S7-1

Challenges and Opportunities of Internet of Things

Yen-Kuang Chen

Intel Labs
Intel Corporation
Santa Clara, CA, USA

E-mail: [email protected]
and nuclear disaster in Japan, robots were used to explore the
damaged nuclear plants due to radiation fears [5]. Third,
wireless communication and broadband internet technologies
Abstract - To date, most Internet applications focus on providing enhance our communication capabilities, a power
information, interaction, and entertainment for humans. that will only grow when 4G wireless [42] and greater internet
However, with the widespread deployment of networked, bandwidth become available. Fourth, emerging cloud
intelligent sensor technologies, an Internet of Things (IoT) is
computing and machine intelligence will enhance our
steadily evolving, much like the Internet decades ago. In the
analytical capabilities. With massive computation capabilities
future, hundreds of billions of smart sensors and devices will
interact with one another without human intervention, on a and more mature machine learning techniques, we may begin
Machine-to-Machine (M2M) basis. They will generate an analyzing information that could not be processed in the past.
enormous amount of data at an unprecedented scale and Further, humans keep demanding technologies to improve
resolution, providing humans with information and control of their lives. Fundamentally, people want to be happier. How can
events and objects even in remote physical environments. The people be happier? First, humans want more time, money, and
scale of the M2M Internet will be several orders of magnitude quality of life. Technologies can help humans save money,
larger than the existing Internet, posing serious research enhance their appearance, eat better, and sleep better. Second,
challenges. This paper will provide an overview of challenges and most people want to avoid being in threatening or troublesome
opportunities presented by this new paradigm. situations. Technologies such as weather forecasting or fire
warning systems help predict future events. Third, they want to
be healthier. Technologies help humans to provide better and
I. Introduction safer environments for healing, to take care of aging and sick
people, and to avoid accident and injury. Fourth, most people
Technologies change our life. Out of many emerging desire companionship. Email, internet phones and social
technologies, Internet of Things (IoT), also known as networks connect people. Finally, people want to feel special or
machine-to-machine (M2M) (where smart devices that collect be respected. Personalized services and products have become
data, relay information to one another, process the information very popular.
collaboratively, and take action automatically) is a new Many of the above human desires can be further fulfilled by
paradigm offering both challenges and opportunities. emerging M2M and related technologies [12][32][36]. For the
Over last twenty years innovative information technologies same example given earlier, imagine that you were to give a
have wrought significant change in human civilization. For speech in another city and you got stuck on the highway. With
example, imagine that you were to give a speech in another the improvement in M2M technology, your calendar and your
city which is normally one hour away from where you live. car can communicate. If you are expected to be late, your
Unfortunately, there was a traffic jam on the highway and you smart phone will send a message to your audience
were expected to be late. If this happened 20 years ago, there automatically telling them approximately how much time they
was literally no way you can communicate the news to your have to wait. Or even better, the calendar planner can look up
audience. Nowadays, if you get stuck on the highway, you can the traffic condition in advance and suggests what time you
simply use your cell phone to tell your audience. This is should leave. Sensors can monitor the traffic conditions along
because cell phones and telecommunication service are the routes to your destination so that you are able to select the
affordable and available to almost everyone. Question: can best route to get to the venue on time. There are many similar
technologies do better? examples. The nexus of human needs and emerging computing
More game-changing capabilities are emerging. First, new and sensor technologies is bringing about a new digital
types of sensors enhance our perceptual abilities by detecting revolution.
information that humans cannot detect and collecting such Section II will give an overview of the new paradigm.
information anytime and anywhere. Second, robots similarly Section III will describe the challenges of the new paradigm.
enhance our ability to act, overcoming our physical Section IV will discuss the opportunities created by these
limitations. They can be given greater-than-human strength challenges.
and may go where humans cannot. After the recent earthquake
 IEEE
383

978-1-4673-0772-7/12/$31.00 ©2012
S7-1

Personal Enterprise Government

Personal nterprise Government

Assisted living drug tracing and authentication i

Smart home Smart
Intelligent building
Economical agricultu
breeding Supply cha
route planning automation
Fleet management Chronic d
Factory automation managementh
Good/pr

Traffic management
Smart metering and Efficient
billing ilii & Efficient e

ti & ti

Ff
Factory safety

m Building safety

Save money
Vehicle and time

safety V hi l f

Avoid

Health monitoring Natural disaster warning Infrastructure

 monitoring
Homeland security
unhappiness

Unmanned defense

Smart entertainment
Environmental
conservation
Stay
Healthy
Intelligent signage &
Feel special shopping recommendation lli i & h
ili

Figure 1: M2M application examples. Different rows reflect different human wishes. Different columns reflect who pay for the
services/applications
understand the potential risk of running into each
II. New Paradigm: Connected Context Computing other; in particular, sensors and inter-vehicle
communication can help us see what we cannot see.
Recently, sensor networks, cyber physical systems, and With timely and proper warning, we can reduce the
internet of things have become more common as sensing, vehicle collision rate.
communication, and analytics technologies have matured. In 4. Assisted living: The population is aging. There is an
the future, digital sensing, communication, and processing increasing need to take care of more elders. Sensors
capabilities will be ubiquitously embedded into everyday can help us monitor the health condition of elders and
objects, turning them into the Internet of Things (IoT, or properly provide help (e.g., reminders of missing a
machine-to-machine, M2M). In this new paradigm, smart dose, warning of high blood pressure, requesting
devices will collect data, relay the information or context to medical emergency).
each another, and process the information collaboratively In short, connected embedded sensors help humans
using cloud computing and similar technologies. Finally, either “hear/see” things that they could not hear or see in the past and
humans will be prompted to take action, or the machines do something that we could not do in the past. This paradigm
themselves will act automatically. shift creates numerous challenges and opportunities for
The world is on the edge of revolutions in both quantity and engineering.
quality. Today more things than humans are connected to the
internet [9]. In the near future, the number of connected
devices will be tens or hundreds of times larger than the III. Challenges
number of connected people. It is estimated that 20 households
in 2012 years will generate more internet traffic than the entire M2M can be broken down into four major layers as shown
internet in 2008 [10]. Furthermore, the computation in Figure 2. Sensors collect data, communication units relay
capabilities per device are also increasing. For example, the information collected, computing units analyze the
today’s smart phone is about 100,000 times faster than the information, and service layers take action.
ENIAC I computer, built in the 1940s [4]. In the future, enormous numbers of sensors will be
There are many applications of this new paradigm deployed. The costs of servicing such sensors will be a major
[14][16][17], as shown in Figure 1. Some examples are concern. Hence, one challenge is sensor technology that
highlighted below: requires minimal or even zero effort to deploy and maintain.
1. Smart home: At home, embedded sensors can According to [3], many domestic ubiquitous computing
understand the human activities and properly adjust projects have failed because the complexity of sensor
the air temperatures or lighting to reduce our energy deployment. Additionally, one important sensor service cost is
usage without sacrificing human comforts. battery replacement. It is often almost impossible to replace
2. Economical agriculture: In a farm field, remote bug sensor batteries once they are in the field. Therefore, another
traps can detect the outbreak of pests and initiate challenge is low power sensor design, or designs which do not
spreading the right amount of pesticide. This will require a battery change over the lifetime of the sensor. For
reduce the chance of overspreading and potential example, if a sensor is deployed on an animal for tracking
damage to the crops. purposes, the battery of the sensor
3. Vehicle safety: Sensors on a car can help drivers

384

• Answers are computed ahead of the questions

• Optimum distribution of device & cloud intelligence

• Machines work for people frictionlessly & robustly Computation • Zero effort to connect large, dense
• Standard interface to foster innovation in the populations of
Wisdom
ecosystem Service
 stationary and moving devices with high energy
efficiency • Complete data security and privacy
Knowledge
• Low-power so that no need to change battery
Sensors Data
S7-1

Useful service Figure 3: Refinement process of

transforming sea of raw data

Information

Communication

• “Zero-touch” to deploy and manage devices Sensors

power and bandwidth to communicate the data to the server.
As M2M devices become more powerful, intelligent
Figure 2: Major components in M2M and their challenges computation must be distributed across both the devices and
the cloud.
should outlive the animal. Finally, after understanding the contexts, machines should
After the sensors collect the data, the next step is to either take proper action or prompt humans for proper action.
communicate the information collected. Even today the Ideally, machines should work for people. Today, when using
number of devices connected to the internet exceeds the search engine, a significant portion of the work is done by the
number of humans; in the future this gap will only increase. users instead of the machines. The machines give us a list of
Many of the sensors will be connected wirelessly through possible results and ask users to refine the search results. As the
systems like Bluetooth, WiFi, or 3G/4G cellular networks. users search through the list of results, the search engine uses
Connecting the growing number of devices is a huge the results to refine future search results. Users actually work
challenge. Most base stations are designed to provide a certain for the machines in this case. Ultimately, we want the machines
quality of service up to a given number of users. When there to work for users, not vice versa. Furthermore, to build a
are too many simultaneous users, some users will not receive successful M2M ecosystem, it is crucial to have unified
service. Since the number of devices will be orders of standards for everyone to follow. However, major standards are
magnitude larger than the number of human users, this still under development [1][7][13][18]-[29][38][41] and many
problem will become even more serious. Too, as the number emerging applications are using their own standards. For
of devices connected to the internet grows, so do security and example, EPC Global and Ubiquitous ID are two different,
privacy issues [6]. non-compatible ways of identifying objects [37]. Existing
Connected devices (sensors) can produce oceans of data. solutions are highly fragmented and technology is typically
According to Cisco, the number of objects on the internet dedicated to a single application. Since the solutions are being
exceeded the number of humans in 2008 or 2009 [9], a trend designed from scratch, such projects take longer than
that accelerates every year. Thus, in the future, the amount of conventional IT projects [11]. Diverse technical solutions and
data generated by machines will be orders of magnitude standards are slowing the development of the global M2M
greater than that generated by humans. However, we need market [39].
layers of intelligence to transform this data into wisdom
(Figure 3). In this new computing era, analysis of data and its
context will play a key role [2]. IV. Opportunities
Sensor generated data are different from human generated
data: While there are many challenges, they are also many
1. There are often real-time requirements for data research and development opportunities. The following
processing, e.g., natural disaster warning. Stream data subsections will discuss these opportunities.
processing or mining is a critical component in the
analysis. A. Low-power wireless sensors
2. There is often a huge amount of temporal and spatial
redundancy in the information. It will be more One of the challenges is to design low-power sensors which
efficient if the analytical algorithm can take advantage do not need battery replacement over their lifetimes. This
of such redundancy. However, synchronization of data creates a demand for energy-efficient designs. A typical
from different sensors may be inaccurate. wireless sensor has 4 major components: a sensing unit, a
3. Reliability or accuracy of the data may be unpredictable. processing unit, a transmitting/receiving unit, and a power unit,
Separating in reliable from unreliable signals will be a as shown in Figure 4.
critical component of the analysis. One of the opportunities is to design low-power sensing
4. The goal for the sensor data processing is either machine unit. Highly accurate sensor modules often consume great
or human action, meaning that the requirement for amounts of power. One alternative is to use an array of
data processing is more stringent. low-accuracy modules with lower power consumption, and
Generating accurate and timely answers is a tremendous then use data fusion to create high-accuracy information.
challenge. Further, many analytic approaches assume that all Another opportunity is to design a new video processing
of the data is present on the server. However, it takes both
into information, knowledge, and finally useful wisdom
 385
S7-1 or remove redundancy of the information,
e.g., via distributed coding, to further
reduce bandwidth usage.
Second, many connected devices are
mobile, such as sensors install on vehicles.
Sensing unit
These sensors not only need to
communicate with other sensors via
intra-vehicle networks, but also
Power unit inter-vehicle networks. Existing radios
Processing unit
nearby cluster head, which in turn pass on most likely underperform in on-road
to the base station. This will reduce the wireless channels. To provide reliable
machine transmission power demand and vehicle-to-vehicle communication, we
Communication
increase spatial reuse of the spectrum. should first
unit Another possibility is to fuse information
Figure 4: Schematic of a typical wireless sensor study the vehicle mobility model and then develop the optimal
communication protocols based on the channel model. Finally,
and encoding algorithm. Traditionally, video encoders have wireless communication consumes large amounts of
higher complexity than video decoders. This is because bandwidth. A recent study shows that up to 70% of the power
encoders must analyze the redundancy in the video in order to used when a person is playing an on-line game on a mobile
compress the video efficiently. It is acceptable conventionally device goes to wireless communication [33]. As Moore’s Law
because usually the video is encoded only once while it is continues to scale down the power consumption of the
decoded multiple times, e.g., DVD or video on-demand. computation circuits, the power consumption of the
However, in the new paradigm, due to low power budgets, the communication circuits will become even more dominant.
redundancy cannot be analyzed completely at the video sensor Many connected devices are battery-powered, and current
node. A good video encoding algorithm must shift the human-to-human communication designs do not consider
computational complexity from encoders to decoders [8]. energy efficiency as the first priority. Yet, energy-efficient
After the collected data is processed, the information must machine-to-machine communication is crucial. In particular,
be transmitted to the gateway or even to the backend server. signaling costs are high. Self-organizing hybrid distributed and
While the power of digital circuits scales reasonably with centralized competition and cooperation framework may be
Moore’s Law, the power of analog circuits does not scale well. one approach to reducing signaling overhead.
Therefore, analog circuits in wireless communication will
consume relatively larger amounts of power than digital C. “Zero-touch” management & analysis
circuits. For low-power sensor node, design of low-power
transmitter circuits should also be considered. One solution is Today, the application development infrastructure of the
to use digitally intensive circuits to replace analog circuits in M2M is highly fragmented and proprietary [30]. Lack of
wireless communication. common tool kits for application developers actually hinders
To prolong the battery life of the sensor, we can harvest the growth of M2M. Development of a flexible, scalable,
energy from the ambient environment from sources such as robust, and optimized applications development framework
lights, heat, vibration, or radio frequency. The efficiency of for data analytics, data security, and sensor management is
today’s RF energy harvesting solutions is around 16.3% [35]. also important.
More efficient solution are thus an urgent need. Further, the Turning the sea of data into useful contexts or wisdom is
amount of energy that can be harvested from RF is at the uW extremely critical. Massive multi-modality, heterogeneous
level, still to low to power a wireless sensor. We also need a sensors create great complexity in analyzing data. Since data
high-efficiency energy harvesting circuit adaptable to different may be collected by hundreds or thousands of different
sources. sensors, the incoming data rate may be billions or trillions of
bytes per day, with the information noisy and unreliable in
B. Better connectivity nature. An API should be defined so that (1) common machine
learning and data mining techniques can be easily used and (2)
Another challenge in M2M is autonomous networking to application specific problems can be effectively addressed. To
connect large, dense populations of stationary and moving increase the reliability of the information, the temporal and
devices with efficient energy usage. spatial redundancy should be leveraged to detect anomalies.
First, many existing wireless standards are optimized for To reduce the computation time of network inferences, fast
human-to-human applications, but may not be able to support hashing techniques may be used, as the majority (95%) of the
large numbers of devices in a limited spectrum. Moreover, in data may not be useful.
the future, the rate of increase in the available spectrum will be Another great opportunity is to design a context-aware,
slower than the rate of increase in the number of wireless decentralized analysis algorithm. Many analytic algorithms
sensors. Fortunately, machine-to-machine communication has assume that the system has all the data on the server. However,
several unique characteristics: the data rate is often lower, the it takes power and bandwidth to communicate the data to the
information from different sensors or at different time steps server. Moreover, not all the data are important or useful.
may have strong correlations, and some messages do not Architecture-wise, it is natural for sensor data to be processed
require real-time delivery. Therefore, one approach to these in a hierarchical and distributed fashion. Data may be analyzed
problems is to form clusters of machines. Instead of and fused in sensors or gateways before arriving at the data
communicates with the base station directly, machines talk to center to save energy and bandwidth. If the system can first
understand what context is important, it

386
detect deviations from expected system communications in
real-time. Finally, distributed peer-to-peer machine
interactions and wireless communications demand
only needs to transmit the relevant information to the backend attack-defense systems that collect and analyze incoming
server or the cloud. As devices become more computationally attack information in a distributed fashion. However, usually
capable, intelligent computation may easily be distributed the amount of available data is limited and sometimes only a
between the sensors and the backend servers. Furthermore, it partial view of the entire system is possible.
will help if the computation can easily migrate from one piece Most current wireless sensor networks require great human
of hardware to another piece of hardware. This is because the effort to configure and deploy applications. Thus, human
computational capabilities of the devices and the bandwidth beings are often the bottleneck in large-scale deployment and
availability and energy consumption of the communication will long-term sustainability of systems in the field. One of the key
improve over time. However, we need to explore programming components that will increase the adoption rate and breadth of
techniques to improve software development processes for M2M across multiple vertical markets is a configurable
distributed heterogeneous middleware that supports intelligent devices
S7-1 Figure 5: An inter-discipline research effort to quest for
unified system architecture and standard interface, which
can increase the rate and breadth of M2M adoption
Services and applications research
management with self-configuration, self-optimization,
… self-healing, and self-protection capabilities. For example,
gateway devices should be able to detect, diagnose, decide, and
defuse faulty end nodes. Master devices should be able to
Context computing research auto-configure devices by deploying in-situ program updates
based on context changes or user commands. In this case, the
Communication research whole system can be reloaded and upgraded remotely. Further,
if the middleware is designed well, it is not only easy for
Sensor research
application developers to reprogram devices after deployment,
but also easy to develop the applications/services in a
computing. Furthermore, open challenges in this direction distributed heterogeneous computing environment before
remain. These include (1) network protocols, and data formats deployment.
are not yet compatible across different devices, applications,
and servers; (2) sharing devices resources across multiple D. Smart service for People
applications is not a common practice; and (3) infrastructure
will become incompatible over time. Diverse and fragmented standards and interfaces between
Data security and privacy is always a major concern, even layers of the systems hinder the innovation capabilities of
more important in M2M, which touches many aspects of M2M application developers and service providers. If we want
human life. Some low-cost devices have a limited budget to the M2M industry to grow as reliably and powerfully as
implement strong security or cryptography features. These Moore’s Law, we will need to develop the same kind of
lightweight devices can become the weakest links in the interdependent, mutually advantageous industry structure that
system. Conventional firewalls that provide network security causes the PC software and hardware industry spiral to
by blocking malicious traffic can no longer work in M2M function. Standard interfaces are the key to the spiral. When
because of its decentralized nature. If the lightweight devices each individual component provider has a fixed boundary
are not properly secured, the data they produce cannot be fully condition, it is easier and faster to create innovative solutions.
trusted. Attackers may provide false information that alters the It is not easy for one single component or solution provider to
behaviors of the system. Designing low-cost and scalable define a generic system architecture and the interface. The best
crypto algorithms and hardware accelerators is crucial. way is to bring community efforts together, as shown in Figure
A system-level security analytics and self-adaptive security 5. While some works on across-the-broad sensor,
policy framework are also needed. Security policies specify communication, and context computing technologies, some
trust relationships between entities, information creation and should develop a few key vertical applications and interact
protection rules, and access control rules to any target asset in deeply with the horizontal components. After understanding
the system. Security policies also define requirements of the the requirements from multiple verticals, we may be able to
actual security mechanism that enforces security policies. In create standardized platforms to better serve human beings.
the current common practice, the security policies are either The broader the diverse set of applications is (use fixed and
administered by IT departments or statically configured with mobile sensors to monitor both human and non-human objects
fairly naïve rules. However, often these policies do not satisfy and perform energy saving, vehicle safety, health/wellness
the requirements of the applications. This leads to an monitoring, and environmental monitoring functions), the
incomplete system. Further, unique challenges are faced in easier we can define the generic architecture and standard
creating suitable security policies for M2M. First, the expected interfaces. After that, we can focus on building innovative
behaviors of a M2M system may evolve over time due to M2M services to server humans.
environmental and/or policy changes. It is thus necessary to
develop solutions for detecting system behaviors outside
original specifications while adapting to changes yet be able to V. Summary
 As growing numbers of devices are added to the internet,

387
S7-1 Internet of things," December 2010.
[18] IoT-A, Internet of Things - Architecture,
http://www.iot-a.eu
M2M will transform the way we live, play, and work. An [19] IoT-i, Internet of Things Imitative, http://www.iot-i.eu
exciting area for innovation, it offers numerous challenges and
opportunities, from scaling applications and services from
billions to trillions of connected devices, and from tera to zeta
bytes of data. [20] IPSO Alliance, IP for Smart Objects
[21] ISO/IEC JTC 1/SC 6 Telecommunications and
information exchange between systems
References [22] ISO/IEC JTC 1/SC 31 Automatic identification and data
[1] 3GPP MTC, Machine-type communication capture techniques
[2] M. Balazinska, A. Deshpande, M.J. Franklin, P.B. [23] ITU-T IoT-GSI, Internet of Things Global Standards
Gibbons, J. Gray, S. Nath, M. Hansen, M. Liebhold, A. Initiative
Szalay, V. Tao, "Data Management in the Worldwide [24] ITU-T JCA-IoT, Joint Coordination Activity on Internet
Sensor Web," IEEE Pervasive Computing, vol.6, no.2, of Things
pp.30-40, April-June 2007 [25] ITU-T Study Group 2 Operational aspects of service
[3] C. Beckmann, S. Consolvo, and A. LaMarca, “Some provision and telecommunications management [26] ITU-T
Assembly Required: Supporting End-User Sensor Study Group 11 Signaling requirements, protocols and test
Installation in Domestic Ubiquitous Computing specifications
Environments,” UbiComp 2004, LNCS 3205, pp. [27] ITU-T Study Group 13 Future networks including mobile
107-124, 2004. and NGN
[4] M. Bellis, "The History of the ENIAC Computer," [28] ITU-T Study Group 16 Multimedia coding, systems and
About.com Guide applications
[5] G. Brumfiel, "First Eyes Inside Nuclear Plant May Be A [29] ITU-T Study Group 17 Security
Robot's," NPR, March 23, 2011. [30]K. Johnson, “Machine-to-Machine: Reinventing
[6] I. Cha, Y. Shah, A. U. Schmidt, A. Leicher, M. V. Embedded Devices for Smart Services,”
Meyerstein, "Trust in M2M communication," Vehicular http://www.intel.com/content/dam/doc/technology-brief/
Technology Magazine, IEEE , vol.4, no.3, pp.69-75, Sept. m2m-reinventing-devices-for-smart-services.pdf
2009 [31] A. M. Kelly, "Telecom Milestone: More Cellphone-Only
[7] K. Chang, A. Soong, M. Tseng, and Z. Xiang, "Global Than Landline-Only Households," Mediamark Research
Wireless Machine-to-Machine Standardization," IEEE Inc., Sept 2007.
Internet Computing, vol.15, no.2, pp.64-69, March-April [32] G. Lawton, "Machine-to-machine technology gears up for
2011. growth," Computer , vol.37, no.9, pp. 12- 15, Sept. 2004 [33]
[8] C.-C. Chiu, S.-Y. Chien, C.-H. Lee, V. S. Somayazulu, and T. Pering, Y. Agarwal, R. Gupta, and R. Want, “Coolspots:
Y.-K. Chen, "Distributed video coding: a promising reducing the power consumption of wireless mobile devices
solution for distributed wireless video sensors or not?" in with multiple radio interfaces,” in Proc. of ACM Mobile
Proc. Visual Communications and Image Processing 2011, Systems, Applications and Services, pp. 220–232, 2006.
Nov. 2011. [34] Public Polls from EE5158 students, National Taiwan
[9] CISCO, “How the Internet of Things Will Change University, Sept. 2011.
Everything,” http://youtu.be/mf7HxU0ZR_Q [35] A. Sample and J. R. Smith, “Experimental results with
[10]CISCO, “Imagine the Possibilities,” two wireless power transfer systems,” Proceedings of the
http://youtu.be/RGk7llP9Z8k 4th international conference on Radio and wireless
[11] T. Cory, “Building M2M services means overcoming new symposium, pp. 16-18, Jan. 2009.
challenges”, Telecom Engine, December 13, 2009 [12] D. [36] J. Santa, A.F. Gomez-Skarmeta, "Sharing Context-Aware
Estrin, "Participatory sensing: applications and architecture Road and Safety Information," Pervasive Computing,
[Internet Predictions]," Internet Computing, IEEE , vol.14, IEEE , vol.8, no.3, pp.58-65, July-Sept. 2009
no.1, pp.12-42, Jan.-Feb. 2010 [37] J. Sen, "Internet of Things-A Standardization
[13] ETSI TC M2M, European Telecommunications Standards Perspective," July 2010,
Institute technical committee machine-to-machine [14] www.gisfi.org/wg_documents/GISFI_IoT_2010062.pdf
European Commission, Directorate-General for the [38] SENSEI, Integrating the Physical with the Digital World
Information Society and Media, "Vision and challenges for of the Network of the Future,
realising the Internet of things," 2010. http://www.sensei-project.eu/
[15] L. Greenemeier, "Can Mobile Phone Networks Be [39]J. O. Smith, “The Coming of Age of M2M Standards,”
Improved to Better Cope with Emergencies?" Scientific Connected World Conference, Keynote presentation,
American, August 2011. 2010.
[16] IEEE Communications Magazine, special issue on [40] R. Staszewski, T. Jung, R. B. Staszewski, K. Muhammad,
"Recent progress in machine-to-machine D. Leipold, T. Murphy, S. Sabin, J. Wallberg, S. Larson,
communications," April 2011 M. Entezari, J. Fresquez, S. Dondershine, and S. Syed,
[17] IEEE Wireless Communications, special issue on "The "Software Assisted Digital RF Processor for Single-Chip
GSM Radio in 90 nm CMOS," IEEE Custom Integrated
 Circuits Conference, 2006. uniteChannel, “The simple truth about LTE,”
[41] TIA TR-50, Telecommunications Industry Association http://youtu.be/asxYJw7wlHE
Smart Device Communications Standards Committee [42]

388