5G – Personal Mobile Internet Beyond What Cellular Did to Telephony

Accepted for publication in IEEE Communications M agazine, to appear February 2014.
5G – Personal Mobile Internet
Beyond What Cellular Did to
Gerhard Fettweis, Vodafone Chair Professor, TU Dresden, Germany
Siavash Alamouti, Entrepreneur, London, UK
Cellular technology has dramatically changed our society and the way we communicate. First it impacted voice telephony, and then has
been making its inroad into data access, apps, and services. However, today potential capabilities of internet have not yet been fully
exploited by cellular systems. With the advent of 5G we will have the opportunity to leapfrog beyond current internet capabilities.
Life without modern wireless communications, as it existed before 1990, is now difficult to imagine. It empowers our modern life, enables
modern societies to operate efficiently, and has had a major impact on modern politics, economy, education, health, entertainment,
logistics, travel, and all industries.
In order to have a productive life, we need to obtain and share information from various sources. Thankfully, most of the information and
data known to mankind has been digitized in one form or another and is available for consumption and sharing. Hence, we require reliable
and fast data communications for real-time access and sharing of this information which is stored and consumed in various formats and
structures (text, image, audio, and video).
First and second generation mobile communications networks were dominated by analog and then digital audio signals and text messaging.
The third generation was more about scaling the numbers of users on the network for voice communications and text messaging but was
overwhelmed by an unpredictable tsunami of image and video content. This trend is sure to continue. Video, audio, and image formats are
going to become richer and will require even more data most probably beyond improvements in codec technologies. Therefore, the thirst
for data communications is going to continue and our transmission networks will most probably remain the bottleneck. Hence, we need to
provide as much capacity as we can and ensure that we build an efficient and smart architecture that can accommodate future demands for
data communications.
M oreover, the landscape of information has expanded greatly to machines. There are many new gadgets and devices (hereby referred to by
machines) that are potential sources of valuable information. Thanks to advances in communications technology, machines can be
connected and reached cost-effectively and will fast become an integral part of the global information network. This will definitely drive an
ever-increasing demand for monitoring and sensing data and applications. The nature of most of this data will be different than
conventional human generated content, and will mostly be short, bursty, and asynchronous. Nevertheless, the number of these devices will
be orders of magnitude larger than the conventional communications devices of today. Hence, we need to architect our networks to handle
10s of billions of devices growing rapidly and limited only by the capabilities and capacities of our networks.
Beyond our need for communication and sharing is the need to steer/control elements of our surrounding and environment, as gadgets,
sensors, and machines that will help us carry out our day-to-day life more efficiently. Once machines become connected, the next natural
leap is to have them controlled remotely. This will generate a completely new paradigm for control communications.
Today we enjoy the power of telephony and data communications. Our 4G networks will enable real-time access to richer content and will
enable early application of machine type communication while control communications will be at its infancy. This article provides a
recollection of our current state for motivating and sketching a vision of our future.
The mobile industry has had a chronology of revolutionary applications and technologies that have shaped everyday life of their customers.
First and foremost, the need for untethered telephony, and therefore real-time wireless communications, dominated the success of cordless
phones, followed by cellular communications. Soon thereafter, two-way paging implemented by text messaging became another killer
application. With the success of wireless local area network technology (WiFi based on the IEEE 802.11 standard), internet browsing, and
the widespread market adoption of laptop computers, untethered internet data connectivity became a reality and ultimately a necessity for
everyone. This phenomenon opened the market for cellular broadband wireless data connectivity. The logical next step was to invent a
better user experience for a subset of laptop functions for mobile use and merging it with the cellular telephone, which evolved into today’s
smartphone. We now enjoy access to the world’s information at our fingertips, anytime, anywhere. But, is this the end game? Is everything
else going to be evolutionary? As difficult as it is to predict, history has shown that future is ripe for transformations and inventions
especially since we are far away from an ideally connected world.
When the cellular concept was first introduced, it was not adopted immediately and took more than 20 years to be deployed. There were
commercial concerns against business viability of a consumer mobile communication service. One major concern with the cellular concept
Accepted for publication in IEEE Communications M agazine, to appear February 2014.
was that telephony was already ubiquitous and available everywhere: in peoples’ homes, offices, hotels, public venues, on every block in
every street, and even in automobiles if you were willing to pay for it (and in fact very few people did; strengthening the commercial
concern with the cellular concept). In other words, there initially was no perceived business case or “killer application” for cellular
communications and hence reluctance to make larger investments in infrastructure required to make it happen. Another reason for the long
delay in recognizing the cellular opportunity early on was the underestimation of the prospects of semiconductor scaling. This enabled
mobile phones at a size, price, and cost which make them portable and affordable for everyone. As a result, the industry could not predict
that cellular communications would create a paradigm shift that would forever change all our lives. The paradigm shift was the
personalization of communication and the ability to call people instead of places. Nobody had predicted the scale of impact of cellular
communications and that it would be one of the largest businesses in the history of mankind with an immense impact on all economies [1].
Is the world ripe for another paradigm shift? It sure feels like that. Internet and mobile communication have become such as integral part of
our lives. Yet, our experience with it is fragmented and far from ideal; very much like in the early days of telephony and before mobile
Starting with connectivity, the experience is poor and unpredictable. We need to be aware of the networks (3G, 4G, WiFi, Bluetooth, etc.)
and need to sign in and pair in order to get access and at times we cannot get connectivity required for the simplest applications even
browsing. The fourth generation cellular will go a long way to address that but the thirst for data will continue and we need to provide more
and more capacity as time goes by far beyond a 2 to three times the spectral efficiency and order of magnitude capacity improvement from
Another area of opportunity is the form and context of communications. We have many ways of reaching others which can become very
confusing. We need to reach a state where we will seamlessly connect to the right individuals at the right time and through the right means.
As a consequence an individual driving should never be enticed to physically pick up a phone to answer a phone call or to text. The context
should be known and the best method of communication should be chosen based on the context of the users. Our devices and gadgets need
to evolve to be able to present information in the most convenient form possible using wearables (watches, glasses, etc.) and many physical
objects should be capable of sending, receiving, and rendering information (smart tables, whiteboards, screens, etc.)
Content discovery is another area ripe for fundamental change. Because of the colossal amount of information on the internet and intranets,
it is becoming more difficult to find relevant information and content. Internet is an improvement over the old broadcast model of content
distribution as it has increased our choice and reach but it is still far from ideal. Even the vast improvements in speed and quality of search
engines have not been able to keep up with the demand for relevant information. The business model for search and discovery has further
impacted the quality of content discovery. The more you pay, the more “relevant” your content will become for the users and will appear on
the top of their list. But more importantly, you have to search for your relevant content repeatedly and inefficiently.
We believe this is where the next paradigm shift in mobile communications may lie. Instead of the consumers going to internet, the internet
will come to them and in fact we will become nodes on the internet. We will become both the source for valuable information and the sink
for highly personalized information and content. For this to happen, eventually, all people and the information on their context of their
environment need to be continuously available for one another. Also, the context needs to be understood and communicated across multiple
nodes, and our gadgets and networks need to become smart enough to understand what content is relevant to whom. Clearly, this requires a
new level of security, integrity, and safety to be implemented for achieving the necessary privacy for the technology to be accepted and
launched successfully.
A well-recognized trend is connecting humans with machines and with other machines, referred to as “M 2M ” (or “M TC” for machine type
communications). During the past years we have seen a multitude of wireless M 2M applications being deployed, for example in public
transport systems and vending machines. However, the commercial success has been somewhat limited. Why? The previous revolutionary
applications have been clearly driven by addressing human needs for basic communications and our networks have been designed for voice
and text applications, whereas M 2M , by nature, has a very different set of requirements. The application of cellular communications to
M 2M has been an after-thought. Undoubtedly, we will see major breakthroughs in this area and we should thrive to understand these
requirements on their merits. For M 2M to reach its full potential, it needs a network optimized for it. The big question that our industry
needs to answer is whether we will have the same network designed for both human and machine communications; need a new dedicated
network for machines, or a hybrid.
In the next section, we discuss the history of wireless and attempt to draw some conclusion to help plan for the future roadmap.
Content: The Wireless Roadmap
Up until now, telecommunications has been mainly used for transmission of various kinds of content, namely voice, video, data, email, and
other. The majority of this content is generated or stored in devices. In the past decades, advances in mobile electronic devices have been
staggering. Within mobile devices, content is mainly stored in flash memory. The storage capacity of flash memory has kept growing
according to M oore’s Law, i.e. doubling every 18 months. This is equivalent to an order of magnitude increase (10×) every 5 years, as
depicted in Figure 1. As embedded memories grow, so does the need for transmitting more and more data [2].
Accepted for publication in IEEE Communications M agazine, to appear February 2014.
Figure 1: The market introduction of single-chip flash memory capacity showing a doubling in capacity every 18 months along Moore’s
Law, leading to a 10× increase every five years.
In the past two decades, wireless communications has advanced greatly. The simplest measure of performance increase has been advances
in achieving higher data rates. This trend for increased bandwidth has been following the same exponential increase as M oore’s Law for
semiconductors, as shown in Figure 2.
Figure 2: The Wireless Roadmap, showing the market entry of technologies as tracked by the authors.
WLAN systems are intended for local area broadband communication services in homes, offices, and hot spots with peaks data rates
reaching hundreds of M bit/s. With WiGig on the horizon, yet another order of magnitude improvement in speed and latency of data transfer
is possible.
The phenomenal success of cellular has been based on providing ubiquitous and reliable wide area coverage for voice and text
communications. With 4G and the success of mobile computing devices the industry is targeting to bring the same reliability and ubiquity
of access to mobile internet applications such as web browsing and audio and video streaming. LTE today already is providing effective
data rates of around 50 M b/s.
The gap between the data rate speed of cellular and WLAN has been 100× (Figure 2). However, the use cases for mobile data on devices
have converged. Consumers only care about applications and services which demand that the underlying network to become completely
transparent. In other words, the distinction between applications for local area and wide area communications will disappear and we will
need a truly heterogeneous network, leading to a new inflection point of technology [3].
Accepted for publication in IEEE Communications M agazine, to appear February 2014.
Another significant driver is the vast amount of spectrum available in the millimeter range beyond 100 GHz for unlicensed WLAN systems
with reasonable propagation properties [4], whereas licensed cellular systems might be restricted to sub-100 GHz frequencies only. Recent
results in the sub-100 GHz band show promising results for cellular hot-spot coverage [5]. The physical layer design for cellular systems is
constrained by availability of licensed spectrum whereas WLAN technologies will have an abundance of frequencies in the unlicensed
band. What is however very clear is that with the next generation of wireless technologies, the various access technologies will have to be
highly coordinated and integrated into a seamless experience unlike the fragmented experience of today.
Setting the Stage for LTE and Beyond: Considering technology and market forces in 10 years, we must be able to address cellular speeds
of 10 Gb/s or more. Current wide area access technologies do not meet this requirement. Using OFDMA in wireless systems requires an
analog/digital conversion with a higher bit-resolution due to its Gaussian amplitude distribution. Hence, for upcoming high data rates,
analog/digital conversion with 10-bit resolution alone would most likely represent a power consumption challenge that is difficult to be
resolved with currently available or projected technologies [6]. This suggests that a new physical layer approach may be needed for 5G
cellular communications.
Collecting information by monitoring our environment for understanding the current status or predicting the future is of great value to our
future progress in almost all industries. M any of today’s smartphone apps fulfill some of our needs, e.g. providing information on location,
speed, prices of goods, and weather status. M any new devices such as connected wrist bands collect sensory information on activity and
vital signs. Obviously, we are at the very early stages of this trend. There are applications that can have significant impact on our daily
lives. For example, every individual plant could be monitored and classified. We could ensure proper light and moisture for optimum
growth and health and increase productivity in agriculture and food production. This is at the top of the food chain which seemingly is
furthest from the technology. You can extend this to many other applications.
For new M 2M applications to become technically and financially feasible, adding connectivity to devices must be very cheap and central
management of devices needs to be reliable and inexpensive. Given that many of these devices have mobile use cases (for instance, body
sensors, or sensors inside cars); most of these devices need wide area connectivity. There is a collective (or social) network aspect of these
devices as well which could further enhance use cases. For instance, for winter sports, if every ski or snow board could measure and report
the status of the run, the information could be aggregated to provide real-time information on the status of every run so that skiers could
select their runs based on their desired profile (powder, ice, moguls, etc.) Expanding this vision to our daily lives you could have 100s of
billions of connected devices assisting human beings in their daily lives, chores, and work [7].
We need to translate the industry vision for M 2M into requirements to the extent that we can envision and attempt to draw some
conclusions on technology requirements. An ideal M 2M device is one that can be activated only when needed, can run “forever” on a small
battery or on an energy scavenging unit, and transmits the required amount of information only when needed. The rate at which the data is
transmitted and the amount of information communicated depends highly on the particular use case.
An example specification for the use case of a smart meter is to transmit a packet of 25B payload every 100 seconds, to operate on an
average 200µW power budget (a 2” solar cell), and be able to provide a 10dB better link budget than GSM .
Setting the Stage for LTE and Beyond: We believe it is technically and financially feasible to provide M2M services at a very large scale.
However, this is not true with current cellular systems as their protocols require too much communications overhead for synchronization,
channel allocation, and mobility/connectivity management. As a result, they are far away from meeting the power requirements for a
broad-scale M2M cellular sensing system. Hence, a new (5G) standard beyond LTE is needed.
Immersive Steering & Control: The Tactile Internet
In this section we focus on some new use cases for real-time services. Real-time is a highly subjective term and depends on the use case.
We define a service to be real-time when the communication response time is faster than the time constants of the application. We consider
four types of physiological real-time constants: muscular, audio, visual, and tactile. For literature on human reaction analysis we refer to
[8], and the overview given in [9].
Humans have the ability to react to sudden environmental changes using our muscles. For example, when reacting to a sudden unforeseen
incident by hitting the brakes in a car, or quickly pulling back a hand after touching a hot platter on a stove. There are two distinctly
different time scales of reaction, depending on being prepared or unprepared for the situation. If unprepared, the sensing to muscular
reaction time is in the range of 500ms to 1 second. When translating this to comparable situations in technical applications this sets the
targets for specification. An easily understood example is interactive browsing of the Web. The page build-up after clicking on a link
should be a fraction of this time so that we get a sense of immediacy. Henceforth, real-time browsing interaction is experienced if new Web
pages can be built up after clicking on a link within a few hundred milliseconds. A shorter latency, that is, a faster reaction time of the Web
is not necessary for creating a real-time experience as the difference is not perceivable by humans. This reaction time has generally been
met by initial 802.11b and 3G cellular systems.
Clearly, if humans are prepared for a situation, faster reaction times are needed, e.g. when driving a formula-1 car in a race, or during highspeed typing on a keyboard. In a communications setting this would be the case e.g. if while clicking a link the new Web page is known
and the user is preempting where to click next. The IAAF (International Association of Athletics Federation) has come to the conclusion
that the fastest “prepared” neural-muscular physiologic reaction time is 100ms. For instance, in track and field competitions, any sprinter
Accepted for publication in IEEE Communications M agazine, to appear February 2014.
who reacts faster than 100ms after the shot is considered to have a false start, even though some studies show that 85ms could be possible
The next shorter real-time latency constant is experienced with the hearing system. To humans, conversations appear as real-time when we
receive the audio signal within 70 to 100 ms. This is why the International Telecommunications Union (ITU) has set this as a minimum
latency requirement for telephony and speech delays on telephone lines have to be in that order of magnitude. This is not only applicable to
two-way communications. For instance, lip synchronization between the video stream and the sound track needs to be within the same time
lag, otherwise the sound seems disconnected to movements on the video. Assuming the core network is provisioned properly and the
wireless network is far from being fully loaded, an overprovisioned LTE network or even later generation 3.x generation networks meet
this requirement. As a result, internet video conferencing (e.g., Skype) is generally viable over averagely loaded cellular networks today.
When immersed in watching a scene, our vision has a typical temporal resolution of 100 Hz. M odern TV sets therefore have a minimum
picture refresh rate of 100Hz to allow for a seamless video experience, translating into a maximum10 ms inter-picture latency requirement.
Figure 3: Coarse categories of physiological real-time constants.
The most challenging latency requirement comes from tactile/haptic action of human limbs with visual or audio feedback. Latency
requirements for tactile action are on the order of 1ms. On some results for haptic interaction see [9]. A good example is moving or
dragging an object on a touch screen. At a relatively fast speed, our finger moves at 1m/s over the screen. To create a seamless experience
the object must follow the movements of the finger without experiencing a displacement between finger and object. If we assume 1 ms of
latency, and 1 m/s speed for the movement of the finger, the resulting displacement is 1mm which is acceptable. M uch larger latencies
would be unacceptable. For instance, a latency of 25ms would lead to 2.5cm, or 1 inch of displacement which is clearly noticeable and
results in very poor experience.
Also, extreme situations where the 1ms latency requirement can be experienced are when moving a 3D object with a joy stick or in a virtual
reality environment. If the time lag between the virtual picture and the movement is above 1ms, motion sickness can occur which within
this context is referred to as cyber sickness [11]. Hence, a real-time cyber physical experience requires an extreme latency constraint of
below 1 ms. Current wireless cellular systems miss this target by nearly two orders of magnitude (see Figure 3).
The Real Time Challenge at 1ms – “The Tactile Internet”
When translating the latency restrictions into building electronic systems, we need to consider the speed of light. Light travels 300 km
within 1 ms. In order to meet a 1ms roundtrip latency constraint, we need to consider limitations in hardware and software design. Figure 4
shows one possible latency budget over a communications chain, taking into account the latencies from the sensor through the operating
system, the wireless/cellular protocol stack, the physical layer of terminal and base station, the base station’s protocol stack, the trunk line
to the compute server, the operating system of the server, the network within the server to the processor, the computation, and back through
the equivalent chain to the actuator.
Accepted for publication in IEEE Communications M agazine, to appear February 2014.
Figure 4: The impact of breaking down the 1ms roundtrip delay.
Setting the Stage for LTE and Beyond: Each and every element of this communications and control chain must be optimized for latency.
Assuming this, the latency budget for the physical layer is at most 100 µs. LTE has an OFDMA symbol duration of 70 µs. Therefore, if we
were to meet the latency requirements for tactile communications, we would need to design a different physical layer. Hence 5G addressing
these new interesting application areas of the “Tactile Internet” [12] requires a completely new physical layer for or an overlay on top of
OFDMA for these applications.
We have envisioned the high level requirements for 5G cellular communications with a user-centric approach, considering 3 dimensions of
innovation: higher data rates to address network traffic demands of the future, operations for M 2M sensing devices, as well as a 1 ms realtime latency (see Figure 5) to meet requirements for tactile control.
Figure 5: The wireless network moving from communications of content to monitoring to control. The dashed lines indicate that
applications allow for a trade-off between the three depicted criteria of optimization: speed, response time, and operation endurance.
Accepted for publication in IEEE Communications M agazine, to appear February 2014.
As all three dimensions may not need to be addressed simultaneously for any class of service, it can be assumed that it is feasible to design
a new 5G system that can meet these differing requirements, and that it may differ greatly from 4G LTE.
Cellular communications with data rates of 10 Gb/s can enable ubiquitously available immersive virtual reality at levels not
Context awareness and personalization built into all devices and services can transform the way we communicate as humans.
It is most likely, that we will have various access schemes to fulfill all our communications needs. These schemes need to
become highly integrated and coordinated to provide a seamless experience to users and applications.
Communicating sensors embedded in our environment can enable a host of new services. We can move from cellular/wireless
data communications to a new level of wireless monitoring.
A roundtrip latency of 1ms can potentially move the world from enjoying today’s wireless communication systems into the new
world of wireless control systems – the Tactile Internet. It will dramatically change our life, impacting all aspects of application
areas such as health, safety, traffic, education, sports, games and energy. For a first overview on application areas and challenges
see [12].
Discussions on setting the stage for LTE and beyond shows that a new physical layer may be needed possibly parting from
OFDM and embracing new concepts such as Generalized Frequency Division M ultiplexing or Filter Bank M ulti-Carrier [13].
The result may be a revolutionary leap from today’s wireless communications to personalized context based communications and future
monitoring and control networks. The future opportunities in mobile communications are larger than anyone can foresee. What we
experience today is only the very first glimpse. Obviously, with challenges as pointed out here, 5G cellular communication will be another
paradigm shift that will redefine our future, impacting out societies in ways which cannot be foreseen.
[1] Vodafone Institut, “Mobile Technologies – the digital fabric of our lives,” www.vodafoneinstitut.de/uploads/media/Studies_komplett_FINAL_von_250913.pdf, September 2013.
[2] G. Fettweis, “ A 5G Wireless Communications Vision ,” in Microwave Journal, December 14, 2012.
[3] Clay Christensen, “The Innovator’s Dilemma,” HarperBusiness Essentials, 1997.
[4] Jonathan Well, “Faster T han Fiber: The Future of Multi-Gb/s Wireless,” IEEE Microwave Magazine, May 2009, pp. 104-112.
[5] T . Rappaport et al., “ Millimeter Wave Mobile Communications for 5G Cellular: It Will Work!,” IEEE Access, Volume 1, Digital Object Identifier
10.1109/ACCESS.2013.2260813, May 2013, pp. 335-349.
[6] S. Krone, “ Wireless Communications with Coarse Quantization: Information-Theoretic Analysis and System Design Aspects,” Vogt Verlag, ISBN
9783938860519, 2012.
[7] M.A. Uusitalo, “Global Vision for the Future Wireless World from WWRF,” IEEE Vehicular Technology Magazine, Vol. 1, issue 2, 2006, pp. 4-8.
[8] Jeff Johnson, “ Designing With the Mind in Mind,” Morgan Kaufman Publ., 2010.
[9] E. Steinbach, S. Hirche, M. Ernst, F. Brandi, R. Chaudhari, J. Kammerl, I. Vittorias, “Haptic Communications,” Proceedings of the IEEE, Vol. 100,
No. 4, April 2012, pp. 937-956.
[10] Matthew T . G. Pain, Angela Hibbs , “ Sprint starts and the minimum auditory reaction time,” Journal of Sports Sciences, Volume 25, Issue 1 January
2007 , pages 79 – 86
[11] T om DeFanti and Rick Stevens, “Teleimmersion,” Chapter 6 within T he Grid: Blueprint for a New Computing Infrastructure, T he Elsevier Series in
Grid Computing, pp 131-155.
[12] Gerhard P. Fettweis, “The Tactile Internet – Applications & Challenges,” accepted for the IEE Vehicular T echnology Magazine, to appear 2014.
[13] G. Wunder et al., “ 5GNOW: Non-Orthogonal Asynchronous Waveforms for Future Mobile Applications,” to appear in the same issue of the IEEE
Communications Magazine, February 2014.
Gerhard P. Fettweis, born M arch 16, 1962, received his Ph.D. from RWTH Aachen in 1990. Thereafter he joined IBM Research and then
TCSI, California. Since 1994 he is Vodafone Chair Professor at TU Dresden, Germany. So far he has co-founded 11 startup.
He runs the German Science Foundation’s CRC HAEC and COE cfAED. He helps organize IEEE conferences, e.g. TPC Chair ICC 2009
and TTM 2012, General Chair VTC Spring 2013 and DATE 2014.
S iavash Alamouti, born M arch 16, 1962, received B.A.Sc. and the M.A.Sc. degrees in EE from the University of British Columbia. He
is currently an entrepreneur helping build new companies focused on personalization. He was the Group R&D Director for Vodafone from
2010 to 2013. Before that he was an Intel Fellow and CTO of the Mobile Wireless Group. Prior to joining Intel in 2004, he held positions
at various companies including Vivato, Cadence, and AT&T. He is most well-known for the invention of the Alamouti Code.
We would like to acknowledge the contributions of our teams at Vodafone Group R&D and at TU Dresden in shaping the vision, and we
also would like to thank the reviewers for their valuable comments.