C H A R T I N G T H E F U T U R E O F I N N O V A T I O N | # 8∙2 0 1 7

ERICSSON
TECHNOLOGY

Request to prioritize vehicle data traffic

Ericsson Cloud
Network traffic prioritization

Scania
command
Data traffic center
prioritization interface Remote
driver

(5G) core
network

Internet
Ericsson test
Scania bus network at Scania
test track Control plane
User plane Prioritized video traffic
User plane Prioritized control traffic
User plane Non-prioritized traffic
(e.g. mobile broadband, infotainment)

5G NETWORK
PROGRAMMABILITY
✱ 5G NETWORK PROGRAMMABILITY

5G network
programmability FOR MISSION-CRITICAL APPLICATIONS

5G will make it possible for mobile network operators to support

enterprises in a wide range of industry segments by providing cellular
connectivity to mission-critical applications. The ability to expose policy
control to enterprise verticals will create new business opportunities for
mobile network operators by enabling a new value chain through the
integration of telecom with other industries.

R AFIA INAM , A key differentiator of 5G systems from a unique and personalized experience.
ATHANASIOS previous generations will be a higher degree A number of use cases in mission-critical
K AR APANTEL AKIS, of programmability. Instead of a one-size-fits- scenarios can benefit from QoS programmability
LEO N ID M O KRUSH IN , because a cellular network’s connectivity
all mobile broadband service, 5G will provide
ELENA FERSMAN requirements – including latency, throughput,
the flexibility to tailor QoS to connectivity
service lifetime and cost – vary widely across
services to meet the demands of enterprise
different use cases. To support them all, we have
customers. This enables a new range of developed an application programming interface
mission-critical use cases, such as those (API) that allows third parties to specify and
involving connected cars, manufacturing request network QoS. We have also demonstrated
robots, remote surgery equipment, precision the usefulness of this API on a test mobile network
agriculture equipment, and so on. using a transport-related use case.
As part of this use case, we have been collaborating
■ Network programmability can support rapid with commercial vehicle manufacturer Scania to
deployment of new use cases by combining develop the QoS requirements for teleoperation.
cloud-based services with mobile network Teleoperation is the remote operation of an
infrastructure and taking advantage of new levels autonomous vehicle by a human operator in cases
of flexibility. Further, network programmability will where the vehicle encounters a situation that the
enable a greater number of enterprise customers to autonomous system cannot overcome by itself (a
use such services, and consumers will benefit from road obstacle or malfunction, for example).

2 ERICSSON TECHNOLOGY REVIEW ✱ JAN UARY 26, 2018

 5G NETWORK PROGRAMMABILITY ✱

Drivers of network programmability by using a software service orchestration function

The key drivers behind the creation of a instead of manual provisioning by engineers.
programmable network are the need to accelerate As orchestration will also be used for provisioning
time to market, and the desire to reduce operational connectivity services to mission-critical
costs and take advantage of the business applications, mobile networks need to support
opportunities presented by a new mission-critical QoS programmability.
service market. In a programmable network,
traditional network functions requiring specialized Mission-critical Intelligent
hardware are replaced with software functions Transportation System use cases
hosted on commercial off-the-shelf infrastructure. 5G will support a diverse range of use cases
Technologies such as software-defined networking in different industry sectors, each putting its own
and network functions virtualization are essential QoS requirements on the mobile network [1]. It is
to cutting operational and capital costs in mobile possible to use network programmability to realize
networks. mission-critical use cases with QoS requirements
Cloud-based services and applications are by creating highly specialized services tailored
enablers for programmability. Service provisioning to industrial needs and preferences.
in the cloud and managed access to the provisioned Features like lower latency (reaction times that
services and applications are important. This are five times faster), higher throughput (10 to 100
requires collaboration between telecom and other times higher data rate) and an enormous increase in
industries (IT application and content providers, the number of connected devices (10 to 100 times
and automotive original equipment manufacturers, more) can support the large-scale use of massive
for example). One way to simplify and accelerate the machine-type communication (mMTC) and
deployment of services and applications from mission-critical MTC (MC-MTC) use cases for the
industry verticals is the automatic translation of first time. Further, a dedicated network slice would
industrial requirements to service requirements, meet the specific requirements of each use case.
and then on to resource-level requirements (in In mMTC use cases, a large number of sensors
other words, network requirements). Network and actuators are connected using a short-range
slicing provides a dedicated, virtualized mobile radio (capillary network) to a base station (eNodeB)
network containing a set of network resources, and using a low protocol overhead to save the battery
provides guaranteed QoS. The network slices are life of the devices.
not only beneficial but also critical to support many This requires a network slice with broad coverage,
applications in vertical industries. small data volumes from massive numbers of
New network communication services can also devices. MC-MTC use cases emphasize lower
be provisioned programmatically; that is, latency (down to a level of milliseconds),

Terms and abbreviations

AAR – Authentication Authorization Request | AF – application function | API – application programming interface
eNB – eNodeB | EPC – Evolved Packet Core | EPS – Evolved Packet System | HSS – Home Subscriber Server |
ITS – Intelligent Transportation System | MME – Mobility Management Entity | mMTC – massive machine-type
communication | MTC – machine-type communication | PCRF – policy and charging rules function | PDN – packet
data network | PGW – PDN gateway | QCI – QoS class identifier | Rx – radio receiver | SAPC – Service-Aware
Policy Controller | SGW – service gateway | UDP – User Datagram Protocol | UE – user equipment

JAN UARY 26, 2018 ✱ ERICSSON TECHNOLOGY REVIEW 3

✱ 5G NETWORK PROGRAMMABILITY

in mobile networks. However, the bandwidth

WE ENVISION REALIZING requirements to make this possible are not
THESE USE CASES WITH A excessive, as only vehicle control data needs to be
communicated. This capability is expected in 5G.
FLEXIBLE NETWORK
PROGRAMMABILITY Teleoperation of vehicles
The ability to control a self-driving vehicle from a
TECHNIQUE distance is an important use case that is needed in
robust transmission and multilevel diversity due public transportation when an onboard, autonomous
to their mission-critical nature and, consequently, system faces a difficult situation, such as a traffic
need a network slice of very low latency, high accident, an unexpected demonstration, unscheduled
reliability and availability (packet loss down to roadworks or flooding. These scenarios require the
10-9). This is possible by creating a slice of very high planning of an alternate route, and an operator
priority. We envision realizing these use cases needs to drive the vehicle remotely for a short time.
with a flexible network programmability technique. Another case could be a mechanical malfunction or
The current focus of our research is within the an injury on a bus that requires remote intervention
Intelligent Transportation System (ITS) domain to mitigate the risk of danger to others. Network
and includes a few 5G use cases in mMTC and requirements for remote monitoring and control
MC-MTC, including transportation and logistics, include broad coverage, high data throughput and
autonomous cars and teleoperated vehicles. low latency to enable continuous video streaming
and the ability to send commands between a remote
Transportation and logistics operations center and a vehicle [2].
The lower latency and high throughput of 5G
will support multiple use cases related to connected Why an API?
cars, transportation and retail logistics that To guarantee QoS for the three ITS cases described
consist of fleets of connected/driverless vehicles above (and mission-critical use cases in general)
transporting people and goods. The key network mobile network operators typically go through
requirements for mission-critical automotive manual network planning and configurations.
driving are high throughput and low latency up to Examples include configuring manual data routes
100ms. Failure is not an option in these cases. via different routers, configuring Differentiated
There are also many potential sub-use cases. Services and allocating dedicated spectrum ranges
For example, a journey from A to B in a driverless to each use case. However, doing this is costly because
vehicle could involve vehicle-to-vehicle connections, it requires the configuration and deployment of
connections between vehicles and street network equipment. Nor is it particularly feasible,
infrastructure for traffic management, and as this kind of configuration deployment cannot be
high-speed reliable connectivity to support cloud done merely in parts of the transport network (such
applications. as backhaul). If, on the other hand, the resources
were virtualized and there was software that could
Autonomous vehicles set up these routes over the same physical network
The vision of fully autonomous vehicles aims to link, both of the limiting factors would be eliminated:
reduce the risks associated with human error. the cost of configuration and the deployment of
A system to achieve this vision would need to multiple routes. As a result, it would be both
connect the cars and the road infrastructure with financially and technically feasible to support
1ms latency in all areas (100 percent coverage). these use cases concurrently.
Unfortunately, 1ms latency is currently not possible It is clear that operators will not be able to

4 ERICSSON TECHNOLOGY REVIEW ✱ JAN UARY 26, 2018

 5G NETWORK PROGRAMMABILITY ✱

support the volume and diversity of use cases Figure 2 illustrates the architecture of the system
with the current network management approach. on which we have implemented the API. In addition
They need a different means of managing the to deploying a standard EPC and LTE band-40 RAN,
network to stay competitive. In our view, an AF is deployed on an OpenStack-managed cloud.
developing an API is the logical first step toward This application functionality allows third parties
exposing a programmable network to the industry to set up QoS for their UEs through an API.
verticals. This approach will result in a solution that The AF consists of the following components:
is more responsive than rigid commercial offerings, a knowledge base module, an API endpoint module
such as preconfigured subscription packages. and a transformer.

Architecture of the Ericsson-Scania project Knowledge base module

Teleoperating a bus requires data from sensors on This module maps domain-specific concepts
the bus, including a video feed from a camera at the to the generic concepts. The knowledge base is
front that is streamed to a remote operations center implemented as a graph database, has a schema
over LTE radio access with an evolved 5G core of general concepts and can be extended with
network. The commands to drive the bus are sent additional domain concept documents that
from the center to the bus using Scania’s command instantiate the general concept schema. The schema
system. includes a basic vocabulary of general concepts that
Figure 1 illustrates the data streams that need to model QoS requests. These concepts can be
be prioritized to meet QoS demands: sensor data, instantiated in domain concepts for a specific
the video feed originating from the vehicle user enterprise. In our case, the enterprise
equipment (UE) and the commands to remotely is automotive.
drive the bus. Sending these data streams over Within the knowledge base module, an “agent”
low-priority data traffic (like infotainment) is a string that is semantically related to the mobile
is a critical requirement. We used QOS class device for whic h QoS is requested. In our case, the
identifier (QCI) bearers, as detailed in the agent is instantiated with the “vehicle” domain
corresponding 3GPP standard, to enforce this concept. QoS class identifiers (QCIs) are indicators
prioritization. We assigned QCI class 5 and 2 to of network QoS for a given agent. The QCI concept
video and sensor data respectively and lowest- was introduced in 3GPP TS 23.203 Release 8,
priority QCI class 9 to infotainment. In our lab with additional classes being introduced in
environment, we have confirmed that the high- Release 12 and Release 14.
priority streams (QCI 2 and 5) can be kept Every QCI class has an integer identifier, for
regardless of the amount of low-priority example QCI1 or QCI2, and is mapped to a set of
background data traffic in the network [3, 4]. QoS metrics such as an indicator of priority of data
The next step will be to test our testbed setup traffic, an upper ceiling for network latency and, in
for the prioritized video stream in the presence some cases, guaranteed bit rate. In our case,
of the network load due to infotainment-type QCIs are instantiated with domain concepts for
background traffic. real-time vehicle traffic domain concepts.
For example, QCI3 is instantiated as “vehicle_
How it works control_traffic” and QCI4 is instantiated as
A cloud-hosted application function (AF) “vehicle_video_traffic.” For users browsing their
dynamically sets up virtual connections between mobile devices in the vehicles, we instantiate a
vehicles and the 5G Evolved Packet Core (EPC) low-priority class QCI9 as vehicle_web_browsing.”
network, with specific QoS attributes, such as Data traffic descriptors are generic contents that
designated latency levels and guaranteed throughput. can configure each QCI. The configuration

JAN UARY 26, 2018 ✱ ERICSSON TECHNOLOGY REVIEW 5

✱ 5G NETWORK PROGRAMMABILITY

Request to prioritize vehicle data traffic

Ericsson Cloud
Network traffic prioritization

Scania
command
Data traffic center
prioritization interface Remote
driver

(5G) core
network

Internet
Ericsson test
Scania bus network at Scania
test track Control plane
User plane Prioritized video traffic
User plane Prioritized control traffic
User plane Non-prioritized traffic
(e.g. mobile broadband, infotainment)

Figure 1 Prioritized data streams to meet QoS demands

pertains to a characterization of the traffic in terms concept document, while each specific enterprise
of required throughput for both “uplink” and has more than one document. For example, in our
“downlink,” the former being data traffic case, there is an “automotive/teleoperation”
transmitted from the agent and the latter being the document. However, other documents for
opposite. Optionally, descriptors may also contain automotive can also exist, such as “automotive/
the type of data packets exchanged (for example, autonomous drive” or “automotive/remote fleet
UDP/IP or TCP/IP), as well as potentially the port management.” Because the data is stored as linked
or port range. For example, in the case of data, concepts from one domain concept document
“vehicle_control_traffic,” the data traffic can be reused in another.
descriptor identifies an uplink bandwidth
of 1Mbps and a downlink bandwidth of 1Kbps. API endpoint module
A combination of agents, QCIs and their This module composes API specifications from
associated data traffic descriptors are stored every domain concept document in the knowledge
in the knowledge base as a domain concept base. This API specification is RESTful, uses
document. Every use case has its own domain symmetric encryption (HTTPS) and can be called

6 ERICSSON TECHNOLOGY REVIEW ✱ JAN UARY 26, 2018

 5G NETWORK PROGRAMMABILITY ✱

HSS
S6a

MME
S1-MME S1-MME

eNB eNB
S1-U S1-U
SGW

Uu Uu S5/S8 Uu Uu

UE UE PGW Gx SAPC UE UE
EPC

Rx

Transformer
Knowledge
base
Domain Generic API
semantics call

API endpoint

Domain
knowledge
definition
AF
Ericsson
research
cloud QoS setup

Third party

Figure 2 Architecture for programmable QoS in existing an LTE EPC network

from any third party. These API calls get translated specified in 3GPP TS 29.214. The Rx requests are
into generic concept calls that are subsequently sent directly to the PCRF node in order to set up
sent to the transformer module. Note that, in the “EPS bearer” (in other words, the data tunnel
addition to an API call for setup of specialized QoS, with the requested QoS). As is the case with the
there is another API call for teardown of this QoS. endpoint module, the transformer module can also
For example, when a vehicle is decommissioned translate a teardown request to an Rx request to
or does not need to be teleoperated, there can be revert to the lowest-priority default bearer
a call to tear down the QoS tunnel, so network (in most cases, QCI9).
resources can be allocated to UEs in other vehicles
or devices. Figure 3 provides an overview of
domain-specific and generic requests.
CONCEPTS FROM
ONE DOMAIN CONCEPT
Transformer module
The transformer module translates generic requests
DOCUMENT CAN BE REUSED
for QoS to Rx AAR requests, as these requests are IN ANOTHER

JAN UARY 26, 2018 ✱ ERICSSON TECHNOLOGY REVIEW 7

✱ 5G NETWORK PROGRAMMABILITY

Domain specific request Generic request Description of the request

GET /vehicle GET /agent Retrieve QoS information

for all UEs

GET /vehicle/<IP> GET /agent/<IP> Retrieve QoS information

for one UE, based on its IP
address

GET /qos GET /qos Retrieve QoS for all UEs

GET /qos/vehicle_control_traffic GET /qos/QCI3 Retrieve all UEs with QCI3

bearer setup

POST /qos/ POST /qos Set QoS for UE with IP src_

{ { IP and port src_port toward
“source_IP”:<src_IP>, “source_IP”:<src_IP>, destination with IP dst_IP
“source_port”:<src_port>, “source_port”:<src_port>, and port dst_port. Protocol
“destination_IP”:<dst_IP>, “destination_IP”:<dst_IP>, in this example is TCP but it
“destination_port”:<dst_port>, “destination_port”:<dst_port> can also be UDP.
“qos_class”:”vehicle_control_ “qos_class”:”QCI3”,
traffic”, “protocol”:”TCP”,
“protocol”:”TCP”, “max-requested-bandwidth-
“type”:”vehicle_video_stream” UL”:”1024,
} “max-requested-bandwidth-
DL”:”100”
}

DELETE /vehicle DELETE /agent Remove QoS for UE

{ { with given source and
“source_IP”:<src_IP>, “source_IP”:<src_IP>, destination IP and port
“source_port”:<src_port>, “source_port”:<src_port>,
“destination_IP”:<dst_IP>, “destination_IP”:<dst_IP>,
“destination_port”:<dst_port>, “destination_port”:<dst_port>,
“protocol”:”TCP” “protocol”:”TCP”
} }

Figure 3 Overview of requests

8 ERICSSON TECHNOLOGY REVIEW ✱ JAN UARY 26, 2018

 5G NETWORK PROGRAMMABILITY ✱

100.000

50.7720697 45.764087 47.333316

Color interpretation
<= 0.08% packet loss: unnoticeable in video
10.000
>0.08 – 0.5%: ghosting effect
5.086253 0.5% – 1%: artificial movement/dropped frames
1% – 5%: long pauses
1.3418724 5%+ impossible to follow
1.000

0.100
Percentage of packets lost

0.049715 0.045638 0.053639 0.057156

0.042681
0.048628 0.045809 0.050838 0.049232 0.046540 0.051253 0.046033 0.044295

0.010

2Mbps 3Mbps 6Mbps 2Mbps 3Mbps 6Mbps 2Mbps 3Mbps 6Mbps 2Mbps 3Mbps 6Mbps 2Mbps 3Mbps 6Mbps 2Mbps 3Mbps 6Mbps
video video video video video video video video video video video video video video video video video video
0.001
No background traffic Some background traffic Extreme background traffic No background traffic Some background traffic Extreme background traffic
(0 Mbps) (4.2Mbps) (8.55Mbps) (0 Mbps) (4.2Mbps) (8.55Mbps)
QCI9 QCI5

Figure 4 Packet loss (in percentage of total packets) in best effort (QCIŁ9) and prioritized (QCI5) bearers

Testbed results We measured both throughput and one-way

To assess QoS, we performed experiments on the network delay under these traffic conditions.
uplink prioritized video stream using QCI5 in the We also measured the ratio of packets lost
presence of the network load due to infotainment- versus packets sent to test the throughput
type background traffic using QCI9. The total quality of the network for three different
measured available bandwidth on the network qualities of video streams:
was approximately 8.55Mbps. We tested several
network load scenarios and measured the results 〉〉 excellent (6Mbps or 70 percent of the available bandwidth)
against three background traffic conditions: 〉〉 good (3Mbps or 35 percent of the available bandwidth)
〉〉 borderline drivable (2Mbps or 23 percent of the available
〉〉 none (0Mbps) bandwidth)
〉〉 some (4.2Mbps or 49 percent of the available bandwidth)
〉〉 extreme (8.55Mbps or 100 percent of the available Borderline drivable is the minimum requirement to
bandwidth) perform teleoperation. We obtained the packet

JAN UARY 26, 2018 ✱ ERICSSON TECHNOLOGY REVIEW 9

✱ 5G NETWORK PROGRAMMABILITY

drop requirements from empirical observations Conclusion

during test driving. We took a total of 160 5G attributes such as network slicing and low
measurements for each experiment and plotted latency will soon make mission-critical use cases
the graphs based on the respective average value. such as safe, autonomous public transport a reality.
Measurements from the 5G-network testbed Automated network resource prioritization via a
show that resource prioritization can assure programmable API can support network QoS for
predefined QoS levels for mission-critical diverse use cases with different connectivity
applications, regardless of background traffic. requirements on the cellular network. By developing
Figure 4 illustrates guaranteed uplink packet loss an API that allows a third party to request network
for a critical application, in which the acceptable resources and implementing it on a test mobile
packet loss of less than or equal to 0.08 percent is network, we have demonstrated how the technology
unnoticeable in the video stream. This is true even works in an urban transport-related use case with
with extreme background traffic when the system is Scania. The initial results show that throughput and
congested – the critical traffic is still served with no latency are maintained for high-priority streams
performance degradation. regardless of the network load.
However, as Figure 4 also illustrates, for the
non-prioritized infotainment traffic (QCI9) the
packet loss increases heavily with the increase Further reading
in the background traffic, introducing long pauses 〉〉 YouTube, Remote bus driving over 5G,
and making teleoperation impossible even for a November 2016 : https://www.youtube.com/
lower level of congestion. watch?v=lPyzGTD5FtM
When we measured the uplink delay, we found 〉〉 Ericsson Research blog, 5G teleoperated
that it is preserved (remaining at less than 34ms) vehicles for future public transport, June
for the critical video traffic even when the system 8, 2017, Berggren, V; Fersman, E; Inam, R;
exhibits congestion. For the non-prioritized traffic, Karapantelakis, A; Mokrushin, L; Schrammar,
the delay reaches up to 600ms during congestion. N; Vulgarakis, A; Wang, K : https://www.
The next step is to develop the concept for a ericsson.com/research-blog/5g/5g-teleoperated-
self-service portal where network customers vehicles-future-public-transport/
could specify QoS requirements on their own terms; 〉〉 Ericsson Mobility Report, June 2017 :
for example, to prioritize 4K video traffic for https://www.ericsson.com/assets/local/mobility-
40 buses in an urban scenario. The software would report/documents/2017/ericsson-mobility-report-
then translate this specification into instructions june-2017.pdf
for network resource prioritization.

10 ERICSSON TECHNOLOGY REVIEW ✱ JAN UARY 26, 2018

 5G NETWORK PROGRAMMABILITY ✱

Rafia Inam management for 5G Leonid Mokrushin

the authors ◆ joined Ericsson Research networks”, won a best ◆ is a senior specialist
in 2015. She works as a paper award in 2015. in the area of cognitive
senior researcher in the technologies. His current
area of machine intelligence focus is on investigating
and automation, and her what technological
research interests include opportunities artificial
5G management, service intelligence may bring to
modeling, virtualization Ericsson by creating and
of resources, reusability prototyping innovative
of real-time software and concepts in the context Elena Fersman
ITSs. Inam received her M.S. of new industrial and ◆ is head of machine
from Chalmers University of telco use cases. He joined intelligence and automation
Technology in Gothenburg, Ericsson Research in 2007 at Ericsson Research and
Sweden, in 2010. She Athanasios after postgraduate studies an adjunct professor in
received her Licentiate Karapantelakis at Uppsala University in cyber-physical systems
and doctoral degrees from ◆ joined Ericsson in Sweden. He holds an M.Sc. at KTH Royal Institute of
Mälardalen University in 2007 and currently works in software engineering Technology in Stockholm.
Västerås, Sweden in 2012 as a master research from Peter the Great St. She holds a Ph.D. in
engineer in the area of computer science from
machine intelligence and Uppsala University in
automation. He holds a B.Sc. Sweden and did post-
in computer science from doctoral research at École
the University of Crete in normale supérieure Paris-
The authors
Greece and an M.Sc. and Saclay, France, before
would like to
Licentiate of Engineering starting her industrial
acknowledge
in communication systems career. Her current research
the work of
from KTH Royal Institute of interests are in the areas of
Keven (Qi) Wang
on this project Technology in Stockholm, modeling and analysis and of
during his stay at and 2014 respectively. Her Sweden. His background is software- and knowledge-
Ericsson. paper, “Towards automated in software engineering. Petersburg Polytechnic intensive intelligent systems
service-oriented lifecycle University. applied to 5G and IoT.

References
1. IEEE, International Conference on Intelligent Transportation Systems (November 2016) – Feasibility
Assessment to Realise Vehicle Teleoperation using Cellular Networks – Rafia Inam, Nicolas Schrammar, Keven
Wang, Athanasios Karapantelakis, Leonid Mokrushin, Aneta Vulgarakis Feljan and Elena Fersman, available at:
http://ieeexplore.ieee.org/document/7795920/
2. IEEE, International Conference on Future Internet of Things and Cloud (August 2016) – DevOps for IoT
Applications Using Cellular Networks and Cloud – Athanasios Karapantelakis, Hongxin Liang, Keven Wang,
Konstantinos Vandikas, Rafia Inam, Elena Fersman, Ignacio Mulas-Viela, Nicolas Seyvet and Vasileios
Giannokostas, available at: http://ieeexplore.ieee.org/document/7575883/
3. Ericsson Mobility Report, Improving Public Transport with 5G, November 2015, available at: https://www.
ericsson.com/res/docs/2015/mobility-report/emr-nov-2015-improving-public-transport-with-5g.pdf
4. IEEE, Conference on Emerging Technologies and Factory Automation (September 2015) – Towards automated
service-oriented lifecycle management for 5G networks (Best Paper) – Rafia Inam, Athanasios Karapantelakis,
Konstantinos Vandikas, Leonid Mokrushin, Aneta Vulgarakis Feljan, and Elena Fersman, available at: http://
ieeexplore.ieee.org/document/7301660/

JAN UARY 26, 2018 ✱ ERICSSON TECHNOLOGY REVIEW 11

✱ 5G NETWORK PROGRAMMABILITY

ISSN 0014-0171
284 23-3306 | Uen

© Ericsson AB 2018 Ericsson

SE-164 83 Stockholm, Sweden
Phone: +46 10 719 0000

12 ERICSSON TECHNOLOGY REVIEW ✱ JAN UARY 26, 2018