5G Network Slicing for Mission-critical use cases

Full text

5G Network Slicing for Mission-critical use cases
Mark Roddy
Dept. Computer Science
Cork Institute of Technology
Cork, Ireland
Mark.Roddy@cit.ie
Dr. Mustafa Al Bado
Dell EMC Research Europe
Emerging Technologies and
Ecosystems, Office of the CTO
Cork, Ireland
Mustafa.AlBado@dell.com
Dr. Thuy Truong
Dell EMC Research Europe
Emerging Technologies and
Ecosystems, Office of the CTO
Cork, Ireland
Thuy.Truong@dell.com
Yanxin Wu
Dept. Computer Science
Cork Institute of Technology
Cork, Ireland
jasonpandaboy@gmail.com
Sean Ahearne
Dell EMC Research Europe
Emerging Technologies and
Ecosystems, Office of the CTO
Cork, Ireland
Sean.Ahearne@dell.com
Prof. Paul Walsh
Dept. Computer Science
Cork Institute of Technology
Cork, Ireland
Paul.Walsh@cit.ie
Michael Healy
Dept. Computer Science
Cork Institute of Technology
Cork, Ireland
michael.healy2@mycit.ie
Abstract—The demand for prehospital emergency care has
increased during the last decades throughout the Western
world, in terms of numbers of emergency calls and dispatched
ambulances. This development represents a challenge for both
the prehospital emergency systems and the emergency
departments at the hospitals [1]. Stroke is the fourth single
leading cause of death in the UK and an accurate recognition of
stroke by in-ambulance or emergency medical services (EMS)
or prehospital ambulance paramedics, offers significant
potential to reduce delays in presentation and treatment in acute
stroke [2]. This paper demonstrates Proof-of-Concept (PoC)
approaches for 5G network slicing in mission-critical use cases
from the H2020 5G PPP SliceNet project, which is implementing
an End-to-End (E2E) cognitive network slicing and slice
management framework in virtualised multi-domain, multi-
tenant 5G Networks. The paper shows how the PoC’s key
enablers, such as QoS-aware network slicing, edge computing
and hardware acceleration, can assist with a continuous
collection, processing and streaming of patient data that could
shorten the time to assess and provide optimal clinical treatment
pathways for potential stroke patients.
Keywords—5G, proof-of-concept (PoC), quality of service
(QoS), quality of experience (QoE), network slicing, mission-
critical services (MCX), eHealth, connected ambulances, artificial
intelligence (AI), machine learning (ML), end-to-end (E2E),
mobile edge computing (MEC), hardware acceleration, Digital
Service Provider (DSP), Network Service Provider (NSP),
Business Verticals, Service Level Agreement (SLA), Radio Access
Network (RAN).
I. INTRODUCTION
Legacy public safety and mission-critical communication
systems (e.g. ETSI’s TETRA) have been designed primarily
as purpose-built mobile radio networks for the delivery of
mission-critical voice, as well as a select number of narrow-
band data services (e.g. text-based messaging). They have
been costly to design, deploy and service, and without a
fundamental redesign, will not be able to deliver and exploit
the media-rich type services currently accessible over public
broadband networks.
At the same time, with early prototype deployments well
underway, next generation 5G networks are expected to start
commercial launches across the world by 2020, working
alongside existing 3G and 4G technology to provide faster
speeds and more reliable connections than ever before.[3]
Innovative solutions, such as Network Slicing, will enable
network operators to provide highly secure dedicated virtual
networks, such as mission-critical networks, to specific
vertical customers over the 5G network infrastructure.[4]
Combining Network Slicing and the very latest research
innovations, future 5G networks will offer connections that
are significantly faster than current connections and will help
power innovative use cases for a smarter and more connected
world.
In addition, Mobile edge computing (MEC) technologies
will bring application intensive processing closer to the data
source, thereby reducing latency for real-time applications.
Gartner [5] expects a steady increase in the embedding of
sensor, storage, compute and advanced artificial intelligent
(AI)/machine learning (ML) capabilities in edge devices.
Intelligence will move towards the edge in a variety of
endpoint devices, from industrial devices, to screens, to
smartphones, and to automobile power generators.
This paper forms part of the research from the H2020 5G
PPP SliceNet project [6], where three key use cases act as
showcase test-bed demonstrators: The Smart Grid Use Case;
The eHealth Use Case; and The Smart City Use Case.
The eHealth Use Case approach, presented here, proposes
a generalised 5G ambulance telemedicine system that
supports MEC-based artificial intelligence (AI) and machine
learning (ML) to provide paramedics and hospital clinicians
with rapid prehospital in-ambulance diagnostics that could
enhance and improve patient treatment pathways.
II. BACKGROUND RESEARCH
A. Background to Mission-critical Emergency
Communications
This section provides a brief overview of public
emergency communications systems and some background to
mission-critical services (otherwise known as MCX).
The world’s first formal policing service was introduced
in 1829 onto the streets of the United Kingdom by Sir Robert
Peel and by the end of the 19th century, inventions such as the
telegraph, the telephone, two-way radio systems, offered new
forms of communications that the emergency services (police

and fire) would quickly embrace to increase service
effectiveness and efficiency.[7]
The 999/112 emergency telephone service was introduced
in 1937 and initially was only available to a small number of
exchanges around central London; it would not be available
across the entire UK until 1976.
Around 1,300 callers used the 999 service in the first week
of launch, compared to the nearly half a million per week in
2017 [8].
Today the public safety agencies worldwide rely on
private land mobile radio (LMR) systems (e.g. Tetra radio)
deployed in dedicated, narrowband spectrum to support
mission critical voice services provided by personnel in
Emergency Dispatch Centres to emergency response
personnel in the field.
With the rollout of ultra-high bandwidth and ultra-low
latency 5G broadband technologies, coupled with the
allocation of new public safety (broadband) spectrum in some
countries, public safety agencies are deploying mission
critical broadband systems as an overlay to existing
narrowband systems and for replacement of commercial
mobile data services. While voice over broadband was
initially expected to act as a back up to the legacy LMR voice
service, representatives of public safety agencies have
suggested that broadband networks ultimately replace
narrowband LMR networks. [9] For example, in 2015 the UK
Home Office awarded a £1.2Bn Emergency Services Network
(ESN) contract to network operator EE and Motorola
Solutions to migrate from the dedicated UK TETRA network
to a public safety Secure-MVNO, over a public broadband
LTE network. But there have been significant problems - the
original migration date from tetra was due to start in 2017 but
rollout delays have caused the UK Home Office to rethink
their strategy, and the original tetra-based network is not now
due to be phased out until at least 2022 and by the summer of
2109 the budget had escalated to over £3Bn.
B. State-of-the-Art for in-ambulance connected health
Summary of an in-ambulance connected health trial in
Scotland - The health of rural people is generally similar to
the health of urban people in most of the developed world but
due to location, rural people are disadvantaged in time-critical
medical and surgical emergencies. In response to inadequate
mobile broadband coverage in most parts of rural highland
Scotland, in 2018 the University of Aberdeen launched a trial
to study the delivery of in-ambulance ultrasound over satellite.
The aim of the trial was the delivery of better quality in-
ambulance emergency care to remote rural locations. The trial
used satellite communications and point-of care ultrasound to
help expedite care delivery in difficult to reach rural areas of
the highlands of Scotland.
From inside the ambulance, point-of-care high-definition
ultrasound images of patients were relayed back to the
Emergency Department consultants via satellite. Consultants
then texted back their care advice to the paramedics inside the
ambulance.
Over the duration of the trial, around 1000 adult patients
with trauma, breathlessness, chest pain, abdominal pain,
shock, were transported to Raigmore Hospital, Inverness, by
the Scottish Ambulance Service.
In the absence of adequate mobile broadband coverage,
satellite communications were used as a positive intervention
to remotely support point-of-care ultrasound and clinical
advice in real time.
The trial concluded that remotely supported broadband
services, such as relay of portable high-definition ultrasound
images and real time expert clinical support has the potential
to streamline care in time-critical emergencies in rural and
hard to reach locations [10].
Summary of an in-ambulance Telestroke Assessment
Service in Belgium - In recent years significant progress has
been made through in-hospital stroke management, but a
telemedicine solution that optimizes prehospital in-ambulance
stroke diagnostics has not yet been widely adopted. Solving
this problem could speed up treatment initiation by early
activation of the in-hospital stroke response, thereby curtailing
the risk of misdiagnosis, reduce the proportion of missed
opportunities for treatment with intravenous thrombolysis
and/or endovascular treatment, and avoid patient admission to
inadequate clinical facilities[6].
The social impact of stroke related illness is stark and
worldwide stroke is the number one cause of acquired
disability, which leads significantly towards dementia and
death.
Stroke is a sudden disturbance in brain circulation, either
through an arterial blockage or bleed. Blockages account for
just over 85% of cases and prompt treatment within the first 3
hours of onset considerably reduces disability and improves
patient outcome. However, only a small percentage of patients
with blockage get definitive care within the recommended 3
hour window [11].
Use of a remote teleconsultant to assist the ambulance
personnel to correctly manage the stroke patient’s treatment
pathway is therefore desirable. Teleconsultant heuristics are
required to efficiently engage in the complex patient–doctor
interactions, and to make appropriate medical decisions under
time pressure. For these reasons, stroke diagnosis and patient
selection by paramedics is not optimal, and real-time
intervention by clinical experts is necessary. Not only does
this offer more appropriate care for patients, but also offers
substantial savings in terms of reduced responses to
ambulance services.
This study helps progress the ambulance service from
being purely a transport provider, to being a key provider of
care in a patient’s clinical pathway, and one further example
of this is was a Prehospital Stroke Study, discussed next.
A Prehospital Stroke Study [12] in Belgium, developed
a mobile broadband bi-directional video consulting system,
whereby for every patient with suspicion of acute stroke, a
stroke expert located at the hospital, was brought virtually
inside the ambulance to effectively provide an innovative
prehospital treatment pathway, which included: remote
patient triage; prehospital preparation of the in-hospital team;
collection of key patient data (e.g. patient identity and
demographics, vital parameters, clinical presentation, medical
history, premorbid functional state, concomitant medication);
assessment of stroke severity (based on an in-hospital stroke
severity scale, which had been adapted for in-ambulance
assessment); and identification of patient for specific in-
hospital stroke treatment (e.g. thrombolytics, endovascular

approaches, aggressive blood pressure lowering,
neurosurgical interventions).
The study concluded that, subject to clinical validation,
this connected health approach could halve the delay from
stroke onset to initiation of specific stroke therapy.
However, a key problem remained that adequate and
resilient mobile broadband coverage could not be provided,
which ultimately led to intermittent delivery of this potentially
life-saving service.
SliceNet Connected Health Use Case Study - As
evidenced from the previous two studies in this section, in-
ambulance telemedicine is a promising approach to improving
health related emergency care but reliable mobile broadband
is proving to be a deterrent to widespread service adoption.
The European funded SliceNet project has built on these
two studies and aims to facilitate widespread use of in-
ambulance Telestroke diagnostics and improved patient
treatment pathways, through the design of a 5G network
slicing framework, as this could prove to be a solution for the
delivery of reliable real time audio-video communication
from inside high-speed moving ambulances.
The Prehospital Stroke Study referenced, had
implemented a system in which a teleconsultant could guide a
patient through a protocol known as the Unassisted Telestroke
Severity Scale (UTSS) [12]. The scale required that patients
perform a series of body movements and verbally answer
questions or repeat phrases, which are then analysed by a
remote clinical teleconsultant.
A potential to extend the state-of-the-art arises when using
5G network slicing and edge computing to develop a machine-
learning application that can automate the UTSS.
The SliceNet Telestroke Assessment application took a
selection of these UTSS protocols, and applied machine
learning algorithms; for example, “Please spread the fingers
of your right hand as far apart as you can”. The goal of the
analysis here was to see if the patient was capable of
separating the fingers on each hand. Using machine vision, the
following prototype screenshots (Fig.s 1, 2 and 3)
demonstrated how this step could be automated.
Figure 1: Example input: Fingers spread
Figure 2: Example input: fingers not spread
Figure 3: Telestroke output
Partial automation of the UTSS could be used as a pre-
teleconsultant step when an initial indication of ‘stroke’ or ‘no
stroke’ is provided, and then used to assist in a decision
whether further diagnosis is needed from a teleconsultant. It
might be beneficial in a situation where no teleconsultant is
available, and a full automation of the UTSS could provide
life-saving time-gains and essential pre-hospital diagnostics
needed in emergency situations.
In order to achieve acceptable QoE and QoS in a 5G
eHealth scenario, recommended latency (time to travel the
network from the sensor to the end user) is 30ms to 100ms
latency end-to-end [13].
III. SLICENET APPROACH
This section will discuss the SliceNet business model,
which shows the vertical and digital/network service provider
perspectives and how SliceNet can meet MCX requirements.
Fig. 4 shows a SliceNet business model, as applied in an
eHealth use case. The end-to-end (E2E) slice customer is a
national/regional health service organization (e.g., National
Ambulance Service in Ireland), which operates multiple
(static) hospitals, dispatch centres, and (moving) ambulances.
The E2E eHealth slice offered to the customer by the Digital
Service Provider (DSP) consists initially of a “base” Network
Slice Instance (NSI) containing the minimal set of network
functions & services. The base slice is fairly static and is
centered around a geographical area in the vicinity of the
hospital. The hospital/dispatch hosts experts who provide real-
time support to the paramedics. As ambulances are
dispatched, additional Network Sub-Slice Instances (NSSIs)
may be instantiated in order to increase the geographical
coverage of the slice (e.g., by adding RAN) or to guarantee the
latency and availability requirements of the slice. For the
latter, additional processing functions may be dynamically
instantiated at suitable MEC locations. Dispatch may trigger a
handover of a paramedic’s communication stream to a
different hospital. Handover between domains might be
needed while the ambulance is moving.

Figure 4: eHealth Business Model from a vertical/service provider
perspective
Vertical and DSP - The verticals only see services
provided by their DSP. In the scope of SliceNet, the DSP will
provide to the vertical E2E eHealth services: i) E2E Telestroke
Assessment service and ii) E2E Video Relay service.
The Telestroke Assessment service consists of the
Gateway application running in the ambulance, capturing the
video of the patient, and the Telestroke Assessment
application running at the edge, collecting the data from the
Gateway and running a ML algorithm to analyse the images
for the stroke condition. The assessment application sends the
analysed result back to the in-ambulance paramedic and to the
hospital for a clinical consultant to examine.
The Video Relay service is initiated with the Video Relay
glasses worn by a paramedic in the ambulance capturing a live
video of the scene. The glasses connect to the Video Relay
Application running at the edge for authentication and to set
up the communication with the hospital to provide a direct link
from the Video Relay glasses to the hospital where the doctors
can view video of the scene.
The DSP will add into its catalogue the Telestroke
Assessment service descriptor and the Video Relay service
descriptor, and then advertises to the vertical these two
services. However, to specify what the DSP can deliver to the
vertical, there are two scenarios from the DSP role:
• The DSP has established partnerships with a set of
Network Service Providers (NSPs) where it has agreed
what each NSP can offer, and based on existing
partnership agreements, the DSP will advertise the
services which are also associated with a set of offering
service assurances to the verticals.
• When the DSP has new customers, or new requirements
from existing customers, and the requirements cannot be
delivered with current DSP resources, the DSP will have
to negotiate new terms with the NSPs or it will have to
find new NSPs/new partnerships to meet these new
requirements.
For the eHealth use case, the requirements from the
vertical are specified in a Service Level Agreement (SLA),
e.g., E2E eHealth slice with minimal bandwidth bw,
maximum latency l, reliability r%, coverage c%, and secure
communication.
DSP and NSP - Within an intra-domain network, the NSP
will provide the network services that it guarantees to the DSP
in the SLA between this NSP and DSP. For example, the DSP
in Fig.4 will create an NSI from NSP A that consists of RAN
NSSI, MEC NSSI and Core NSSI and allocate the resources
for these instances in a way that meets the QoS requirements,
while the NSP B with an NSI of RAN NSSI and Core NSSI,
and resources will be used for inter-domain if needed.
Monitoring and optimisation will also be running in the
domains to maintain service assurance.
SliceNet key innovations - As a public safety sliced
service takes priority over all other network traffic (e.g.,
industries 4.0, smart city, ad hoc access), it is crucial to
guarantee the SLA for the service, e.g., availability, delay,
bandwidth, coverage, security, etc. SliceNet is meeting these
requirements with the approaches below:
• One-stop API towards the vertical, with Plug and Play
(PnP) functionalities for service monitoring,
reconfiguring and autoscaling.
• Cross-domain, cross-plane orchestration to provide
dynamic slicing, dynamic reconfiguration based on
priority level.
• Cognitive, agile QoE management of slices for service
assurance of vertical business.
• E2E slice FCAPS management to manage fault,
configuration, accounting, performance and security of
all slices across multiple planes and network operators
domains.
More details of the SliceNet approach can be found at the
project website (slicenet.eu).
IV. TESTBED INFRASTRUCTURE
This section will discuss a high-level description of the
eHealth use case and the infrastructure built for that use case.
Fig.5 shows the eHealth use case with the Video Relay and
Telestroke Assessment services running, and the ambulance
connected to the hospital over the 5G network.
Fig.6 shows the eHealth infrastructure at Dell premises in
Ovens, Ireland. On top of the infrastructure, the open source
security software, pfSense, is running to provide different
security services including the firewall, NAT rules,
rule/policy-based traffic control and OpenVPN server with
TLS protocol.
Figure 5. eHealth UC high level description
System access through web portal/one-stop API will run
an OpenVPN client software and authorised with credentials
and certificate to open connections with OpenVPN server
configured and running at pfSense server. After this
authentication phase, e.g., the user is successfully initiating
the sequence, OpenVPN allocates an IP address to this user to
enable the user in the tunnel, accessing the LAN network. The
configuration in the OpenVPN server allows which LAN
network the user is tunnelled into, and with firewall rules,
pfSense controls which servers, VMs, services in that LAN

network that the user can have access to. For eHealth, the VPN
connections are configured to have AES-256-CBC/SHA1 for
cryptography and 2048 bits parameter length Diffie-Hellman
for key exchange, remote access with SSL/TLS protocol.
Figure 6: eHealth UC high level description
Below the firewall and security services in pfSense, the
infrastructure is spanning across 3 racks for different
purposes: i) management rack with 3xR510 PowerEdge
servers and 2xR610 (one for primary controller node and the
other for secondary controller node to enable the HA feature),
the servers in this rack are running management services such
as OSM (release FIVE), MAAS and Juju, the rack will be
dedicated for other management components/services that
will be integrated in future; ii) core/enterprise cloud rack has
an R430 server running oai-CN components (MME, HSS,
SPGW) and 4xR640 servers are configured as cloud operating
system, running VIO VMware (abstracting the VMware
vSphere with NSX environment to have the outbound
interfaces as Openstack API). The rack is dedicated to run core
components/services, the VIO cloud is also for demonstration
purpose where the services running here (as core/enterprise
cloud services) should have longer delay compared to the
services running at the edge; and iii) the edge rack with
4xR640 servers running Openstack. This edge VIM is hosting
the two eHealth services (Telestroke Assessment and Video
Relay above). LL-MEC and FlexRan controller are also
deployed in this VIM for RAN virtualisation and traffic
control. The two racks of MP and core/enterprise are located
in CIX where the edge rack is located in Dell premises which
is 15km away in geography and these are linked by MPLS
connection with both PE devices enabling VLPS to have L2
encapsulation from the two sites.
Finally, a stand-alone eNB is composed by Dell Precision
5000 that is running oai-RAN software and is attached with
USRP B210 radio. To have access to this eNB, open-cells
SIMs are reprogrammed with uicc/sim programming software
provided by open-cells. One of the SIMs is then inserted into
the Dell Edge Gateway series 3003 with LTE modem built-in.
This device with the SIM card is attached with an Intel
Realsense camera for UE device that is running as the gateway
in the ambulance, capturing the images of the patient for
Telestroke assessment. Another SIM is then used for Video
Relay device which starts to stream the video of surroundings
to a normal device, e.g., doctor’s laptop, connected at the
hospital site to see what has happened at the ambulance site.
V. EXPERIMENTAL RESULTS AND KEY FINDINGS
The experiment is to examine the scenario of possible
stroke patients at a remote location. With OpenMANO OSM,
the experiment shows the seamless composition and
onboarding of the eHealth slice onto Openstack VIM. More
importantly, it shows how edge computing with hardware
acceleration can assist with a continuous collection,
processing and streaming of patient data that shortens the time
to assess potential stroke patients. Experimental results
showing the performance of traditional Telestroke
Assessment service running at the cloud and the performance
of the service running at the edge with accelerated hardware.
Figure 7: eHealth Slice template
Figure 8: eHealth slice design, onboarding and instantiation and its state
transition
Fig.7 and Fig.8 show the service orchestration with i) an
eHealth slice designed and onboarded onto the orchestrator
and ii) the slice instantiated and resources allocated into a
specific VIM (VMware VIO or Openstack). After these steps,
the eHealth service is up and running. As the study focussed
only on the E2E runtime performance of the Telestroke
Assessment application, no consideration was given to the
setup time of the E2E runtime performance, nor the
breakdown of latency for each element.
Fig.9 shows the performance of service running at the
cloud/core and when it is running at the edge with hardware
acceleration. Fig.9.a) represents the Round-Trip Time (RTT)
for all packets sent and received within the service. This RTT
highlights the time taken between transmission of a packet
containing data for the service, and the service responding to
the client that the data has been received. RTT performance
from the core is not optimal for a real-time service such as this,
with several high latency spikes and a high amount of packet
loss. RTT performance at the edge is much improved in
comparison to the core. Latency stabilizes shortly after service
instantiation, with minimal packet loss. The average RTT
latency from client to core was 296.91 milliseconds. The
average RTT from client to edge was 50.68 milliseconds, a
5.86x performance improvement. Average packet loss was
7.2% for the core and 0.1% at the edge, a 7.7x improvement.
Fig.9.b) represents the total number of frames processed
by the service per second. A WebSocket packet is created each
time a new frame is transmitted from the client to the service.
When the service receives the entire frame, it sends another
WebSocket packet in acknowledgement. When the service
has finished processing that frame for facial recognition or
stroke detection, it sends the response to the client and
requests that a new frame is sent. Thus, it can be inferred that
the frame rate of this application is (number of WebSocket

packets/2). The performance of the core as shown in the red
graph is poor, with 1 frame taking several seconds to process.
This performance can be attributed to high packet loss
requiring frame data to be re-transmitted, and lack of hardware
acceleration. The hardware accelerated edge performance can
be seen in the green graph, with a frame rate an order of
magnitude greater than the core. The eHealth application
changes from facial recognition to stroke detection at
approximately the halfway point in both core and edge tests.
a) Latency/RTT
b) Throughput/Frames per second
Figure 9: Performance of Telestroke Assessment running at the cloud
(left) and edge (right)
VI. CONCLUSION
This paper explores Proof-of-Concepts for 5G network
slicing approaches to mission-critical use cases in the SliceNet
project. It examines an experimental scenario for remote in-
ambulance stroke assessment. The experiment shows the
seamless composition and onboarding for a 5G connected
health slice, and how edge computing with hardware
acceleration can assist with a continuous collection,
processing and streaming of patient data that shortens the time
to assess a patient suspected of having a stroke. The
computational workload involves real-time AI/ML image
processing, which is both network and compute intensive, and
can be addressed by 5G technology enablers, such as QoS-
aware network slicing and edge computing.
SliceNet will reach a maximum Technology Readiness
Level 4/5 but full validation will require further field trial and
real user evaluation. Clinical validation will be much longer
in time frame, and there is a complex and intricate stakeholder
roadmap required here to realise any future towards a viable
Telestroke assessment deployment.
ACKNOWLEDGMENT
This work has been funded in part through the European
Union’s H2020 program, under grant agreement No 761913:
project SliceNet. The authors would like to thank all SliceNet
partners for their support in this work.
REFERENCES
[1] Christensen EF, Bendtsen MD, Larsen TM, Jensen FB, Lindskou TA,
Holdgaard HO, Hansen PA, Johnsen SP, Christiansen CF. Trends in
diagnostic patterns and mortality in emergency ambulance service
patients in 2007− 2014: a population-based cohort study from the North
Denmark Region. BMJ open. 2017 Aug 1;7(8):e014508.
[2] Harbison J, Hossain O, Jenkinson D, Davis J, Louw SJ, Ford GA.
Diagnostic accuracy of stroke referrals from primary care, emergency
room physicians, and ambulance staff using the face arm speech test.
Stroke. 2003 Jan 1;34(1):71-6.
[3] https://www.facebook.com/TechRadar (2019) What is 5G? Everything
you need to know, TechRadar. Available at:
https://www.techradar.com/news/what-is-5g-everything-you-need-to-
know (Accessed: February 28, 2019).
[4] What is Network Slicing? (2018) 5g.co.uk. Available at:
https://5g.co.uk/guides/what-is-network-slicing/ (Accessed: February
28, 2019).
[5] Panetta, K. (2018) Smarter With Gartner, Gartner.com. Available at:
https://www.gartner.com/smarterwithgartner/gartner-top-10-strategic-
technology-trends-for-2019/ (Accessed: February 28, 2019).
[6] SliceNet Consortium “End-to-End Cognitive Network Slicing and
Slice Management Framework in Virtualised Multi-Domain, Multi-
Tenant 5G Networks”, H2020 5G PPP https://slicenet.eu/ [Accessed 29
May 2019].
[7] R. Stewart, "The police signal box: a 100-year history", Engineering
Science & Education Journal, vol. 3, no. 4, pp. 161-168, 1994.
Available: 10.1049/esej:19940407.
[8] Andrew Roberts (2017) ‘Emergency services at your fingertips ’, Daily
Telegraph (London), p. 16,17. Available at:
http://search.ebscohost.com.cit.idm.oclc.org/login.aspx?direct=true&
db=ukh&AN=8Q2127062888&site=ehost-live (Accessed: 8 May
2019).
[9] K. Budka, T. Chu, T. Doumi, W. Brouwer, P. Lamoureux and M.
Palamara, "Public safety mission critical voice services over LTE",
Bell Labs Technical Journal, vol. 16, no. 3, pp. 133-149, 2011.
Available: 10.1002/bltj.20526 [Accessed 8 May 2019].
[10] L. Eadie (2018) “SatCare: Remote Support for Ambulance Clinicians
in Medical Emergencies”. Available:
https://clinicaltrials.gov/ct2/show/NCT03323229 [Accessed 27 May
2019].
[11] E. McGrath, M. Canavan, M. O'Donnell, “Stroke” Elsevier,
Haematology (7th Ed) 2018, Pages 2133-2141
https://doi.org/10.1016/B978-0-323-35762-3.00145-1 [Accessed 27
May 2019].
[12] Valenzuela Espinoza, A. et al. (2016) “Development and Pilot Testing
of 24/7 In-Ambulance Telemedicine for Acute Stroke: Prehospital
Stroke Study at the Universitair Ziekenhuis Brussel-Project,”
Cerebrovascular Diseases, 42(1–2), pp. 15–22. doi:
10.1159/000444175.
[13] SliceNet Consortium “Vertical Sector Requirements Analysis and Use
Case Definition”, H2020 5G PPP https://slicenet.eu/ [Accessed 20 July
2019].