<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>5G revolution and the Italian use cases</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <string-name>Alessandro Vizzarri</string-name>
          <xref ref-type="aff" rid="aff1">1</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Francesco Bove</string-name>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <aff id="aff0">
          <label>0</label>
          <institution>Independent Consultant</institution>
          ,
          <addr-line>83059 Vallata, Avellino</addr-line>
          ,
          <country country="IT">Italy</country>
        </aff>
        <aff id="aff1">
          <label>1</label>
          <institution>Radiolabs</institution>
          ,
          <addr-line>Consorzio Università Industria</addr-line>
        </aff>
      </contrib-group>
      <fpage>52</fpage>
      <lpage>58</lpage>
      <abstract>
        <p>The 5G is a breakthrough transformation for fixed and mobile networks. It enables high performances more than the current networks. For this reason, 5G is one of the next cornerstones of digitalization, from the industries' competitiveness and improvement in communication and standards of living point of view. This paper provides a technical analysis of 5G technology, characterizing it as the technological boost to innovate all communication networks. 5G technology is a research, innovation and standardization product that interests the entire industrial world and communities. The network is evolving to manage both broadband connection and very low latency, critical communications. We will assist to an increasing number of hyper-connected devices. Therefore, the approach chosen to build this infrastructure for Gigabit Society is to create new industrial partnerships, signing agreements between technology and academic leaders.</p>
      </abstract>
      <kwd-group>
        <kwd>5G</kwd>
      </kwd-group>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>1. Introduction</title>
      <sec id="sec-1-1">
        <title>In the last years Telecommunications have become mo</title>
        <p>re and more innovative. They led to a very disruptive
technological change in the society. The data transfer
from users has been characterized by a rapid growth
and considerable increase, beyond any prediction. In
fact, in 1995 the GSM reached 10 kbit/s per channel.</p>
        <p>
          Today, with 4G we can reach 100 Mbit/s per
channel [
          <xref ref-type="bibr" rid="ref1">1</xref>
          ].
        </p>
        <p>
          From the managing point of view, the presence of
many digital devices implies powerful and eficient
communication networks. They have to enable high
capacity in order to guarantee low latency during the
data transfer, limited energy consumption and stable Figure 1: 5G use cases envisaged by ETSI [
          <xref ref-type="bibr" rid="ref2">2</xref>
          ].
connectivity. For this reason, the development of 5G
have to be encouraged. As ETSI (European
Telecommunications Standards Institute) states [
          <xref ref-type="bibr" rid="ref2">2</xref>
          ], "5G should tute. The deployment of 5G system foresees to be
dedeliver significantly increased operational performance, ployed in many diferent scenarios. They include
ulofering full mobility and coverage. 5G needs to cater tra dense access points and smart objects belonging to
ofering acceptable levels of energy consumption, equip- the Internet of Things ecosystem [
          <xref ref-type="bibr" rid="ref3">3</xref>
          ]. Moreover, the
ment cost and network deployment, and operation cost. 5G system is able to integrate diferent radio access
It needs to support a wide variety of applications and technologies, including the satellite components [
          <xref ref-type="bibr" rid="ref4 ref5">4, 5</xref>
          ].
services". From the architecture point of view, the first One of the key elements of 5G system is the Network
release of 5G is similar to 4G. It is composed of ra- Virtualization Function (NVF) feature. Authors in [
          <xref ref-type="bibr" rid="ref6">6</xref>
          ]
dio access network and network core, Evolved Packet presented an overview of some 5G initiatives to
exCore and Internet Multimedia System [
          <xref ref-type="bibr" rid="ref2">2</xref>
          ]. In Figure 1 ploit NFV approach. It can facilitate the 5G difusion
it is reported the use cases defined by the ETSI insti- and its integration with terrestrial and satellite system.
From a market point of view, the integration of
sevSYSTEM 2020: Symposium for Young Scientists in Technology, eral communication technologies enabled by the 5G
Engineering and Mathematics, Online, May 20 2020 system brings to new business opportunities in the
en" alessandro.vizzarri@radiolabs.it (A. Vizzarri); bove9@hotmail.it tire Information and Communication (ICT) sector. The
(F. Bove) supplier companies in communication services,
communications services, cloud and software can benefit
© 2020 Copyright for this paper by its authors. Use permitted under Creative
CPWrEooUrckReshdoinpgs IhStpN:/c1e6u1r3-w-0s.o7r3g CCoEmUmoRns WLiceonrsekAsthtriobuptioPnr4o.0cIneteerdnaitniognasl ((CCC EBYU4R.0)-.WS.org)
hardware involved in 5G systems. The IoT nodes need
to optimize data processing in order to reduce energy
consumption and extend battery life. For this reason,
5G will impact also on microprocessors’ design. As
known in the literature, specific or re-configurable
hardware accelerators represent a good solution to
speedup data processing and in some cases also to reduce the
power consumption as shown for example in [
          <xref ref-type="bibr" rid="ref10 ref8 ref9">8, 9, 10</xref>
          ].
        </p>
        <p>
          In fact, Qualcomm pushes on the introduction of
new smartphones supporting Android operating
system and each equipped by 2 chipsets. The new SoC is
characterized by a powerful processor and the
Snapdragon X55 modem. It allows to manage 2 platforms,
Figure 2: Qualcomm Snapdragon 855. one for the 4G and the other for the 5G. Moreover, it
updates the diferent components. In a specific article
by Digital Day, the main processor’s features are
defrom 5G system deployment. On the other side, in gen- scribed: "The block diagram shows how the processor is
eral they can increase the growth of telecommunica- divided: the CPU now occupies a relatively small portion,
tion sector in terms of eficiency, business and user ex- comparing it to the three true processors that handle all
perience. In fact, by the end of 2025, the 5G is expected the usual operations: the DSP hexagon 698 designed to
to reach 2.6 billion subscriptions and then to cover up speed up machine learning calculations, the GPU Adreno
to 65% of the world’s population. It means that 5G will 650 and the Image signal Processor Spectra 480. Together
be able to generate the 45% of total mobile data trafic with the sensing Hub, that is to say with the “safe zone”,
worldwide. Thus makes faster the adoption of 5G on which stores sensitive data in temporary memory, and
a global scale [
          <xref ref-type="bibr" rid="ref7">7</xref>
          ]. with the module that handles sensor data, these
procesThe paper is organized as follow. In Section 2 the hard- sors make up what is today the most advanced SoC in
ware enabling the 5G is described. In Section 3 the the world".
5G experimentation carried in Italy is presented. The Considering the extremely high number of
applicaSection 4 provides a brief list of main communication tions supported by the 5G, Embedded microprocessors
vendor companies pf 5G equipment. The Section 5 the shall be able to perform several operations as for
eximpact of 5G technology in the connected vehicles is ample speech quality enhancement [
          <xref ref-type="bibr" rid="ref11">11</xref>
          ] and machine
analyzed. The Section 6 summarizes the main conclu- learning
sions. In particular for what concerns Machine Learning,
In the last few years, we assisted in its incredible
growing up [
          <xref ref-type="bibr" rid="ref12 ref13 ref14 ref15 ref16 ref17 ref18 ref19 ref20 ref21 ref22 ref23">12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23</xref>
          ].
2. Hardware This has been possible for three main reasons
        </p>
      </sec>
      <sec id="sec-1-2">
        <title>The general standardisation process of a radio mobile</title>
        <p>technology foresees diferent steps.</p>
        <p>One of the most important regards the assessment
of radio interface technologies, proposed by
independent evaluation groups. Qualcomm, an American
research and development company in the wireless
telecommunications field founded in 1985, in the first half
of 2018 proposed several 5G trials with some
American operators, AT&amp;T and Verizon. The Qualcomm
President, Cristiano Amon, announced the
availability of Snapdragon 855 (the dedicated 5G chipset for
high-end smartphones) in the first half of 2019. The
products with Snapdragon 855 are able to connect to
the 5G network (see fig. 2).</p>
        <p>The new applications supported by the 5G require
high level of computation capabilities supported by the</p>
      </sec>
      <sec id="sec-1-3">
        <title>1. The big availability of data introduced by the in</title>
        <p>
          ternet
2. The increase of computational power of digital
circuits that allowed the possibility to eficiently
implement neural networks and more in general
ML systems [
          <xref ref-type="bibr" rid="ref24 ref25">24, 25</xref>
          ].
3. The introduction of new Machine Learning
models that allows emulating the brain-behavior in
a very eficient way [26, 27].
        </p>
      </sec>
      <sec id="sec-1-4">
        <title>A new processor, dedicated to machine learning and</title>
        <p>to models’ acceleration, is called Hexagon 689. The
NPU analog component from Huawei is a module that
gets to 15 trillion operations per second. It reached
with an eficiency that is 35% higher than that of the
Snapdragon 855 Hexagon.</p>
        <p>The Hexagon, however, works in a diferent way
compared to NPU. Thus because it can ask for help
from GPU or CPU based on the diferent kind of
neural network or model that it has to handle. The new
CPU is 25% faster and more eficient than the
previous one. it consists of 8 cores divided into 3 clusters.
One of them from a single core Cortex a77 at 2.85 GHz
and with a 512 KB second level cache. The other three
which work at 2.4 GHz. All the cores have a 256 KB
cache. The remaining 4 cores are deputed for
background tasks and for applications that do not require
an intensive use of CPU work at 1.8 GHz, Cortex A55
4. Each of them have a 128KB cache.</p>
        <p>Each core shares a 4 MB cache. The self-organizing
networks (SON) should take to fewer drops of signal,
higher data speed and lower costs for operators,
adapting dynamically to diferent environments. Regression
techniques are evaluated. In fact estimating route loss
has been analyzed by researchers when they adopted
the machine learning techniques in order to obtain
more eficient and accurate route loss models. It has been
demonstrated that the machine learning use provides
adaptability to network designers that rely on signal
propagation models.</p>
        <p>Diferent vendor companies launched new devices
compatible with the 5G. Among the first ones, Oppo
Reno 5G, talked about Snapdragon X50, joined by SoC
Qualcomm. The chipset characteristics are the same
of the previous version. A second device has been
launched by Samsung, the Galaxy S10 5G. It represents
a special variant, with a design similar to the other
models except for the 5G support sided by modem SoC
Exynos 9820. A bigger display comes thanks to a
Dynamic Amoled panel of 6,7”. The real revolution comes
with Nokia. It started with 5G network adopting a
dedicated device, a mid-rage smartphone. The new Nokia
8.2 5G, in addiction to a reduced price, is equipped by
a waterfall display, a selfie retractable camera and a
Snapdragon 735 5G processor.</p>
      </sec>
    </sec>
    <sec id="sec-2">
      <title>3. Experimentation in Italy</title>
      <sec id="sec-2-1">
        <title>In Italy several experimentations have been carried in</title>
        <p>the last years. In particular, TIM and Ericsson
developed a 5G antenna with millimeter waves (mmW) and
performed several preliminary test in Turin (see fig. 3).</p>
        <p>In 2017 the Italian Ministry of Economic
Development (MISE) launched the first 5G experimentation to
be held in 5 italian cities and in collaboration with the
operators (TIM, Vodafone, Wind Tre) and vendors
(Ericsson, Nokia, Huawei and ZTE).</p>
        <p>In Turin, the project sees the realization of a Smart</p>
      </sec>
      <sec id="sec-2-2">
        <title>City Control Room. It enables several important appli</title>
        <p>cations supported by 5G, as smart trafic, smart
parking and smart lighting. A download datarate of 20
GBbps was reached. The value is 20 times faster than
the 4G. According to Tim CEO’s expectations, the 5G
network is going to involve more and more users over
the years, changing radically the habits, perspectives
and potential. One of the most promising aspect is
represented by virtual reality. It allows to live senses and
emotions from hundreds of kilometers away. Not only
calls and safe navigation, fast and stable, but also the
best use of artistic and cultural heritage, through
designed applications and created for valorize stripes of
territory unknown by tourists, improvement of the
education condition, healthcare more connected with the
possibility to do also small remote surgeries.</p>
        <p>The project, in the south of Italy between Bari and
Matera, is branded Tim, Fastweb and Huawei, with 52
partners involved.</p>
        <p>With the first experiments, in 2018, the 75% of
network coverage of cities has been reached. The high
radio performances have been reached through the use
of “Massive MIMO” antenna, characterized by active
project is experimented in some cities in Tuscany. The
project aim is to digitalize the harbour operations and
to make them more eficient and sustainable using the
5G network.</p>
      </sec>
    </sec>
    <sec id="sec-3">
      <title>4. Vendors in the world</title>
      <sec id="sec-3-1">
        <title>In 2019 Nokia became the leader 5G vendor in North</title>
        <p>Figure 4: 5G antenna installed in Turin by Tim/Ericsson America. As confirmed by Nokia Bells Labs President,
Marcus Weldon: "It is to allow new geographical areas
to experiment for the first time the possibilities of the
elements that enable the coordinated use of an inde- 5G, to push on the technology and redefine the
limpendent number of radiant elements. the resulting ef- its and to ofer excellent performances reaching new
fect allows to increase antenna directivity and to dy- records in the field, highlighting the first transmission
namically orient the transmission of the radio signal data end-to-end on a commercial network 5G New
Rain the direction of the various customers. In this way, dio 3GPP. Nokia has managed to reach 23 records
reeveryone could use the best performance that the an- lated to 5G in multiple technological areas”.
tenna is able to ofer thanks to techniques for spatial The inventors of patents, launched by Nokia, work
distribution of radio resources. Vodafone carried out at Near Bell Labs in Aalborg, in Denmark: “In Nokia
the 5G experimentation in Milan, in collaboration with it has been long understood the importance of open
Nokia. They defined 41 Use Cases and 28 partners. standardization, aimed at enabling others to create,
baAt the end of 2019, the network coverage has reached sed on our innovation, new products for people around
100% of the Lombard metropolitan area. With a total the world. Through licensing, other sectors and
acof 90 Million euros for total investment, several impor- tivities can benefit from Nokia’s innovation, while we
tant applications were tested, as healthcare and well- seek to develop an even more connected world”, said
ness, security and surveillance, mobility and transports, Jenni Lukander, Nokia Technologies President. Nokia
manufacture and industry 4.0. The healthcare sector is the only end-to-end products supplier, that covers all
can benefit a lot from 5G. Thanks to a broadband con- 5G network elements. Nokia 5G products were
connection and high speed data rates ofered by 5G, it tracted by all US national operators, all South Korea
can be possible to guarantee high reliability for new operators and all the three Japanese national
operaambulance intervention protocols avoiding delays and tors. That is to say two-thirds of the global business of
the repetition of examinations. The real time manage- access networks in a typical year.
ment of critical situations during transport can also be China is the most powerful mobile market with 1
enable, together with the remote consultation. The Billion connections from smartphones reached at the
possibility to exchange the trafic information com- end of 2018. 5G will represent a real turning point: it
ing from widespread sensors deployed along the road is the reference point for Asian Countries of this new
infrastructure and vehicles can both enable the smart technology in the world. China worked on 5G in
colmobility and enrich the driving assistance system. Fast- laboration with his three national operators. In July,
web and Ericsson carried out the 5G experimentation despite the USA persistence, Huawei has claimed to
in Rome in order to create a digital laboratory useful have signed 91 more commercial agreements for 5G
for the touristic sector. They developed new applica- network supply all over the world, of which at least
tions in the urban mobility and security, in terms of 28 of them being in Europe. While the USA accuse
innovative high-definition surveillance solutions. The Huawei of telecommunications espionage, the Chinese
development for a new experimentation has been taken giant signed the agreement for the 5G network’s
dein agreement from Fastweb and Samsung for 5G Fixed velopment in Russia. The agreement has been signed
Wireless Access Network (FWA) on commercial fre- close to a meeting between the Russian leader Vladimir
quencies, always on the Italian territory. Putin and the Chinese leader, Xi Jinping.</p>
        <p>This project shows that the 5G FWA networks are
able to replace the FTTH network (filbert-to-the-home)
to realize Gbps-based connections. In this way, the
5G FWA technical and commercial feasibility on
millimetre bands has been proved. The European Corealis
Virtual reality
FWA
e-health
Sensors networks
Smart city
Connected automotive
Cloud robotics
Cloud virtual e augmented reality
Smart manufacturing</p>
        <p>Connected drones</p>
        <p>TIM, connected vehicles experimentation
A powerful management of the people and goods
mobility needs to assure firstly their safety, secondly, the
reduction of wasted time, money losses, and the
environmental impact. It can ofer various points of
evolution for the telephone operator’s business.</p>
        <p>Transport systems are evolving towards connected,
cooperating and automated mobility paradigms CCAM
(Connected, Cooperative and Automated Mobility). In
this scenario the vehicle becomes an integral part of
a connected system that includes intelligent road
infrastructures, distributed sensors, private and public
controls. The network is able to exchange data and
• Network level. The system intelligence should
be moved On this level. The MEC functionalities
can be deployed in order to reduce the round trip
time of the request/response of the connected</p>
      </sec>
    </sec>
    <sec id="sec-4">
      <title>6. Conclusion</title>
      <sec id="sec-4-1">
        <title>In the last years the major telecommunications leaders</title>
        <p>have demonstrated their interest on the 5G
development, not only from a new network standard point of
view. 5G is an important enabler of new applications
and services to ofer to the end users. In the whole
ICT sector, the companies have been forced to change
in order to catch the right opportunities coming from
more challenging the user requirements. In order to
do this, we need an more decentralized network
infrastructure, able to support the elaboration and
storage closer to the end user. It also implies the
compliance with regulations and the necessity to guarantee
the network resilience and scalability. In this paper,
the authors provides an overview of the
telecommunication sector considering the last and the future trends
regarding the 5G. After the analysis of hardware
supporting the 5G, the paper presents the main important
5G experimentation carried out in Italy, together with
the 5G applications for the smart mobility. As the next
steps, the authors will focus on the performance
analysis and comparative analysis with the previous
communication standards.</p>
      </sec>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          [1]
          <string-name>
            <given-names>N.</given-names>
            <surname>Alliance</surname>
          </string-name>
          ,
          <article-title>Ngmn 5g initiative white paper, final deliverable (approved), 5G Initiative Team (</article-title>
          <year>2015</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref2">
        <mixed-citation>
          [2]
          <string-name>
            <surname>ETSI</surname>
          </string-name>
          ,
          <source>Why do we need 5G?</source>
          ,
          <year>2015</year>
          . URL: https:// www.etsi.org/technologies/mobile/5g.
        </mixed-citation>
      </ref>
      <ref id="ref3">
        <mixed-citation>
          [3]
          <string-name>
            <given-names>F.</given-names>
            <surname>Mazzenga</surname>
          </string-name>
          ,
          <string-name>
            <given-names>R.</given-names>
            <surname>Giuliano</surname>
          </string-name>
          ,
          <string-name>
            <given-names>F.</given-names>
            <surname>Vatalaro</surname>
          </string-name>
          ,
          <article-title>Fttc-based fronthaul for 5g dense/ultra-dense access network: Performance and costs in realistic scenarios</article-title>
          ,
          <source>Future Internet</source>
          <volume>9</volume>
          (
          <year>2017</year>
          )
          <fpage>71</fpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref4">
        <mixed-citation>
          [4]
          <string-name>
            <given-names>C.</given-names>
            <surname>Sacchi</surname>
          </string-name>
          ,
          <string-name>
            <given-names>T.</given-names>
            <surname>Rossi</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M.</given-names>
            <surname>Murroni</surname>
          </string-name>
          ,
          <string-name>
            <surname>M.</surname>
          </string-name>
          <article-title>Ruggieri, Extremely high frequency (ehf) bands for future broadcast satellite services: Opportunities and challenges</article-title>
          ,
          <source>IEEE Transactions on Broadcasting</source>
          <volume>65</volume>
          (
          <year>2019</year>
          )
          <fpage>609</fpage>
          -
          <lpage>626</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref5">
        <mixed-citation>
          [5]
          <string-name>
            <given-names>R.</given-names>
            <surname>Giuliano</surname>
          </string-name>
          ,
          <string-name>
            <given-names>F.</given-names>
            <surname>Mazzenga</surname>
          </string-name>
          ,
          <string-name>
            <given-names>A.</given-names>
            <surname>Vizzarri</surname>
          </string-name>
          ,
          <article-title>Satellitebased capillary 5g-mmtc networks for environmental applications</article-title>
          ,
          <source>IEEE Aerospace and Electronic Systems Magazine</source>
          <volume>34</volume>
          (
          <year>2019</year>
          )
          <fpage>40</fpage>
          -
          <lpage>48</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref6">
        <mixed-citation>
          [6]
          <string-name>
            <given-names>G.</given-names>
            <surname>Giambene</surname>
          </string-name>
          ,
          <string-name>
            <given-names>S.</given-names>
            <surname>Kota</surname>
          </string-name>
          ,
          <string-name>
            <given-names>P.</given-names>
            <surname>Pillai</surname>
          </string-name>
          ,
          <article-title>Satellite-5g integration: A network perspective</article-title>
          ,
          <source>IEEE Network 32</source>
          (
          <year>2018</year>
          )
          <fpage>25</fpage>
          -
          <lpage>31</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref7">
        <mixed-citation>
          [7]
          <string-name>
            <surname>Ericsson</surname>
            ,
            <given-names>Mobility</given-names>
          </string-name>
          <string-name>
            <surname>Report</surname>
          </string-name>
          , Ericsson,
          <year>November 2019</year>
          ,
          <year>2019</year>
          . URL: https://www.ericsson.com/en/ mobility-report/reports/november/
          <year>2019</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref8">
        <mixed-citation>
          [8]
          <string-name>
            <given-names>G. C.</given-names>
            <surname>Cardarilli</surname>
          </string-name>
          ,
          <string-name>
            <given-names>L. Di</given-names>
            <surname>Nunzio</surname>
          </string-name>
          ,
          <string-name>
            <given-names>R.</given-names>
            <surname>Fazzolari</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M.</given-names>
            <surname>Re</surname>
          </string-name>
          ,
          <string-name>
            <given-names>F.</given-names>
            <surname>Silvestri</surname>
          </string-name>
          ,
          <string-name>
            <given-names>S.</given-names>
            <surname>Spanò</surname>
          </string-name>
          ,
          <article-title>Energy consumption saving in embedded microprocessors using hardware accelerators</article-title>
          ,
          <source>Telkomnika</source>
          <volume>16</volume>
          (
          <year>2018</year>
          )
          <fpage>1019</fpage>
          -
          <lpage>1026</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref9">
        <mixed-citation>
          [9]
          <string-name>
            <given-names>S.</given-names>
            <surname>Paul</surname>
          </string-name>
          ,
          <string-name>
            <given-names>R.</given-names>
            <surname>Karam</surname>
          </string-name>
          ,
          <string-name>
            <given-names>S.</given-names>
            <surname>Bhunia</surname>
          </string-name>
          ,
          <string-name>
            <given-names>R.</given-names>
            <surname>Puri</surname>
          </string-name>
          ,
          <article-title>Energyeficient hardware acceleration through computing in the memory</article-title>
          , in: 2014 Design, Automation &amp; Test in Europe Conference &amp;
          <string-name>
            <surname>Exhibition (DATE),</surname>
            <given-names>IEEE</given-names>
          </string-name>
          ,
          <year>2014</year>
          , pp.
          <fpage>1</fpage>
          -
          <lpage>6</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref10">
        <mixed-citation>
          <source>[10] INTEL, AN 531: Reducing Power with Hardware Accelerators</source>
          ,
          <year>2008</year>
          . URL: https://www.intel.com/.
        </mixed-citation>
      </ref>
      <ref id="ref11">
        <mixed-citation>
          [11]
          <string-name>
            <given-names>G.</given-names>
            <surname>Capizzi</surname>
          </string-name>
          ,
          <string-name>
            <given-names>S.</given-names>
            <surname>Coco</surname>
          </string-name>
          , G. Sciuto,
          <string-name>
            <given-names>C.</given-names>
            <surname>Napoli</surname>
          </string-name>
          ,
          <article-title>A new iterative fir filter design approach using a gaussian approximation</article-title>
          ,
          <source>IEEE Signal Processing Letters</source>
          <volume>25</volume>
          (
          <year>2018</year>
          )
          <fpage>1615</fpage>
          -
          <lpage>1619</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref12">
        <mixed-citation>
          [12]
          <string-name>
            <given-names>G.</given-names>
            <surname>Capizzi</surname>
          </string-name>
          ,
          <string-name>
            <given-names>G. Lo</given-names>
            <surname>Sciuto</surname>
          </string-name>
          ,
          <string-name>
            <given-names>C.</given-names>
            <surname>Napoli</surname>
          </string-name>
          ,
          <string-name>
            <given-names>R.</given-names>
            <surname>Shikler</surname>
          </string-name>
          ,
          <string-name>
            <surname>M.</surname>
          </string-name>
          <article-title>Woźniak, Optimizing the organic solar cell manufacturing process by means of afm measurements and neural networks</article-title>
          ,
          <source>Energies</source>
          <volume>11</volume>
          (
          <year>2018</year>
          )
          <fpage>1221</fpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref13">
        <mixed-citation>
          [13]
          <string-name>
            <given-names>G.</given-names>
            <surname>Capizzi</surname>
          </string-name>
          ,
          <string-name>
            <given-names>G. Lo</given-names>
            <surname>Sciuto</surname>
          </string-name>
          ,
          <string-name>
            <given-names>C.</given-names>
            <surname>Napoli</surname>
          </string-name>
          ,
          <string-name>
            <given-names>E.</given-names>
            <surname>Tramontana</surname>
          </string-name>
          ,
          <article-title>A multithread nested neural network architecture to model surface plasmon polaritons propagation</article-title>
          , volume
          <volume>7</volume>
          ,
          <year>2016</year>
          . doi:
          <volume>10</volume>
          .3390/ mi7070110.
        </mixed-citation>
      </ref>
      <ref id="ref14">
        <mixed-citation>
          [14]
          <string-name>
            <given-names>F.</given-names>
            <surname>Beritelli</surname>
          </string-name>
          ,
          <string-name>
            <given-names>A.</given-names>
            <surname>Gallotta</surname>
          </string-name>
          ,
          <string-name>
            <given-names>C.</given-names>
            <surname>Rametta</surname>
          </string-name>
          ,
          <article-title>A dual streaming approach for speech quality enhancement of voip service over 3g networks, in: 2013 Wavelet recurrent neural network with semi18th International Conference on Digital Signal parametric input data preprocessing for microProcessing (DSP)</article-title>
          , IEEE,
          <year>2013</year>
          , pp.
          <fpage>1</fpage>
          -
          <lpage>5</lpage>
          .
          <article-title>wind power forecasting in integrated generation</article-title>
        </mixed-citation>
      </ref>
      <ref id="ref15">
        <mixed-citation>
          [15]
          <string-name>
            <given-names>C.</given-names>
            <surname>Napoli</surname>
          </string-name>
          ,
          <string-name>
            <given-names>F.</given-names>
            <surname>Bonanno</surname>
          </string-name>
          , G. Capizzi,
          <article-title>Exploiting systems</article-title>
          , in: 2015 International Conference on
          <article-title>solar wind time series correlation with magne- Clean Electrical Power (ICCEP)</article-title>
          , IEEE,
          <year>2015</year>
          , pp.
          <article-title>tospheric response by using an hybrid neuro- 602-609</article-title>
          . wavelet approach,
          <source>Proceedings of the Interna</source>
          <volume>-</volume>
          [26]
          <string-name>
            <given-names>G.</given-names>
            <surname>Susi</surname>
          </string-name>
          ,
          <string-name>
            <given-names>A.</given-names>
            <surname>Cristini</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M.</given-names>
            <surname>Salerno</surname>
          </string-name>
          ,
          <source>Path multimodaltional Astronomical Union</source>
          <volume>6</volume>
          (
          <year>2010</year>
          )
          <fpage>156</fpage>
          -
          <lpage>158</lpage>
          .
          <article-title>ity in a feedforward snn module, using lif with</article-title>
        </mixed-citation>
      </ref>
      <ref id="ref16">
        <mixed-citation>
          [16]
          <string-name>
            <given-names>F.</given-names>
            <surname>Beritelli</surname>
          </string-name>
          , G. Capizzi,
          <string-name>
            <given-names>G. L.</given-names>
            <surname>Sciuto</surname>
          </string-name>
          , C. Napoli, latency model,
          <source>Neural Network World</source>
          <volume>26</volume>
          (
          <year>2016</year>
          )
          <string-name>
            <given-names>F.</given-names>
            <surname>Scaglione</surname>
          </string-name>
          ,
          <article-title>Rainfall estimation based on the in- 363. tensity of the received signal in a lte/4g mobile</article-title>
          [27]
          <string-name>
            <given-names>A.</given-names>
            <surname>Cristini</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M.</given-names>
            <surname>Salerno</surname>
          </string-name>
          ,
          <string-name>
            <given-names>G.</given-names>
            <surname>Susi</surname>
          </string-name>
          ,
          <article-title>A continuousterminal by using a probabilistic neural network, time spiking neural network paradigm</article-title>
          ,
          <source>in: AdIEEE Access</source>
          <volume>6</volume>
          (
          <year>2018</year>
          )
          <fpage>30865</fpage>
          -
          <lpage>30873</lpage>
          . vances in Neural Networks: Computational and
        </mixed-citation>
      </ref>
      <ref id="ref17">
        <mixed-citation>
          [17]
          <string-name>
            <given-names>R.</given-names>
            <surname>Giuliano</surname>
          </string-name>
          ,
          <string-name>
            <given-names>G. C.</given-names>
            <surname>Cardarilli</surname>
          </string-name>
          ,
          <string-name>
            <given-names>C.</given-names>
            <surname>Cesarini</surname>
          </string-name>
          , Theoretical Issues, Springer,
          <year>2015</year>
          , pp.
          <fpage>49</fpage>
          -
          <lpage>60</lpage>
          . L. Di
          <string-name>
            <surname>Nunzio</surname>
            ,
            <given-names>F.</given-names>
          </string-name>
          <string-name>
            <surname>Fallucchi</surname>
            ,
            <given-names>R.</given-names>
          </string-name>
          <string-name>
            <surname>Fazzolari</surname>
            , [28]
            <given-names>I. Benedetti</given-names>
          </string-name>
          ,
          <string-name>
            <given-names>R.</given-names>
            <surname>Giuliano</surname>
          </string-name>
          ,
          <string-name>
            <given-names>C.</given-names>
            <surname>Lodovisi</surname>
          </string-name>
          ,
          <string-name>
            <given-names>F.</given-names>
            <surname>Mazzenga</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M.</given-names>
            <surname>Re</surname>
          </string-name>
          ,
          <string-name>
            <given-names>A.</given-names>
            <surname>Vizzarri</surname>
          </string-name>
          ,
          <string-name>
            <surname>Indoor lo- F. Mazzenga</surname>
          </string-name>
          ,
          <article-title>5g wireless dense access netcalization system based on bluetooth low energy work for automotive applications: Opportunities for museum applications</article-title>
          ,
          <source>Electronics</source>
          <volume>9</volume>
          (
          <year>2020</year>
          )
          <article-title>and costs</article-title>
          , in: 2017 International Conference 1055. of Electrical and Electronic Technologies for
        </mixed-citation>
      </ref>
      <ref id="ref18">
        <mixed-citation>
          [18]
          <string-name>
            <given-names>S.</given-names>
            <surname>Spanò</surname>
          </string-name>
          ,
          <string-name>
            <given-names>G. C.</given-names>
            <surname>Cardarilli</surname>
          </string-name>
          ,
          <string-name>
            <given-names>L. Di</given-names>
            <surname>Nunzio</surname>
          </string-name>
          , R. Fazzo- Automotive, IEEE,
          <year>2017</year>
          , pp.
          <fpage>1</fpage>
          -
          <lpage>6</lpage>
          . lari, D. Giardino,
          <string-name>
            <given-names>M.</given-names>
            <surname>Matta</surname>
          </string-name>
          ,
          <string-name>
            <given-names>A.</given-names>
            <surname>Nannarelli</surname>
          </string-name>
          , M. Re, [29]
          <string-name>
            <given-names>E.</given-names>
            <surname>Chiocchetti</surname>
          </string-name>
          ,
          <string-name>
            <given-names>F. G. G.</given-names>
            <surname>Larini</surname>
          </string-name>
          ,
          <string-name>
            <surname>G. Zafiro,</surname>
          </string-name>
          <article-title>An eficient hardware implementation of rein- Smart Mobility: come cambia il ruolo forcement learning: The q-learning algorithm, dell'operatore, Notiziario Tecnico TIM, N0</article-title>
          .
          <source>Ieee Access</source>
          <volume>7</volume>
          (
          <year>2019</year>
          )
          <fpage>186340</fpage>
          -
          <lpage>186351</lpage>
          . 3,
          <year>2018</year>
          . URL: https://www.telecomitalia.
        </mixed-citation>
      </ref>
      <ref id="ref19">
        <mixed-citation>
          [19]
          <string-name>
            <given-names>L. M.</given-names>
            <surname>Da Silva</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M. F.</given-names>
            <surname>Torquato</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M. A.</given-names>
            <surname>Fernandes</surname>
          </string-name>
          , com/tit/it/notiziariotecnico/edizioni
          <article-title>-2018/ Parallel implementation of reinforcement learn- n-3-2018/N1-Smart-Mobility-come-cambia\ ing q-learning technique for fpga, IEEE Access 7 -ruolo-dell-operatore</article-title>
          .html. (
          <year>2018</year>
          )
          <fpage>2782</fpage>
          -
          <lpage>2798</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref20">
        <mixed-citation>
          [20]
          <string-name>
            <given-names>P.</given-names>
            <surname>Chi</surname>
          </string-name>
          ,
          <string-name>
            <given-names>S.</given-names>
            <surname>Li</surname>
          </string-name>
          ,
          <string-name>
            <given-names>C.</given-names>
            <surname>Xu</surname>
          </string-name>
          ,
          <string-name>
            <given-names>T.</given-names>
            <surname>Zhang</surname>
          </string-name>
          ,
          <string-name>
            <given-names>J.</given-names>
            <surname>Zhao</surname>
          </string-name>
          ,
          <string-name>
            <given-names>Y.</given-names>
            <surname>Liu</surname>
          </string-name>
          ,
          <string-name>
            <given-names>Y.</given-names>
            <surname>Wang</surname>
          </string-name>
          ,
          <string-name>
            <given-names>Y.</given-names>
            <surname>Xie</surname>
          </string-name>
          ,
          <article-title>Prime: A novel processing-in-memory architecture for neural network computation in reram-based main memory</article-title>
          ,
          <source>ACM SIGARCH Computer Architecture News</source>
          <volume>44</volume>
          (
          <year>2016</year>
          )
          <fpage>27</fpage>
          -
          <lpage>39</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref21">
        <mixed-citation>
          [21]
          <string-name>
            <given-names>R.</given-names>
            <surname>Zhao</surname>
          </string-name>
          ,
          <string-name>
            <given-names>W.</given-names>
            <surname>Luk</surname>
          </string-name>
          ,
          <string-name>
            <given-names>X.</given-names>
            <surname>Niu</surname>
          </string-name>
          ,
          <string-name>
            <given-names>H.</given-names>
            <surname>Shi</surname>
          </string-name>
          ,
          <string-name>
            <given-names>H.</given-names>
            <surname>Wang</surname>
          </string-name>
          ,
          <article-title>Hardware acceleration for machine learning</article-title>
          ,
          <source>in: 2017 IEEE Computer Society Annual Symposium on VLSI (ISVLSI)</source>
          , IEEE,
          <year>2017</year>
          , pp.
          <fpage>645</fpage>
          -
          <lpage>650</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref22">
        <mixed-citation>
          [22]
          <string-name>
            <given-names>M.</given-names>
            <surname>Matta</surname>
          </string-name>
          ,
          <string-name>
            <given-names>G. C.</given-names>
            <surname>Cardarilli</surname>
          </string-name>
          ,
          <string-name>
            <given-names>L. Di</given-names>
            <surname>Nunzio</surname>
          </string-name>
          ,
          <string-name>
            <given-names>R.</given-names>
            <surname>Fazzolari</surname>
          </string-name>
          ,
          <string-name>
            <given-names>D.</given-names>
            <surname>Giardino</surname>
          </string-name>
          ,
          <string-name>
            <given-names>A.</given-names>
            <surname>Nannarelli</surname>
          </string-name>
          , M. Re,
          <string-name>
            <given-names>S.</given-names>
            <surname>Spano</surname>
          </string-name>
          ,
          <article-title>A reinforcement learning-based qam/psk symbol synchronizer</article-title>
          ,
          <source>Ieee Access</source>
          <volume>7</volume>
          (
          <year>2019</year>
          )
          <fpage>124147</fpage>
          -
          <lpage>124157</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref23">
        <mixed-citation>
          [23]
          <string-name>
            <given-names>M.</given-names>
            <surname>Zhang</surname>
          </string-name>
          ,
          <string-name>
            <given-names>Z.</given-names>
            <surname>Liu</surname>
          </string-name>
          ,
          <string-name>
            <given-names>L.</given-names>
            <surname>Li</surname>
          </string-name>
          ,
          <string-name>
            <given-names>H.</given-names>
            <surname>Wang</surname>
          </string-name>
          ,
          <article-title>Enhanced eficiency bpsk demodulator based on one-dimensional convolutional neural network</article-title>
          ,
          <source>IEEE Access 6</source>
          (
          <year>2018</year>
          )
          <fpage>26939</fpage>
          -
          <lpage>26948</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref24">
        <mixed-citation>
          [24]
          <string-name>
            <given-names>M.</given-names>
            <surname>Wózniak</surname>
          </string-name>
          ,
          <string-name>
            <given-names>D.</given-names>
            <surname>Połap</surname>
          </string-name>
          ,
          <string-name>
            <given-names>R. K.</given-names>
            <surname>Nowicki</surname>
          </string-name>
          ,
          <string-name>
            <given-names>C.</given-names>
            <surname>Napoli</surname>
          </string-name>
          ,
          <string-name>
            <given-names>G.</given-names>
            <surname>Pappalardo</surname>
          </string-name>
          , E. Tramontana,
          <article-title>Novel approach toward medical signals classifier</article-title>
          , in: 2015
          <source>International Joint Conference on Neural Networks (IJCNN)</source>
          , IEEE,
          <year>2015</year>
          , pp.
          <fpage>1</fpage>
          -
          <lpage>7</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref25">
        <mixed-citation>
          [25]
          <string-name>
            <given-names>F.</given-names>
            <surname>Bonanno</surname>
          </string-name>
          , G. Capizzi,
          <string-name>
            <given-names>G. L.</given-names>
            <surname>Sciuto</surname>
          </string-name>
          , C. Napoli,
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>