Instituto de Telecomunicações 4TELL Research Group

http://www.av.it.pt/4Tellhttp://www.it.pt

- 2 -

Representing 4TELL Today we have.....Name: Dr Shahid Mumtaz

Degrees:• PhD in Electrical & Electronic Engineering (Telecommunication), Portugal• MSc in Electrical Engineering (Telecommunication), Sweden

Experiences:• Research & industrial experience (Ericsson, Huawei, NSN, Portugal Telecommunication)

(> 10 years)

Publication:Conference, Journal, Magazine (> 60)Patent (2)Book chapters (>8)Books (3)

Editors: IEEE communication and wireless Magazine WorkshopsTutorial talksGeneral chair for conferenceSpeical issue in IEEE journal & Transactions

- 3 -

IT overview (I) Instituto de Telecomunicações (IT) is a private, Portuguese non-profit

organization, of public interest, a partnership of eight institutions with experience in R&D in the telecommunications field Portugal Telecom Inovação

Nokia Siemens Networks* Instituto Superior Técnico (IST);

Universidade de Aveiro (UA);

Universidade de Coimbra (UC);

Nokia Siemens Networks (NSN);

ISCTE - Instituto Universitário de Lisboa (ISCTE-IUL).

Universidade da Beira Interior (UBI);

Universidade do Porto (UP);

IT mission is to create and disseminate scientific knowledge in the field of telecommunications.

- 4 -

IT overview (II) IT is organised around four main sites

Aveiro – ITAV

Coimbra - UC

Lisbon - IST

Porto -PT

Approximately 200 researchers (PhD holders) in total (200 PhDs, 200 Msc Students)

Scientific expertise in IT, from which follow its main research and education activities, spans through Wireless Communications

Optical Communications

Networks and Multimedia

Basic Sciences and Enabling Technologies

- 5 -

IT- Aveiro 150 researchers (40 PhD’s)

Funding 1/3 national research agency

1/3 national industry

1/3 EU (significant involvement in FP7, FP6, ..)

IT Aveiro

- 6 -

4TELL Research Group 4TELL is a visionary and cutting-edge research group within the

“Wireless Communications Scientific Area” at the Instituto de Telecomunicações in Aveiro.

The competence of the group is focused on targeting innovative solutions in: 5G related Technologies (mmWave, Massive MIMO, HetNets, D2D…..)

Cognitive networks

Radio Resource Management

Security

Green Communications

Localization

Media Independent Handover and Mobility

- 7 -

4TELL Projects 4TELL European projects involvement

5G-SPEED (5GPP Project EU 20:20:20)

WHERE 2 (STREP)

ARTEMOS (ENIAC)

CRS-I (CSA) - coordinator

GREENET (MC-ITN)

C2POWER (STREP) - coordinator

COGEU (STREP) - coordinator

GREEN-T (EUREKA CELTIC) - coordinator

ROMEO (IP)

ACCUS (ARTEMIS )

SALUS (STEP) - coordinator

Project Portfolio includes ~ 3 M Euros Research Funding 4TELL Group : 12 post-docs; 8 Research Associates; 10 PhD Students

- 8 -

4TELL – Achievements Since inception in 2008

5 authored Books (5G networking and green communications)

300 scientific works

Successfully concluded as coordinator: ICT-C2POWER (Good) and ICT-COGEU (Excellent) April ´13

Head of RAS (Radio Access Cluster) – EU Cooperation program for FP7 projects

Active Member of ETSI standardization and Commonwealth-ITU

Conference Organization:

General Chair for Mobimedia conference (2010) General Chair for WiCOM conference (2014)

3 patents

Conducted first national TVWS trials in Portugal for rural internet coverage

- 9 -

5G-HETEROGENEOUS CLOUD RADIO ACCESS NETWORKS

Shahid Mumtaz

- 10 -

IntroductionWe examine the idea of Resource sharing in H-CRAN at specturm, infrastruture and network levels.

H-CRAN is combination of HetNets and C-RANs

From HetNets: H-CRANs inherit the tiered deployment of small and macro cells.

This increase spectrum reuse, resulting in significant gains in network capacity

From C-RAN: Centralizing processing enables optimized orchestration of the number of users per base station, energy consumption, and interference.

BBU, RRH

The H-CRAN will to go a step further in performance by incorporating cloud computing into HetNets to accomplish large-scale cooperative signal processing and network functionalities with increased network capacity.

Exploit advanced spatial signal processing techniques in the physical layer (PHY), such as centralized massive MIMO and distributed large-scale spatial cooperative processing

the cloudification of network functionalities enables H-CRANs to perform cooperative radio resource management and cooperative self-organizing networking to schedule and reorganize resources to supply the huge bit rate demand for ultra dense communication scenarios.

10

- 11 -

C-RAN L2L3

L1L2L3

L1L2IP

L1L2IP

L1L2IP

L1

SCTPX2-AP

L2IP

L1

UDPGTP-U

Ir Ir

RF RF

RRH 2RRH 1CPRI/OBSAI/OR1

SRC DUC CFR/DPD DAC

FF

SRC DUC

CPRI/OBSAI/OR1

SRC DUC CFR/DPD DAC

FF

SRC DUC

IQ DL

IQ ULIQ DL

IQ ULSyn

EPC

Backhaul

FronthaulFronthaul

Network Architecture

- 12 -

12

Challenges

Integrating two separate network architectures is a challenge.

The multi-tiered approach of HetNets increases the complexity of interference avoidance

Separation of functionality into RRHs and BBUs, by C-RANs, incurs the need for a higher-capacity backhaul to avoid delays and service degradation .

Fortunately, many of these challenges are offset by the dynamic resource sharing enabled by H-CRANs

Resource sharing in H-CRANs

spectrum

infrastructure

network

12

- 13 -

13

Challenges

Spectrum sharing

H-CRANs facilitate improved spectral efficiency through distributed multiantenna use, intertier interference mitigation, and dynamic spectrum access.

At the infrastructure level, physical entities (e.g. base stations, backhaul, and access points) can be shared among network operators.

At the network level, spectrum and infrastructure from an H-CRAN can be abstracted into network slices defined by higher-level metrics (e.g., throughput and processing).

At each of these levels, an H-CRAN enables resource sharing benefits including a dynamic pool of spectral resources, enhanced infrastructure coverage, and virtual networks tailored to particular service goals.

13

- 14 -

H-CRAN RESOURCE SHARING ANALYSISH-CRAN RESOURCE SHARING ANALYSIS

- 15 -

SIMULATION RESULTS

15

• To quantify the resulting efficiency gains (in bits per second per Hertz), we designed two experimental H-CRAN scenarios.

• For both scenarios, we placed a macrocell in the middle of a 1 km2 area, representing one sector of a cellular network.

• UEs are spread randomly in this area. In addition, inside the same area, small cells are randomly placed varying in number [50, 100, 250, 500].

• For the macrocell (m), we calculated the mean spectral efficiency (M) according to distances of the UE to the RRH (D).

• This calculation was done based on a single-input single-output operational mode of a base station from 3GPP LTE-A.

- 16 -

16

For the small cell (s) the calculation was based on the draft of IEEE 802.11 ac 3.0. Each small cell was designed to support 20 UEs, and the macrocell has no support limit.

SIMULATION RESULTS

- 17 -

17

As a baseline, we measured the average spectral efficiency reached by a solo macrocell of 2.134 b/s/Hz, which remains constant because we did not limit the macrocell capacity. By deploying 50 small cells, the average spectral efficiency grows, reaching 5.235 b/s/Hz on average; nevertheless, as soon as the number of UEs per square kilometer exceeds 1000, the efficiency degrades significantly, down to 2.423 b/s/Hz. With the deployment of 250 to 500 small cells, the average spectral efficiency almost reaches the maximum of 9.75 b/s/Hz, but both efficiencies degrade in the presence of 1000 or more UEs

SIMULATION RESULTS

- 18 -

18

In the same scenario, to better understand why the spectral efficiency degrades for densities larger than 1000 UE/km2, we measured the percentage of saturated small cells, that is, small cells that cannot provide communication to any additional Users . The number of saturated small cells grows quickly between UE densities of 10 and 1000/km2. The average spectral efficiency decreases in proportion to the number of saturated small cells. Therefore, a massive number of small cells is required to reach better average spectral efficiency in some densely populated areas. In this case, exploiting the sharing of small cells becomes inevitable

SIMULATION RESULTS

- 19 -

19

SIMULATION RESULTS (Specturm Sharing)

- 20 -

20

PHY

Clou

d Res

ource

optim

izer

Data

EPC

PDCP

RLC

Unified MAC

PHY

MAC Ctrl syn

I/Q I/Q

RRH1 RRH2

Future work: Unified Mac