Advances in Intelligent Systems and Computing 884

Miguel Botto-TobarLida Barba-MaggiJavier González-HuertaPatricio Villacrés-CevallosOmar S. GómezMaría I. Uvidia-Fassler Editors

Information and Communication Technologies of Ecuador (TIC.EC)

Advances in Intelligent Systems and Computing

Volume 884

Series editor

Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Polande-mail: kacprzyk@ibspan.waw.pl

The series “Advances in Intelligent Systems and Computing” contains publications on theory,applications, and design methods of Intelligent Systems and Intelligent Computing. Virtually alldisciplines such as engineering, natural sciences, computer and information science, ICT, economics,business, e-commerce, environment, healthcare, life science are covered. The list of topics spans all theareas of modern intelligent systems and computing such as: computational intelligence, soft computingincluding neural networks, fuzzy systems, evolutionary computing and the fusion of these paradigms,social intelligence, ambient intelligence, computational neuroscience, artificial life, virtual worlds andsociety, cognitive science and systems, Perception and Vision, DNA and immune based systems,self-organizing and adaptive systems, e-Learning and teaching, human-centered and human-centriccomputing, recommender systems, intelligent control, robotics and mechatronics includinghuman-machine teaming, knowledge-based paradigms, learning paradigms, machine ethics, intelligentdata analysis, knowledge management, intelligent agents, intelligent decision making and support,intelligent network security, trustmanagement, interactive entertainment,Web intelligence andmultimedia.

The publications within “Advances in Intelligent Systems and Computing” are primarily proceedingsof important conferences, symposia and congresses. They cover significant recent developments in thefield, both of a foundational and applicable character. An important characteristic feature of the series isthe short publication time and world-wide distribution. This permits a rapid and broad dissemination ofresearch results.

Advisory Board

Chairman

Nikhil R. Pal, Indian Statistical Institute, Kolkata, Indiae-mail: nikhil@isical.ac.in

Members

Rafael Bello Perez, Universidad Central “Marta Abreu” de Las Villas, Santa Clara, Cubae-mail: rbellop@uclv.edu.cu

Emilio S. Corchado, University of Salamanca, Salamanca, Spaine-mail: escorchado@usal.es

Hani Hagras, University of Essex, Colchester, UKe-mail: hani@essex.ac.uk

László T. Kóczy, Széchenyi István University, Győr, Hungarye-mail: koczy@sze.hu

Vladik Kreinovich, University of Texas at El Paso, El Paso, USAe-mail: vladik@utep.edu

Chin-Teng Lin, National Chiao Tung University, Hsinchu, Taiwane-mail: ctlin@mail.nctu.edu.tw

Jie Lu, University of Technology, Sydney, Australiae-mail: Jie.Lu@uts.edu.au

Patricia Melin, Tijuana Institute of Technology, Tijuana, Mexicoe-mail: epmelin@hafsamx.org

Nadia Nedjah, State University of Rio de Janeiro, Rio de Janeiro, Brazile-mail: nadia@eng.uerj.br

Ngoc Thanh Nguyen, Wroclaw University of Technology, Wroclaw, Polande-mail: Ngoc-Thanh.Nguyen@pwr.edu.pl

Jun Wang, The Chinese University of Hong Kong, Shatin, Hong Konge-mail: jwang@mae.cuhk.edu.hk

More information about this series at http://www.springer.com/series/11156

Miguel Botto-Tobar • Lida Barba-MaggiJavier González-Huerta • Patricio Villacrés-CevallosOmar S. Gómez • María I. Uvidia-FasslerEditors

Informationand CommunicationTechnologies of Ecuador(TIC.EC)

123

EditorsMiguel Botto-TobarDepartment of Mathematics and Computer

ScienceEindhoven University of TechnologyEindhoven, Noord-Brabant, The Netherlands

Lida Barba-MaggiFacultad de IngenieríaUniversidad Nacional de ChimborazoRiobamba, Ecuador

Javier González-HuertaDepartment of Software EngineeringBlekinge Tekniska HögskolaKarlskrona, Blekinge Län, Sweden

Patricio Villacrés-CevallosFacultad de IngenieríaUniversidad Nacional de ChimborazoRiobamba, Ecuador

Omar S. GómezEscuela Superior Politécnica de ChimborazoRiobamba, Ecuador

María I. Uvidia-FasslerFacultad de IngenieríaUniversidad Nacional de ChimborazoRiobamba, Ecuador

ISSN 2194-5357 ISSN 2194-5365 (electronic)Advances in Intelligent Systems and ComputingISBN 978-3-030-02827-5 ISBN 978-3-030-02828-2 (eBook)https://doi.org/10.1007/978-3-030-02828-2

Library of Congress Control Number: 2018958314

© Springer Nature Switzerland AG 2019This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or partof the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmissionor information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilarmethodology now known or hereafter developed.The use of general descriptive names, registered names, trademarks, service marks, etc. in thispublication does not imply, even in the absence of a specific statement, that such names are exempt fromthe relevant protective laws and regulations and therefore free for general use.The publisher, the authors and the editors are safe to assume that the advice and information in thisbook are believed to be true and accurate at the date of publication. Neither the publisher nor theauthors or the editors give a warranty, express or implied, with respect to the material contained herein orfor any errors or omissions that may have been made. The publisher remains neutral with regard tojurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AGThe registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface

The sixth conference on Information and Communication Technologies “TIC-EC”was held in Riobamba—Ecuador from November 21 until 23, 2018. This academicevent is considered as one of the most important conferences about ICT in Ecuador,as it brings scholars and practitioners from the country and abroad to discuss thedevelopment, issues, and projections of the use of information and communicationtechnologies in multiples fields of application. In 2018, the “TIC-EC” conferencewas organized by Universidad Nacional del Chimborazo (Unach) and itsEngineering School, and the Ecuadorian Corporation for the Development ofResearch and Academia (CEDIA). The content of this volume is related to thefollowing subjects:

• Communication Networks• Software Engineering• Computer Sciences• Architecture• Intelligent Territory Management• IT Management• Web Technologies• Engineering, Industry, and Construction with ICT Support• Entrepreneurship and Innovation at the Academy: a business perspective

In its 2018 edition, the TIC-EC conference received 87 submissions in Englishfrom 234 authors coming from nine different countries. All these papers werepeer-reviewed by the TIC-EC 2018 Program Committee consisting of 50high-quality researchers coming from 12 different countries. To assure ahigh-quality and thoughtful review process, we assigned each paper at least threereviewers. Based on the results of the peer reviews, 27 full papers were accepted,resulting in a 31% acceptance rate, which was within our goal of less than 40%.

v

We would like to express our sincere gratitude to the invited speakers for theirinspirational talks, to the authors for submitting their work to this conference, andthe reviewers for sharing their experience during the selection process.

November 2018 Miguel Botto-TobarLida Barba-Maggi

Javier González-HuertaPatricio Villacrés-Cevallos

Omar S. GómezMaría I. Uvidia-Fassler

vi Preface

Organization

Honorary Committee

Nicolay Samaniego Erazo

Presidente de CEDIA/Rector Unach

Juan Pablo Carvallo Vega

Director Ejecutivo CEDIA

Patricio Villacrés-Cevallos

Decano Facultad de Ingeniería, Unach

Organizing Committee

Lida Barba-Maggi, UnachCiro Radicelli García, UnachMaría Isabel Uvidia, UnachGabriela Jimena Dumancela Nina, UnachGalia Rivas Toral, CEDIAAndrea Daniela Morales Rodríguez, CEDIAXimena Lazo Álvarez, CEDIA

vii

Program Committee

Miguel Botto-Tobar Eindhoven University of Technology,The Netherlands

Angela Díaz Cadena Universitat de Valencia, SpainAndrés Robles Edinburgh Napier University, UKAndrés José Cueva Costales Yachay EP, EcuadorYan Pacheco Universidad de las Américas, EcuadorFadloun Samiha University of Montpellier, FranceGuillermo Pizarro Universidad Politécnica Salesiana, EcuadorOrlando Erazo Universidad Técnica Estatal de Quevedo,

EcuadorMaría L. Montoya Freire Aalto University, FinlandErick Cuenca University of Montpellier, FranceDavid Rivera Espín Interamerican Center of Tax Administrations,

PanamáYuliana Jimenez Universidad Técnica Particular de Loja, EcuadorJuan Fernando Balarezo

SerranoRadical Alternativas de Avanzada, Ecuador

Luis Felipe Urquiza Aguiar Escuela Politécnica Nacional, EcuadorGustavo Andrade-Miranda Universidad de Guayaquil, EcuadorWayner Xavier Bustamante

GrandaUniversidad Internacional del Ecuador, Ecuador

Janneth Chicaíza Universidad Técnica Particular de Loja, EcuadorDiego Vallejo-Huanga Universidad Politécnica Salesiana, EcuadorDanilo Jaramillo Hurtado Universidad Técnica Particular de Loja, EcuadorPablo Palacios Játiva Universidad de las Américas, EcuadorMarlon Navia Mendoza ESPAM-MFL, EcuadorPablo Saa Universidad Tecnológica Equinoccial, EcuadorJeffery Alex Naranjo Cedeño Universidad Politécnica Estatal del Carchi,

EcuadorJefferson Ribadeneira Ramírez Escuela Superior Politécnica de Chimborazo

(ESPOCH), EcuadorJulio Proaño Universidad Politécnica Salesiana, EcuadorMaikel Leyva Vázquez Universidad de Guayaquil, EcuadorAlex Cazañas Universidad de Coimbra, PortugalJaime Jarrín AndeanTrade, EcuadorWashington Velásquez Universidad Politécnica de Madrid, SpainMarco Fabricio Falconi

NoriegaCorporación Nacional de Telecomunicaciones,

EcuadorJosé Luis Carrera Villacrés University of Bern, SwitzerlandGermania Rodríguez Morales Universidad Técnica Particular de Loja, EcuadorPatricia Ludeña González Universidad Técnica Particular de Loja, EcuadorMarcia M. Bayas Sampedro Universidad Estatal Península de Santa Elena,

Ecuador

viii Organization

Israel Pineda Universidad Metropolitana del Ecuador, EcuadorTania Jhomara Palacios

CrespoUniversidad Católica de Cuenca, Cuenca

Jaime Meza University of Fribug, SwitzerlandMaría Fernanda Granda Universidad de Cuenca, EcuadorOtto Parra González Universidad de Cuenca, EcuadorJacqueline N. Mejía Luna Escuela Superior Politécnica del Litoral, EcuadorMiguel Zúñiga Prieto Universidad de Cuenca, EcuadorAngel Cuenca-Ortega Universitat Politécnica de Valencia, SpainYuliana Jiménez Gaona Università de Bologna, ItalyLuis Urquiza Aguiar Escuela Politécnina Nacional, EcuadorJohanna Ortega Universidad de las Américas, EcuadorCristhy Jiménez Granizo Pontificia Universidad Católica de Valparaiso,

Chile/Universidad Nacional del Chimborazo,Ecuador

Germania Rodríguez Universidad Técnica Particular de Loja, EcuadorPablo Torres-Carrión Universidad Técnica Particular de Loja, Ecuador

Sponsoring Institutions

Universidad Nacional del Chimborazohttp://www.unach.edu.ec/

CEDIAhttps://www.cedia.edu.ec/es/

Organization ix

Contents

Communication Networks

Millimeter-Wave Channel Estimation Using Coalitional Game . . . . . . . 3Pablo Palacios, José Julio Freire, and Milton Román-Cañizáres

Resource Allocation in WDM vs. Flex-Grid Networks:Use Case in CEDIA Optical Backbone Network . . . . . . . . . . . . . . . . . . 18Rubén Rumipamba-Zambrano, Luis Vargas, Claudio Chacón,Flavio Rodríguez, and Juan Pablo Carvallo

NFC-Based Payment System Using Smartphonesfor Public Transport Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34Diego Veloz-Cherrez and Jaime Suárez

An Open Source Synchronous and Asynchronous Approachfor Database Replication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Marcos Orellana Cordero, Gerardo Orellana Cordero,and Esteban Crespo Martinez

Forensics Analysis on Mobile Devices:A Systematic Mapping Study . . . 57Jessica Camacho, Karina Campos, Priscila Cedillo, Bryan Coronel,and Alexandra Bermeo

Software Engineering

Analytic Hierarchy Process of Selection in Version Control Systems:Applied to Software Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75Javier Vargas, Franklin Mayorga, David Guevara, and Edison Álvarez

Reliability and Validity of Postural Evaluations with Kinect v2 SensorErgonomic Evaluation System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86Christian Mariño, Rafael Santana, Javier Vargas, Luis Morales,and Lorena Cisneros

xi

Improving the Design of Virtual Learning Environmentsfrom a Usability Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100Germania Rodriguez Morales, Pablo Torres-Carrion, Jennifer Pérez,and Luis Peñafiel

The Digital Preservation in Chimborazo: A Pending Responsibility . . . . 116Fernando Molina-Granja

Offensive Security: Ethical Hacking Methodology on the Web . . . . . . . . 127Fabián Cuzme-Rodríguez, Marcelo León-Gudiño, Luis Suárez-Zambrano,and Mauricio Domínguez-Limaico

Identification of Skills for the Formation of Agile High PerformanceTeams: A Systematic Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141Héctor Cornide-Reyes, Servando Campillay, Andrés Alfaro,and Rodolfo Villarroel

Computer Sciences

A Text Mining Approach to Discover Real-Time Transit Eventsfrom Twitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155Belén Arias Zhañay, Gerardo Orellana Cordero, Marcos Orellana Cordero,and María-Inés Acosta Urigüen

Automatic Microstructural Classification with ConvolutionalNeural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170Guachi Lorena, Guachi Robinson, Perri Stefania, Corsonello Pasquale,Bini Fabiano, and Marinozzi Franco

Clustering Algorithm Optimization Applied to MetagenomicsUsing Big Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182Julián Vanegas and Isis Bonet

Intelligent System of Squat Analysis Exercise to PreventBack Injuries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193Paul D. Rosero-Montalvo, Anderson Dibujes, Carlos Vásquez-Ayala,Ana Umaquinga-Criollo, Jaime R. Michilena, Luis Suaréz, Stefany Flores,and Daniel Jaramillo

Architecture

Multifunctional Exoskeletal Orthosis for Hand RehabilitationBased on Virtual Reality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209Patricio D. Cartagena, José E. Naranjo, Lenin F. Saltos, Carlos A. Garcia,and Marcelo V. Garcia

xii Contents

Intelligent Territory Management

Subregion Districting to Optimize the Municipal Solid WasteCollection Network: A Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225Israel D. Herrera-Granda, Juan C. León-Jácome,Leandro L. Lorente-Leyva, Fausto Lucano-Chávez,Yakcleem Montero-Santos, Winston G. Oviedo-Pantoja,and Christian S. Díaz-Cajas

IT Management

Importance of ICT’s Use in Business Management and ItsContribution to the Improvement of University Processes . . . . . . . . . . . 241Johanna Rosalí Reyes Reinoso and Deisy Carolina Castillo Castillo

ICT and Business Inclusion in the Southern Communities of the Cityof Bogotá – Colombia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253Camilo José Peña Lapeira and Cliden Amanda Pereira Bolaños

Edition, Publication and Visualization of GeoservicesUsing Open-Source Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266Pablo Landeta, Jorge Vásquez, Xavier Rea, and Iván García-Santillán

Web Technologies

LOD-GF: An Integral Linked Open Data Generation Framework . . . . 283Víctor Saquicela, José Segarra, José Ortiz, Andrés Tello,Mauricio Espinoza, Lucía Lupercio, and Boris Villazón-Terrazas

Semantic Architecture for the Extraction, Storage, Processingand Visualization of Internet Sources Through the Use of Scrapyand Crawler Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301Ramiro Leonardo Ramírez-Coronel, Ana Cristina Cárdenas,María-Belén Mora-Arciniegas, and Gladys-Alicia Tenesaca-Luna

Use of Apache Flume in the Big Data Environment for Processingand Evaluation of the Data Quality of the Twitter Social Network . . . . 314Gladys-Alicia Tenesaca-Luna, Diego Imba, María-Belén Mora-Arciniegas,Verónica Segarra-Faggioni, and Ramiro Leonardo Ramírez-Coronel

ICT in Education

Sophomore Students’ Acceptance of Social Media for ManagingGeoreferenced Data in a Socially-Enhanced CollaborativeLearning Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329Erika Lozada-Martínez, Félix Fernández-Peña, and Pilar Urrutia-Urrutia

Contents xiii

Engineering, Industry, and Construction with ICT Support

Random Sub-sampling Cross Validation for Empirical CorrelationBetween Heart Rate Variability, Biochemicaland Anthropometrics Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347Erika Severeyn, Jesús Velásquez, Héctor Herrera, and Sara Wong

Robotic Arm Manipulation Under IEC 61499 and ROS-basedCompatible Control Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358Carlos A. Garcia, Gustavo Salinas, Victor M. Perez, Franklin Salazar L.,and Marcelo V. Garcia

EDA and a Tailored Data Imputation Algorithm for DailyOzone Concentrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372Ronald Gualán, Víctor Saquicela, and Long Tran-Thanh

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387

xiv Contents

Communication Networks

Millimeter-Wave Channel EstimationUsing Coalitional Game

Pablo Palacios1(B), Jose Julio Freire2(B), and Milton Roman-Canizares2(B)

1 Departamento de Ingenierıa Electrica, Universidad de Chile, Santiago, Chilepablo.palacios@ug.uchile.cl

2 Departamento de Redes y Telecomunicaciones, Universidad De Las Americas,Quito, Ecuador

{jose.freire,milton.roman}@udla.edu.ec

Abstract. In millimeter-wave (mm-wave) massive MIMO systems, thechannel estimation (CE) is a crucial component to set the mm-wave links.Unfortunately, acquiring channel knowledge is a source of training over-head. In this paper, we propose a CE method leveraging measurementsat sub 6-Ghz frequencies in order to reduce the training overhead. Thissolution extracts spatial information from a sub 6-Ghz channel using avirtual channel transformation, such as the searching space is reducedto the information provided by the low frequency channel. In a secondstage, a multicell system and its interference between cells is analyzed,proposing a coalitional game to deal with the intercell interference. In thesingle cell case, we analyze the proposed method in different SNR scenar-ios, the computational complexity and over user equipment (UE) mobil-ity environment. Finally, we analyze how the coalitional game improvesthe throughput and its performance over UE in mobility cases.

Keywords: mm-wave · High-speed · Coalitional gameChannel estimation

1 Introduction

Large antenna arrays (i.e. Massive MIMO) at both sides eNodeB (eNB) andUE is a promising technology to achieve high-throughput services [1]. Largeantenna arrays at the same time deal with the high path-loss in millimeterfrequencies. By the other hand, channel states information (CSI) in terms ofchannel matrix or beam alignment are needed at the eNB to point the beams inthe UE direction. Both strategies are usually acquired by a training sequence [2].The sequence is used to measure every beamformer and combiner to estimatethe pair of beams that are closer to the desired angles, but this exhaustive-searchmethod need a large number of measurements to estimate the best beam-pair.Additionally, this fact could leads to lower channel rate. In vehicular or trainscenarios where due to the UE speed, the channel coherence time becomes shorterand the training period could occupies all the coherence time, leaving no timefor data transmission [3].c© Springer Nature Switzerland AG 2019M. Botto-Tobar et al. (Eds.): TICEC 2018, AISC 884, pp. 3–17, 2019.https://doi.org/10.1007/978-3-030-02828-2_1

4 P. Palacios et al.

The document is structured as follows: Sect. 2 describes the current work onmillimeter waves, radio cognitive and work proposal. Then, Sect. 3 introduces themodel of the system used for the evaluation of the proposed method. After that,Sect. 4 reports the channel estimation method for the proposed work. Section 5includes the multicell analysis in which the proposed coalition game model isdescribed with its respective algorithm. Finally, in Sect. 6 the numerical resultsobtained in the simulation are shown, to finalize with the conclusions that aredescribed in Sect. 7

2 Related Work

In the literature we can find different mechanism to estimate the Angles ofArrival (AoA) and Angle of Departure (AoD) or the complete CSI with lowertraining, for example in [4] a beam alignment method is carried out taken advan-tage of UE location. Beamforming focused on wireless backhaul in small cellnetworks and wind effect on beam misalignment is studied at [5], in the otherhand a typical mm-wave channel estimation (CE) process is carried out based oncompressive sensing (CS) framework [6–8] that is a useful technique to decreasethe training overhead although could not be appropriate for environments wherefrequent updates are required to estimate the channel, e.g. railway and vehicularscenarios.

Therefore some analysis in models with high mobility have been done in orderto provide a better understanding about mm-wave propagation for vehicular andtrain environments, in [9] an analysis focused on urban areas was deployed, moti-vated by measurements and ray tracing, researchers concluded that interferencefrom a NLOS parallel street is negligible, in [10] was found an optimal beam-width different than zero which maximizes the coherence time, the paper alsoincludes the beam misalignment due to motion in the receiver.

In [11] a location-aided mm-wave channel estimation method was proposedexploiting the eNB and vehicle position to infer the LOS path’s AoA and AoD.Unfortunately, the adaptive approach is likely to fail if the LOS path is blocked.Keeping in mind that beam training due to channel estimation process is a sourceof overhead, another point of view to resolve the problem is by using spatialinformation from different frequencies, specifically measurements from sub-6 Ghzfrequencies which are being broadly used currently, although the number ofcommon paths decreases with larger frequency separation, there is still a strongspatial information congruency among sub-6Ghz and mm-wave frequencies aswas proved in [12], where the researchers provided a mm-wave channel estimationmethod using two transform process to relate the spatial correlation matrix fromsub-6 Ghz to mm-wave frequencies. While [13] took advantage of out-of-bandinformation for beam-selection in a OFDM system, leveraging data from allactive subcarriers to decide the best beam-pair.

In this paper, we propose a channel estimation method using out-of-bandmeasurements. It is based on training and we assume the microwave channelhave been already estimated. This assumption is taken based on several studies

Millimeter-Wave Channel Estimation Using Coalitional Game 5

about channel estimation for static and high speed environments have been donefor single carrier and multi-carrier systems in micro-wave frequencies, such thatthis is a practical assumption, this solution applies virtual channel decomposition[14] to the lower frequency channel in order to extract the dominant paths, andreminding the fact that due to higher path loss and shorter wavelength, highfrequency systems are expected to use a larger antenna array than sub-6Ghz[15], this implies a narrower beam at mm-wave system.

Notation: A is a set, |A| is the cardinality of set A. lower-case a is a scalar,a is a vector, A is a matrix. AT ,AH denote the transpose and Hermitian ofmatrix A.

3 System Model

We consider a mm-wave MIMO downlink system with uniform linear arrays(ULAs) conformed by Nt transmitter antennas in the UE and Nr receiver anten-nas in the eNB, as show in Fig. 1. We consider that both the transmitter and thereceiver have only one RF chain, hence, only analog beamforming/combiningcan be applied.

We use f, and q to denote the beamformer and combiner vector, respectively:

f =1√Nt

[1, ..., ej(Nr−1) 2πλ d cosφ]T , (1)

where φ ∈ [−π/2, π/2], is a quantized angle of departure, besides f has constantmodulus entries and only phase can varying, in similar fashion the combiner:

q =1√Nr

[1, ..., ej(Nr−1) 2πλ d cos θ]T , (2)

where θ ∈ [−π/2, π/2], is a quantized angle of arrival, the AoAs and AoDs canbe taken following regular or non regular sampling strategies, the detail abouthow we choose this angles is discussed later in Sect. 3.2. Then considering anarrowband channel model H ∈ C

Nr×NT , the received signal in the eNB can bemodeled as:

y =√

ρqHHfx + qHv, (3)

where√

ρ is the average transmit power in the training phase, x is the trainingsymbol, and v is the vector of i.i.d. ∼ CN (0, σ2

0I) noise.

3.1 Millimeter-Wave Channel Model

We adopt a geometric channel model with L scatterers, where each scatterercontributes to one propagation path. Accordingly, the channel matrix H, can beexpressed as:

H =√

NtNr

L∑

l=1

αlar(θl)aHt (φl), (4)

6 P. Palacios et al.

Fig. 1. Transmitter and receiver architecture.

Fig. 2. Illustration of virtual channel matrix for: (a) Sub-6 Ghz with Nr = Nt = 16, (b)Mmwave with Nr = Nt = 64.

where L is the number of paths, αl represents the complex path gain of thel -th propagation path, θl ∈ [−π/2, π/2] and φl ∈ [−π/2, π/2] denote the AoAand AoD of the L-th path at transmitter and receiver, respectively. The vectorsat(·) and ar(·) denote the array response vectors for transmitting and receivingantenna arrays. The array response vector of ar(θl) is given by:

ar[θl] =1√Nr

[1, ej 2πλ d cos θl , ..., ej(Nr−1) 2π

λ d cos θl ]T , (5)

where λ is the transmission wave length and d is the antenna spacing. Further-more the array response vector in (5) has a unit norm and the factor

√NtNr in

(4) reflects this normalization. The array response vector at(φl), can be writtenin a similar fashion.

3.2 Beam-Codebook Design

Different codebook models have been designed according to the channel prop-erty for instance Grassmannian codebooks, or to satisfy an hybrid architecturefor example Multi-Resolution hierarchical codebook In this work due to full-analogous architecture we denote the codebooks F and Q at the transmitterand receiver respectively as:

F = {f1, f2, ..., fNt} (6)

Q = {q1,q2, ...,qNr} (7)

Millimeter-Wave Channel Estimation Using Coalitional Game 7

where every beamforming vector has the same form of the array response vector,such that:

fm = at(φm), m ∈ {1, ..., Nt} (8)qn = ar(θn), n ∈ {1, ..., Nr} (9)

The angles are chosen as in [14] such that every beam has same magnitude butdifferent width, that is to say narrower at broadside direction and broader atendfire direction, if we set the inter-antenna spacing as λ/2, the positive andnegative angles of departure are given by:

φ(+)m = arcsin(

2m

Nt), m ∈ {1, ...,

Nt

2} (10)

φ(−)m = arcsin(

2m

Nt), m ∈ {−1, ...,

−Nt

2} (11)

The AoAs design follow the same rule than AoDs.

4 Channel Estimation Method

4.1 Extracting Spatial Information

Here thanks to a matrix transformation we propose an easy methodology toobtain the spatial information from sub-6Ghz channel. Considering a geometricchannel model H6G already estimated, whose structure doesn’t provide clearinformation about AoAs and AoDs in the different paths, therefore a channelrepresentation that provides a simpler geometric interpretation of the scatteringenvironment is needed.

A virtual channel representation (VCR) of H6G will provide spatial infor-mation uniformly spaced over the virtual angles, which are determined by thespatial resolution of the array. Thus, VCR characterizes the MIMO channel viabeamforming in the direction of fixed virtual transmit and receive angles [14],that is:

H6G = UrH6GUHt (12)

where Ur = [ur(θ−k), ...,ur(θk)], −Nr/2 ≥ k ≤ Nr/2, is a matrix Nr × Nr

which carries the receiver response vector in the virtual directions that sat-isfy θk = arcsin( 2k

Nr), likewise Ut = [ut(φ−i), ...,ut(φi)], −Nt/2 ≥ i ≤ Nt/2

is a matrix Nt × Nt that carries the transmitter response vector in the virtualdirections that satisfy φt = arcsin( 2i

Nt), consequently Ut and Ur are unitary

discrete Fourier transform (DFT) matrices reflecting the fixed virtual receiveand transmit angles, and H6G ∈ C

Nr×Nt is the virtual channel matrix and itsentries reveals the desired channel parameters, virtual AoAs, AoDs and pathgain [14]. In addition, typical sub-6 Ghz MIMO systems carry with lower num-ber of antennas than mm-wave, this fact will lead to broader virtual angles inthe low frequency channel compared with the high frequency channel, as shownin Fig. 2 several mm-wave virtual AoAs and AoDs overlap to those at virtual

8 P. Palacios et al.

sub-6 Ghz, reducing the searching space to those overlapped mm-wave angles,then the candidate beam list provided by the spatial information from the sub-6 Ghz channel, is stored in the set S. At this point the task is select the pair ofvirtual angles in the mm-wave virtual representation that provide higher gain inorder in the receiver in order to estimate the mm-wave channel.

Considering the set S is carrying the bunch of possibles arrival and departurevirtual angles θp and φr respectively, where θp, p = 1, 2, ..., P ;P < Nr andφr, r = 1, 2, ..., Nt;R < Nt decreasing the searching space and training overheadto P × R. Then using the set of arrival virtual angles to construct the combinerat the receiver i.e. Q = [q1(θ1),q2(θ2), ...,qP (θP )]. The departure virtual anglesare feedback to the transmitter in order to build its beamformer F, where log2 Pbits are needed, that could be sent using the sub-6 Ghz channel. The new receivedsignal in mm-wave systems is

Y =√

ρQHS HFS + V, (13)

After a vectorization step:

y =√

ρ(QTS ⊗ FH

S )h + v (14)

where y = vec(Y), h = vec(H), and v = vec(V). The largest absolute valueentry in (15) determines the best beam-pair, that is to say iopt = arg max |[y]i|,will match the best pair of AoA and AoD.

Length of signal x5 10 15 20 25 30 35 40

0

0.2

0.4

0.6

0.8

1

1.2

Fig. 3. Complex path gain estimation for a typical value.

4.2 Path Gain Estimation

At this point the AoAs and AoDs for different paths have been estimated bythe qi and fi vectors, therefore the final step is to estimate the nonzero valuesentries of the channel, to simplify the analysis, we assume a single path channel,then re-writing Eq. (3) with a vector training symbol such that x ∈ C

1×s, thereceived signal is:

y =√

ρqHiHfix + v, (15)

Millimeter-Wave Channel Estimation Using Coalitional Game 9

We apply the linear LS estimator to calculate the complex path value associatedto the given AoA and AoD, given by

z = (xxH)−1xyH (16)

where z =√

ρqHiHfi, that carries the path gain estimation. About the vector

training symbol length the Fig. 3 shows the variability of the path gain estimatedαl according to the signal length, we can observe there is a convergence with alength further than 20.

Additionally we focus on analyze the rate between the eNB and an UE indownlink transmission. Then the rate R is affected by changing the channelcoherence time due to MS velocity, as

RSingle =To − Tτ

Tolog2(1 +

ρqHiHfif

HiHHqi

σ2o

) (17)

where the pre log-factor takes account the training overhead necessary to esti-mates the channel, here To = λBo

2vo, λ,Bo, vo, are mm-wave carrier wavelength,

bandwidth and mobile station speed respectively, additionally Tτ is the numberof P × R training blocks needed to estimate the channel.

Another metric choosed to evaluate the performance of the proposed methodis the computational complexity compared with Fast Channel Estimation (FCE)method described in [16]. The complexity cost is given by O(NrNtL + NrL)counting the beam and gain selection, by the other hand the FCE computationalcomplexity is O(NrT + LNtT + LT ).

5 Multicell Analysis

From now, considering an Orthogonal Frequency Division Multiple Access(OFDMA) multicell system as is shown in Fig. 4. We assume in every cell thereis a small cell microwave station and millimeter-wave small station located inthe same position. Also, we assume a microwave channel estimation have beenalready done in every small cell microwave station, such that a CE training-basedmethod is applied in every millimeter-wave small station. We will analyze theperformance of this mm-wave CE proposed method. Therefore, we focus on thisfrequency band. Additionally we assume no interference among the subchannels.

Let’s consider the Fig. 4 where the UE1 is located in the eNB2 coverage areaborder, such that this user must deal with handoff management and interfer-ence from neighboring cells. In order to overcome these problems, we propose amethod based on cooperative model using coalitional games between the con-cerned eNBs. The goals of this section is to increase the channel rate in thetransmission data stage.

5.1 Coalitional Game: System Model

Assuming a mm-wave system, with N cells in the network and N = {1, ..., N}denoting the set of millimeter-wave small cells (MMWSCs) which are connected

10 P. Palacios et al.

with each other via a wire backhaul, e.g., fiber, providing reliable links for trafficcontrol. Each MMWSC i ∈ N works over the same set of channels, that is theavailable spectrum, i.e., is the frequency resource shared between all MMWSCs.Therefore the MMWSCs should share these bandwidth allocated in an oppor-tunistic way in order to avoid handoff recurrent, besides diminishing the inter-ference, leading to an improve in the channel rate.

Fig. 4. Multicell network architecture

In the non-cooperative scenario, the access mode is frequency division duplex-ing (FDD) where each MMWC i ∈ N transmits over a set of subchannels Γ ,which contains |Γ | = M subchannels. The MMWSC i occupies the full timeduration of all its l ∈ Γ subchannels, under this non-cooperative scenario theUE1 (Fig. 4) uses the M subchannels available to transmit its data. In this casea UE located near the coverage area border may suffer an important degradationdue to interference and coalitions in the subchannels from neigboring MMWSCs.Under this setup and without considering UE mobility, we can rewrite Eq. 15 as:

yi,l =√

ρqHi,lHi,l fi,lxi,l + v + IS , (18)

where IS =∑

j∈N ,j �=i

√ρ qH

j,lHj,lfj,lxj,l, is the interference from neighboringcells during the mm-wave CE process in the eNB. Here is easy to notice howthe interference affect the estimation process. By the other hand the rate indownlink transmission of MMWC i ∈ N to an UE is given by

RMulti =∑

l∈Γ

log2(1 +ρi,lqH

i,lHi,lfi,lfHi,lHHi,lqi,l

σ2o + IS

) (19)

where ρi,l denotes the downlink transmit power by MMWSC i to a UE onsubchannel l, fi,l is the beamforming vector in the MMWSC i pointing to UE

Millimeter-Wave Channel Estimation Using Coalitional Game 11

on subchannel l, Hi,l is the massive MIMO channel between the MMWSC i andthe UE on the l-th subchannel, qi,l is the combining vector in the UE pointingto MMWSC i on subchannel l and IS denotes the power interference suffered byan UE from neighboring MMWSCs, as

IS =∑

j∈N ,j �=i

ρj,lqHj,lHj,lfj,lfHj,lH

Hj,lqj,l, (20)

The term IS can significantly reduce the rates achieved. Specifically, depend-ing on the signal to noise and interference ratio (SNIR) feedback from the UE,the MMWCs can decide to form cooperative groups called coalitions, in orderto overcome the interference between neighboring cells and the handoff manage-ment. Under this coalitional game approach, the MMWCs are modeled as playersthat access the spectrum, avoiding coalitions among them by jointly schedulingtheir transmissions, we state the following definition for a coalition:

Definition 1. A coalition S ⊆ N is a non−empty subset of N in which playersinside the set access the spectrum via a coordinated manner.

Therefore we consider that if a coalition S is formed, the transmissions inside Sare managed by a local scheduler as in [20], using Time Division multiple access(TDMA) mode, such that the subchannels are split in time slots allocated forevery MMWSC. As a result no more than one MMWSC will access the everychannel in each time slot, mitigating the interference inside the coalition S.

Although coordination can help to increase the channel rate by decreasingthe interference, it also incurs in a coordination cost. Here we consider this coor-dination cost in terms of transmit power. Thus, the power spent by a MMWSCi to reach the farthest MMWSC j in a coalition S is ρj,i. Then the power costneeded to form a coalition S is:

ρS =∑

i∈S

ρj,i (21)

In addition, we define a maximum tolerable power cost ρlim for every coalitionS as in [21]. By considering the coalition cost in this way we take account thespatial distribution of the MMSCs and the coalition size.

5.2 MMWCs Cooperation as a Coalitional Game

The main goal is to deal with interference from neighboring MMWSC in theborder coverage area by forming coalitions, i.e., using coalitional game theory[18], we denote as B the set of all partitions GN of N , this problem can bemodeled as coalitional game in partition form with transferable utility as [19]:

Definition 2. A coalitional game in partition form with transferable utility(TU) is defined by the pair (N , v) where N is the set of players in the game,and a value function v(S,GN ) assigning a real value to each coalition S.

12 P. Palacios et al.

We also assume that v(∅) = 0. Thus, the definition above imposes a dependenceon the coalitional structure N when evaluating the value of S ⊆ N , i.e. to theplayers N\S as well. By the other hand the TU property implies that the totalutility represented by a real number, (in our case the channel rate) can be dividedin any manner between the coalition members.

Clearly, the cooperation model can be stated as a game in partition formwhere the MMWSC are the players, since the channel rate of a coalition isaffected by the interference from others players, thus there is a dependencebetween the coalition and the players which do not belong to the coalition(coalitional structure). Therefore the utility achieved by the coalition S canbe expressed in terms of the channel rate as:

U(S, GN ) =∑

i∈S

∑

l∈Γ

αli log2(1 +

ρi,lqHi,lHi,lfi,lfHi,lH

Hi,lqi,l

σ2o + IS

), (22)

where αli ∈ [0, 1] denotes the fraction of time duration during which MMWSC i

transmits on the subchannel l to the UE. In the non-cooperative scenario, i.e.,FDD transmission mode, each transmission occupies a whole subchannel, henceαl

i = 1. In addition IS denotes the co-tier interference suffered by the UE servedby MMWSC i on subchannel l from players j ∈ N\S as follows:

IS =∑

j∈GN \S, j �=i

ρj,lqHj,lHj,lfj,lfHj,lH

Hj,lqj,l, (23)

Therefore thanks to transmission scheduling the interference from players withinthe coalition S is suppressed, while inter-coalition interference still remain andleads to a game in partition form. Given the power cost and utility function forany coalition S ∈ N , we can define the value of any coalition, i.e., the totalbenefit as:

v(S, GN ) ={ |S| U(S, GN ) if ρS ≤ ρlim

0 otherwise,(24)

As the utility in (24) represents a sum rate, then the proposed coalitionalgame has a transferable utility, since the sum rate can be shared among the coali-tion members by dividing the frequency resource in any manner, while meeting afairness criterion, consequently our aim is to maximize the sum rate while takingaccount the constraints in terms of transmit power.

We can define the payoff of a MMWSC i ∈ S as:

xi(S, GN ) =1

|S|

⎛

⎝v(S, πN ) −∑

j∈Sv({j}, GN )

⎞

⎠ + v({i}, GN ) (25)

5.3 Proposed Algorithm

For the stated coalitional game is important to notice that due to the powerconstraint the grand coalition seldom forms. Therefore cooperation will occurs

Millimeter-Wave Channel Estimation Using Coalitional Game 13

when the interfering MMWSCs are closely located in a way that ρS ≤ ρlim, thuscoalition with many members is unlikely to happen, in this sense we are not focuson analyze the stability of the grand coalition. Finding an optimal coalitionalstructure for games in partition form have been studied in, [22] and [23], wewill apply the concept of recursive core as in [23]. The concept of recursive corestudies the behaviour of dynamics coalition formation but also considering theinterference from neighboring MMWSCs. The detail about how the recursivecore works is provided in [23].

In the algorithm proposed (Algorithm1), at the first step an UE is sens-ing interference which will trigger the coalitional game. Secondly to resolve thecoalitional game, i.e., achieving the recursive core, we propose three phases: envi-ronment sensing, coalition formation and scheduling transmission. First of all,the network is partitioned in N single coalitions, this is the non-cooperativecase. Then by discovering neighbors stage, the MMWSCs can create a list ofexisting neighbors in the network, once each MMWSC has a neighboring list,they can start a recursive coalition formation to find a recursive core. Hereevery MMWSC establishes negotiations with the discovered neighbors, to iden-tify potential partners for cooperation, this information is exchanged by usingthe wire reliable control channel. Then the cooperation cost for every coalitionis calculates as in Eq. 21, and the potential payoff is computed as in Eq. 25 forevery member of a coalition. To reach the recursive core, each MMWSC joinsto the coalition which provides the highest revenue, i.e., payoff. Then, once thecoalition is formed, coalition-level scheduling occurs in each coalition.

6 Numerical Results

In this section, simulations are carried out to evaluate the performance for theproposed channel estimation strategies. We assess the channel estimation error,+the effect of SNR and speed on the method’s performance of the proposedmethods, additionally the computation time is analyzed.+

The sub-6Ghz channel works at f6G = 3 GHz and mm-wave channel atfmm = 28 GHz, bandwidth Bo = 10 MHz alike than mm-wave case, the distancebetween the UE and eNB is set at 50m, the path loss exponent at sub-6Ghzis equal to 2 while for mm-wave has been set to 3. The angles of arrival anddeparture for both environments are limited at [−π

2 , π2 ), the antennas number at

sub-6Ghz and mm-wave will be changing according to every experiment at theUE and eNB, the inter antenna element distance at both cases is half-wavelength.Additionally the signal-to-noise ratio is set as SNR = Po

σ2o.

To assess the estimation error we express the mm-wave virtual channel Hvirt

as sparse [16], such that Hvirt = JTLNr

Λ JLNtwhere Λ is an L × L diagonal

matrix and the L are the entries different than zero of Hvirt, the binary matricesJT

LNr,JLNt

are L × Nr and L × Nt selection matrices, generated by keeping Lrows of Nr × Nr and Nt × Nr identity matrices respectively. Therefore we cancompute the mean square error as MSE= E{‖Hvirt − Hvirt‖2F /‖Hvirt‖2F }.

14 P. Palacios et al.

Algorithm 1. The Proposed MMWSC cooperation algorithmStep 1: UE interference sensing

The UE sense the interference UEint, once it overpass a threshold Ithr, the UEfeedback the information to its attached eNB in order to initiate the cooperationprocess, thus:if UEint ≥ Ithr then

Step 2: Coalitional Game StartsAt the beginning when players are not cooperating GN = {1, ..., N} =

{S1, ..., SN }.Three stages in each round of the algorithmStage 1 - Discovering Neighbors:

– Each MMWSC discovers the neighboring coalitions.

Stage 2 - Recursive Coalition Formation:repeat

– Each MMWSC establishes negotiations with discovered neighboring FAPs, inorder to identify potential coalition partners.

– Each MMWSC create a list of the feasible coalitions which ensure ρS ≤ ρlim

– The payoff for the feasible coalitions is computed and each MMWSC joins tothe coalition which ensures the maximum payoff.

– The resulting coalition is included in the recursive core.

until convergence to a stable partition in the recursive core.Stage 3 - Inner-coalition scheduling:

– The scheduling information is gathered by each MMWSC i ∈ S from itscoalitions members, and transmitted within the coalition S afterwards.

end ifStep 3: High Speed mmWave Communications

– And high data rate transmission starts.

In order to explore the performance of the proposed method, in the firstsimulation we set the Sub 6-Ghz channel with 16 transmitter antennas and 16receiver antennas, in the other and we set the mm-wave channel with Nr = Nt =64, Nr = Nt = 32 and Nr = Nt = 16 antennas, in the Fig. 5 we can see how thespectral efficiency change according to different SNR values for every antennaarray, as is expected under the beam codebook design, increasing the number ofantennas lead to an increasing in the beam-resolution.

For exhaustive-search this occurs due to the number of Nr×Nt blocks neededfor channel estimation, here the training time occupies most of the channelcoherence time leaving no time for data sending, by the other hand under thismethod thanks to prior spatial information obtained from sub-6 Ghz the numberof beam candidates decrease, such that P � Nr and R � Nr therefore the blocksfor training are highly reduced, leaving more time for data transmission.

Millimeter-Wave Channel Estimation Using Coalitional Game 15

Fig. 5. Rate achieved by CE training-based over different SNR values.

The last experiment takes account the system complexity between the refinedCE method and FCE detailed at [17] in term of their computation time, thecomplexity cost are O(NrT + LNtT + LT ) and O(NrNtL + NrL) respectively.All simulations are conducted at Matlab R2015a by the Intel Core i5 CPU, inFig. 6(a) the Nr is set to 16, while Nt is increasing, then in Fig. 6(b) theNt is setto 16 and Nt is changing. For the sake of fairness we comparative both method

Fig. 6. Computation time of CE based on training and FCE, varying: (a) Nr and(b) Nt.

16 P. Palacios et al.

setting the parameters to obtain a higher accuracy, that is L=18 according to[16]. In the Fig. 6 we observe the computation time for both methods growslinearly when the number of antennas increase, although this proposed methodrun faster than FCE, this is mainly because the searching space is diminishedthanks to prior information obtained from sub-6 Ghz channel.

7 Conclusions

In this work, we proposed a channel estimation method based on coalitionalgame for a multicell case that improves the throughput and its performance overUE in mobility cases. The prior based on an algorithm that improves intercellinterference. We analyze how the coalitional game improves the throughput andits performance over user equipments (UEs) in mobility cases. The proposedalgorithm allows sharing the bandwidth allocated to the UE in an opportunisticmanner to avoid recurring handover, in addition to reducing interference, whichleads to an improvement in the channel speed. In addition, authors expect thatthe proposed method can be applied to reduce complexity and improve efficiencyin terms of probability of non-detection of the system for NOMA (non-orthogonalmultiple access systems).

References

1. Wang J, Zhu H, Gomes NJ (2012) Distributed antenna systems for mobile com-munications in high speed trains. IEEE J Sel Area Commun 30(4):675–683

2. Hassibi B, Hochwald BM (2003) How much training is needed in multiple-antennawireless links? IEEE Trans Inf Theory 49(4):951–963

3. Heath J, Robert W (2013) What is the role of MIMO in future cellular networks:massive? Presentation delivered at IEEE international conference on communica-tions (ICC)

4. Maschietti F, Gesbert D, Kerret P, Wymeersch H (2013) Robust location-aidedbeam alignment in millimeter wave massive MIMO. In: GLOBECOM 2017-2017IEEE global communications conference, pp 1–6

5. Hur S, Kim T, Love DJ, Krogmeier JV, Thomas TA, Ghosh A (2013) Millimeterwave beamforming for wireless backhaul and access in small cell networks. IEEETrans Commun 61(10):4391–4403

6. Alkhateeb A, El Ayach O, Leus G, Robert RW (2014) Channel estimation andhybrid precoding for millimeter wave cellular systems. IEEE J Sel Top SignalProcess 8(5):831–846

7. Alkhateeb A, Ayach OE, Leus G, Heath RW (2013) Hybrid precoding for mil-limeter wave cellular systems with partial channel knowledge. In: Proceedings ofinformation theory and applications workshop (ITA), February 2013, p 15

8. Ayach OE, Rajagopal S, Surra SA, Pi Z, Heath RW (2013) Spatially sparse pre-coding in millimeter wave MIMO Systems. In: Proceedings of information theoryand applications workshop (ITA), February 2013, p 15

9. Wang Y, Venugopal K, Andreas F, Heath RW Jr (2017) Analysis of urban mil-limeter wave microcellular networks. IEEE Trans Veh Technol 66(10):8964–8978

Millimeter-Wave Channel Estimation Using Coalitional Game 17

10. Va V, Heath RW Jr (2017) Basic relationship between channel coherence time andbeamwidth in vehicular channels. IEEE Trans Veh Technol 66(10):8964–8978

11. Garcia N, Wymeersch H, Strom EG, Slock D (2016) Location-aided mm-wave chan-nel estimation for vehicular communication. In: IEEE 17th international workshopon signal processing advances in wireless communications (SPAWC), July 2016, pp1–5

12. Ali A, Gonzalez-Prelcic N, Heath RW Jr (2016) Estimating millimeter wave chan-nels using out-of-band measurements. In: Proceedings of information theory andapplications (ITA) workshop, February 2016, p 15

13. Ali A, Gonzalez-Prelcic N, Heath RW Jr (2018) Millimeter wave beam-selectionusing out-of-band spatial information. IEEE Trans Wirel Commun 17:1038–1052

14. Sayeed AM (2002) Deconstructing multiantenna fading channels. IEEE Trans Sig-nal Process 50(10):2563–2597

15. Rappaport TS, Sun S, Mayzus R, Zhao H, Azar K, Wang GN, Wong JK, Schulz M,Samini F Gutierrez (2013) Millimeter wave mobile communications for 5G cellular:it will work! IEEE Access 1:335–349

16. Sayeed AM, Raghavan V (2016) A fast channel estimation approach for millimeter-wave massive MIMO systems. In: Proceedings of IEEE global conference on signaland information processing, Arlington, VA, USA

17. Va V, Choi J, Shimizu T, Bansal G, Heath RW Jr. Inverse multipath fingerprintingfor millimeter wave V2I beam alignmen. IEEE Access

18. Palacios P, Castro A (2018) Cognitive radio simulator for mobile networks: designand implementation. i-manager’s J Commun Eng Syst 7(2):1–9

19. Saad W, Han Z, Debbah M, Hjorungnes A, Basar T (2009) A distributed coalitionformation framework for fair user cooperation in wireless networks. IEEE TransWirel Commun 8(9):4580–4593

20. Pantisano F, Ghaboosi K, Bennis M, Verdone R (2010) Interference avoidancevia resource scheduling in TDD underlay femtocells. In: Proceedings of the IEEEPIMRC workshop on indoor and outdoor femto cells, Istambul, Turkey, September2010

21. Zhang Z, Song L, Han Z, Saad W (2014) Coalitional games with overlapping coali-tions for interference management in small cell networks. IEEE Trans Wirel Com-mun 13(5):2659–2668

22. Huang C-Y, Sjostrom T (2006) Implementation of the recursive core for partitionfunction form games. J Math Econ 42:771–793

23. Pantisano F, Bennis M, Saad W, Verdone R, Latva-Aho M (2011) Coalition for-mation games for femtocell interference management: a recursive core approach.In: Proceedings of 2011 IEEE wireless communications and networking conference,Cancun, Quintana Roo, Mexico, March 2011, p 2831

24. Bogomonlaia A, Jackson M (2002) The stability of hedonic coalition structures.Games Econ Behav 38:201–230