Online]. Available: http://ieeexplore.ieee.org/lpdocs/epic03/
be BIG!, Statecast Perspectives & Insight for Executives, SPIE 2010,” wrapper.htm?arnumber=4554519
Frost & Sullivan, Tech. Rep., 2010. [21] Z. Shelby, D. Sturek, and B. Frank, “Constrained Application Protocol
[9] Ericsson, “Device Connectivity Unlocks Value (White Paper),” Ericsson, (CoAP),” orgiddraftietfcorecoap01 txt 0807, pp. 1–81, 2010. [Online].
Tech. Rep., 2011. Available: http://tools.ietf.org/html/draft-ietf-core-coap-08
[10] L. Ericsson, “More than 50 Billion Connected [22] C. Bormann, A. P. Castellani, and Z. Shelby, “CoAP: An application
Devices,” www.ericsson.com/res/docs/whitepapers/wp-50- protocol for billions of tiny internet nodes,” IEEE Internet Computing,
billions.pdf, no. February, 2011. [Online]. Available: vol. 16, no. 2, pp. 62–67, 2012.
http://scholar.google.com/scholar?hl=en\&btnG=Search\&q=intitle: [23] Google, “Thread - Google.” [Online]. Available: http://www.
more+than+50+billion+connected+devices\#0$\backslash$nhttp: threadgroup.org/about.aspx
//scholar.google.com/scholar?hl=en\&btnG=Search\&q=intitle: [24] V. Mobile, “IoT megatrends 2015.” [On-
More+than+50+Billion+Connected+Devices\#0 line]. Available: http://emma.intel.com:8089/Documents/
[11] A. B. I. Research, “Cellular M2M Connectivity Services - The Market 15-04-08VISIONMOBILEIoT-Megatrends-2015-VisionMobile-Topp(1)
Opportunity for Mobile Operators, MVNOs and other Connectivity .pdf
Service Providers 2010,” ABI Research, Tech. Rep., 2010. [25] “LoRa Technology.” [Online]. Available: http://lora-alliance.org/
[12] S. Analytics, “A Brave new World in Mobile Machine to Machine What-Is-LoRa/Technology
(M2M) Communications,” Strategy Analytics, Tech. Rep., 2008. [26] SigFox, “About SIGFOX.” [Online]. Available: http://www.sigfox.com/
[13] L. Atzori, A. Iera, and G. Morabito, “The Internet of Things: A survey,” en/#!/about
Computer Networks, vol. 54, no. 15, pp. 2787–2805, 2010. [27] “SigFox-specs-Atmel.” [Online]. Available: http://www.atmel.com/
[14] C. Perera, A. Zaslavsky, P. Christen, and D. Georgakopoulos, “Context images/atmel-9372-smart-rf-ata8520 datasheet.pdf
aware computing for the internet of things: A survey,” IEEE Communi- [28] “Semtech-SX1272.” [Online]. Available: http://www.semtech.com/
cations Surveys and Tutorials, vol. 16, no. 1, pp. 414–454, 2014. images/datasheet/sx1272.pdf
[15] H. Sundmaeker, P. Guillemin, and P. Friess, “Vision and [29] “AX-Sigfox.” [Online]. Available: http://www.axsem.com/www/sigfox
challenges for realising the Internet of Things,” . . . the

444

S-ar putea să vă placă și