newcom# - cordis · 2017. 4. 22. · the main asset of the euwin@cnrs/eurecom lab is the...

PROPRIETARY RIGHTS STATEMENT This document contains information, which is proprietary to the NEWCOM# Consortium.

Network of Excellence

NEWCOM#

Network of Excellence in Wireless Communications#

FP7 Contract Number: 318306

WP 2.3 – Flexible communication terminals and net-works

D2.3.3

Second report on tools and their integration on the experi-mental setups

Contractual Delivery Date: October 31, 2014 Actual Delivery Date: November 21, 2014 Responsible Beneficiary: CNRS/Eurecom Contributing Beneficiaries: CNIT/PoliTO, CNIT/UniBO, INOV, TUD, Bilkent, IASA, Estimated Person Months: 15 Dissemination Level: Public Nature: Report Version: 1.0

Ref. Ares(2014)4050202 - 03/12/2014

PROPRIETARY RIGHTS STATEMENT This document contains information, which is proprietary to the NEWCOM# Consortium.

This page is left blank intentionally

FP7 Contract Number: 318306 Deliverable ID: WP2.3 / D2.3.3

Distribution Level: Public

Document Information

Document ID: D23.3 Version Date: November 21, 2014 Total Number of Pages: 32 Abstract: Second report on tools and their integration on the experimental

setups: description of the hardware and software developments carried out by the partners during the second year of the NoE and their integration into the three experimental setups. Pro-gress report on dissemination activities.

Keywords: EuWIn, CNRS/Eurecom, OpenAirInterface, experimental re-search, cloud RAN, TVWS, 4G/5G coexistence, localization, channel reciprocity

Authors

IMPORTANT: The information in the following two tables will be directly used for the MPA (Monitoring Partner Activity) procedure. Upon finalisation of the deliverable, please, ensure it is accurate. Use multiple pages if need-ed. Besides, please, adhere to the following rules:

• Beneficiary/Organisation: For multi-party beneficiaries (CNIT) and beneficiaries with Third Parties (CNRS and CTTC), please, indicate beneficiary and organisation (e.g., CNIT/Pisa, CNRS/Supelec).

• Role: Please, specify: Overall Editor / Section Editor / Contributor. Full Name Beneficiary /

Organisation e-mail Role

Florian Kaltenberger CNRS/Eurecom Florian.kaltenberger

@eurecom.fr Editor

Raymond Knopp CNRS/Eurecom [email protected] Contributor

Riadh Ghaddab CNRS/Eurecom [email protected] Contributor

Navid Nikaein CNRS/Eurecom [email protected] Contributor

Geroge Arvanitakis CNRS/Eurecom [email protected] Contributor

Xiwen Jiang CNRS/Eurecom [email protected] Contributor

Carlo Condo CNIT/PoliTO [email protected] Contributor

Andrea Biroli CNIT/PoliTO [email protected] Contributor

Guido Masera CNIT/PoliTO [email protected] Contributor

Lucio Studer Ferreira INOV [email protected] Contributor

Andreas Festag TUD [email protected] Contributor

Martin Dannenberg TUD [email protected]

Contributor

Erdal Arikan Bilkent [email protected] Contributor

Ioannis Dagres IASA [email protected] Contributor

Andreas Polydoros IASA [email protected] Contributor

Roberto Verdone CNIT/UniBO [email protected] EuWIn Director



Reviewers

Full Name Beneficiary / Organisation

e-mail Date

Roberto Verdone CNIT/UniBO [email protected] 10/11/2014 Marco Luise CNIT/UniPI [email protected] 17/11/2014

Version history

Issue Date of Issue Comments 0.1 24/7/2014 Initial Toc 0.2 8/8/2014 First draft 0.3 14/10/2014 Ready for review 0.4 10/11/2014 Final version 1.0 17/11/2014 Final Revision



Executive Summary

The objective of Work-package 2.3 is to set up, maintain and operate the EuWIn facilities at CNRS/Eurecom (EuWIn@CNRS/Eurecom). The general focus is on networked signal pro-cessing for collaborative communications, development methodologies for massive radio networks in support of the Internet of Things, and technical enablers for white-space exploita-tion in the presence of a primary communication system. The main asset of the EuWIn@CNRS/Eurecom lab is the OpenAirInterface.org experimental wireless platform. The platform provides an open-source software defined radio solution for 4G networks. The software modem runs on CNRS/Eurecom’s own hardware target Ex-pressMIMO2 and since October 2014 also on the National Instruments/ETTUS B2x0 and X3x0 targets. The latter development was made possible through a close collaboration with National Instruments. During the course of NEWCOM# the OpenAirInterface platform and software modem has gained significant attention from both industry and academia inside and outside the network. This is partially due to the training and dissemination activities conducted by NEWCOM#. More details and some of the achieved results are reported in this deliverable. This delivera-ble further gives an update on the various joint research actions (JRAs) that have been de-fined in the previous deliverable (D2.3.2). The Deliverable first provides a summary of JRA achievements (described in more detail in Annex I), then provides conclusions based on the many results of year two. Annex I provides details about the JRA results. Annex II and III are in common to the three Deliverables D21.3, D22.3 and D23.3, as they refer to activities carried out in cooperation among the three sites of EuWIn, and are not reported here (see D22.3).



Table of Contents

1. Introduction ......................................................................................................... 8 1.1 Glossary ..................................................................................................................... 8 1.2 List of Joint Research Activities (JRAs) ............................................................... 10 1.3 Description of the Main WP Achievemnts in the Reporting Period .................... 10

1.3.1 Most significant results ....................................................................................... 10 1.3.2 Success Stories .................................................................................................. 11 1.3.3 Spreading of results ............................................................................................ 11

2. Detailed Activity and Achieved Results .......................................................... 12 2.1 JRA #1 Cloud RAN .................................................................................................. 12

2.1.1 Description of Activity ......................................................................................... 12 2.1.2 Relevance with the identified fundamental open issues ..................................... 12 2.1.3 Main Results Achieved in the Reporting Period and planned activities ............. 12 2.1.4 Publications ........................................................................................................ 13

2.2 JRA #2 Cellular Broadcasting ................................................................................ 14 2.2.1 Description of Activity ......................................................................................... 14 2.2.2 Relevance with the identified fundamental open issues ..................................... 14 2.2.3 Main Results Achieved in the Reporting Period and planned activities ............. 14 2.2.4 Publications ........................................................................................................ 14

2.3 JRA 3 4G/5G coexistence using spectrum overlay .............................................. 14 2.3.1 Description of Activity ......................................................................................... 14 2.3.2 Relevance with the identified fundamental open issues ..................................... 15 2.3.3 Main Results Achieved in the Reporting Period and planned activities ............. 15 2.3.4 Publications ........................................................................................................ 15

2.4 JRA 4 OAI Lab setup for joint teaching activities ................................................ 15 2.4.1 Description of Activity ......................................................................................... 15 2.4.2 Relevance with the identified fundamental open issues ..................................... 15 2.4.3 Main Results Achieved in the Reporting Period and planned activities ............. 15 2.4.4 Publications ........................................................................................................ 16

2.5 JRA 5 Multihop coding for networks ..................................................................... 16 2.5.1 Description of Activity ......................................................................................... 16 2.5.2 Relevance with the identified fundamental open issues ..................................... 16 2.5.3 Main Results Achieved in the Reporting Period and planned activities ............. 16 2.5.4 Publications ........................................................................................................ 16

2.6 JRA 6 Localization with distributed antennas ...................................................... 16 2.6.1 Description of Activity ......................................................................................... 16 2.6.2 Relevance with the identified fundamental open issues ..................................... 17 2.6.3 Main Results Achieved in the Reporting Period and planned activities ............. 17 2.6.4 Publications ........................................................................................................ 17

2.7 JRA 7 Exploiting Channel Reciprocity in MIMO TDD channels .......................... 18 2.7.1 Description of Activity ......................................................................................... 18 2.7.2 Relevance with the identified fundamental open issues ..................................... 18 2.7.3 Main Results Achieved in the Reporting Period and planned activities ............. 18 2.7.4 Publications ........................................................................................................ 18

3. General Conclusions and Prospects ............................................................... 19

4. Annex I: Detailed Description of Main technical WP Achievements ............ 20 4.1 Achievement 1: Exploiting TV white spaces with OpenAir4G ............................ 20 4.2 Achievement 2: Design of HW decoders for Polar Codes. .................................. 25



4.3 Achievement 3: 4G/5G coexistence studies ......................................................... 26 4.4 Appendix: Massive MIMO Experimental Testbed ................................................. 29


Security: Public

1. Introduction 1.1 Glossary 3GPP Third generation Partnership ACK Acknowledgement ACLR Adjacent channel leakage ratio ADC Analogue digital convertor AM Acknowledged Mode AMBA Advanced Microcontroller Bus Architecture ASIC Application Specific Integrated Circuit AVCI Advanced VCI BB Baseband BCH Broadcast Channel CCCH Common Control Channel CFI Control Format Indicator CPU Central Processing Unit CQI Channel Quality Indicator CRC Cyclic Redundancy Code CTS Clear to send DAB Digital Audio Broadcast DAC Digital analogue convertor DCCH Dedicated Control Channel DCI Downlink Control Information DD Digital Dividend DDR Double data rate DLSCH Downlink Shared Channel DMA Direct memory access DoW Description of Work DSP Digital Signal Processor DTCH Dedicated Transport Channel EC European commission eNB Enhanced Node B EuWIn European Laboratory of Wireless Communications for the Future Internet FDD Frequency division duplex FE Front-end FFT Fast Fourier Transform FPGA Field programmable gate array GPIO General Purpose Input Output GSM Global System for Mobile Communications HARQ Hybrid Automated Repeat Request HI Hybrid ARQ indicator IEEE Institute of Electrical and Electronics Engineers IoT Internet of Things


Security: Public

IP Internet Protocol JRA Joint Research Activities LNA Low noise amplifier LTE Long term evolution LXRT Linux Real Time MAC Medium access control MIMO Multiple-Input Multiple-Output NAK Non-Acknowledgement NAS Non Access Stratum NFS Network file system NITOS Network Implementation Testbed using Open Source code OAI OpenAirInterface OEDL OMF Experiment Description Language OFDM Orthogonal Frequency division multiplexing OMF cOntrol and Management Framework OS Operating System OSD OpenAir Scenario Descriptor PA Power amplifier PBCH Physical Broadcast Channel PC Personal Computer PCFICH Physical Contrl Format Indicator Channel PCI Peripheral Component Interconnect PCIe PCI express PDCCH Physical Downlink Control Channel PDCP Packet Data Convergence Protocol PDSCH Physical downlink Shared Channel PDU Packed Data Unit PHICH Physical Hybrid-ARQ Indicator Channel PHP Hypertext Preprocessor PHY Physical PMI Precoding Matrix Indicator PRACH Physical Random Access Channel PSS Primary Synchronization Signal PUSCH Physical Uplink Shared Channel RA Random Access RB Resource Block RF Radio frequency RI Rank Indicator RLC Radio Link Controller RRC Radio Resource Controller RRM Radio Resource Manager RTAI Real-time Application Interface RTS Ready to send RX Receiver


Security: Public

SIMD Single Instruction Multiple Data SMA SubMiniature version A SNR Signal to Noise Ratio SRS Sounding Reference Signal SSS Secondary Synchronization Signal SVN Subversion TB Transport Block TDD Time division duplex TM Transmission Mode TVWS Television white space TX Transmitter UCI Uplink Control Information UE User Equipment ULSCH Uplink Shared Channel UM Unacknowledged Mode UMTS Universal Mobile Telecommunications system VCI Virtual Component Interface WCDMA Wideband Code division multiple access WiMAX Worldwide Interoperability for Microwave Access 1.2 List of Joint Research Activities (JRAs)

• JRA #1 Cloud RAN • JRA #2 Cellular Broadcasting • JRA #3 4G/5G coexistence using spectrum overlay • JRA #4 OAI Lab setup for joint teaching activities • JRA #5 Multihop coding for networks • JRA #6 Localization with distributed antennas • JRA #7 Exploiting Channel Reciprocity in MIMO TDD channels

1.3 Description of the Main WP Achievemnts in the Reporting Period 1.3.1 Most significant results

• A batch of 20 new ExpressMIMO2 cards was delivered to CNRS/Eurecom. Several cards were further sold to partners, such as IASA and Alcatel-Lucent Bell Labs Nozay (Paris).

• In terms of JRAs, the most active ones were o JRA#1 on cloud RAN, which produced 2 joint papers and led to the integra-

tion CNRS/Eurecom-INOV work on CRAN transport mechanisms in OAI. Moreover an exchange student from CNIT/PoliTO visited Bilkent for joint work on efficient polar decoding architectures.

o JRA#3 on 4G/5G coexistence, which resulted in a joint experiment between CNRS/Eurecom and TUD.

o JRA#6 on localization, where measurements collected with OpenAirInterface were exploited (publication pending).

o JRA#7 on exploiting channel reciprocity, which produced also 1 paper. • WP2.3 showed a strong presence at EUCNC 2014 in Bologna with


Security: Public

o An exhibition booth showing for the first time the OpenAirInterface eNB com-municating with an off-the-shelf UE

o A presentation of Joint IASA-CNRS/Eurecom localization measurement JRA o A presentation of Joint INOV-CNRS/Eurecom-Bilkent-CNIT/PoliTO CRAN

JRA 1.3.2 Success Stories

• CNRS/Eurecom has initiated contact with National Instruments regarding the integra-tion of OpenAirInterface.org software on the newest generation of USRP platforms. National Instruments is primarily interested in exploiting OAI’s open-source develop-ment to provide a large user base in the USRP community and in particular to help promote the development. Both NI and CNRS/Eurecom will demonstrate OpenAir-Interface LTE eNB running on the Ettus/NI B210 platform at the Wireless Innovation Forum in November 2014 in Rome.

• CNRS/Eurecom has been officially integrated as a partner in Mobile Cloud Network

FP7 project to strengthen work on CRAN architectures. This will lead to more joint work within N# with Inov and Orange.

• OAI equipment currently being used in the OFCOM TV White Space Trial in London,

England, organized by KLC (Associate Partner Type II)- 1.3.3 Spreading of results

• OAI’s LTE basestation software and ExpressMIMO2 were demonstrated by Agilent and IBM China at the Mobile World Congress in Barcelona (Feb. 24-27) as an ena-bling technology for CloudRAN.

• Integration of OAI’s LTE basestation software with two industrial core network (EPC) implementations has be inititated. Firstly, with Alcatel-Lucent’s “LTEBOX” which is a single-server EPC solution. Secondly, the German network testing company, NG4T, has provided their EPC solution for the purpose of evaluating OAI’s basestation im-plementation.

• CNRS/Eurecom has also recently performed interoperability testing of the OAI LTE basestation with the French company, ERCOM, who provides a full-RF synthetic test-ing environment for stimulating commercial basestations with emulated terminals. This will allow for future development on abstraction methodologies for real-time test-ing systems, in particular for interference scenarios in hetnets.

• Presentation of WP2.3 JRAs at 3 dissemination events (NEC Heidelberg, Alcatel-Lucent Bell Labs Stuttgart, AVEA Labs Istanbul)

• Membership of OAI team within the NGMN (Next Generation Mobile Networks) asso-ciation.

• Alcatel-Lucent Bell Labs New Jersey now a user of OAI in addition to Bell Labs – Par-is

• China Mobile to publicise the use of OAI in CRAN trial in Shanghai. Regular interac-tions with Agilent and China Mobile of integration of OAI in their CRAN demonstrator using commercial remote radio-head technologies

• Discussions with IBM Watson Laboratory in New York for collaboration around OAI emulation methodologies


Security: Public

2. Detailed Activity and Achieved Results 2.1 JRA #1 Cloud RAN CNIT/PoliTO, CNRS/Eurecom, Bilkent, INOV.

2.1.1 Description of Activity This JRA is a joint effort between CNIT/PoliTO, CNRS/Eurecom, Bilkent, and INOV. Some industry collaboration notably with Alcatel-Lucent (France, Germany) and members of the Chinese C-RAN project (Agilent, Orange, IBM), as well as collaboration with the FP7 Mo-bileCloud Networking project should also be mentioned, although for the latter they are out-side the scope of Newcom#, both the project and the external partners benefit from the de-velopment that is injected into OAI. The main motivation for this work is twofold. First to pro-vide an experimental all software CRAN solution using only x86-based processing. Secondly we provide a framework for studying the feasibility of using hardware accelerators for the channel decoders in software defined radios, with target architectures CRAN or basestation pooling applications. 2.1.2 Relevance with the identified fundamental open issues Cloud RAN (also called basestation pooling) is one of the emerging arcitectures for 5G net-works. It aims to simplify a radio network by centralizing signal processing in servers, much like data centers in the internet. One of the potential scenarios is a basestation server in a downtown core driving several so-called remote-radio heads (RRH) from a common pro-cessing center. Some could be small-cells and others macro-cells. Yet another solution would be to centralize signal processing at macro-cell locations (i.e. in the technical room below a cell-tower) where all small-cells co-located with the macro-cell are also processed. Both architectures rely on efficient optical networks using transport protocols such as CPRI or OBSAI to relay I/Q samples to from the server machines. The RRH is a key enabling technology for such systems and is now almost a commodity piece of hardware in the cellu-lar industry. It contains all the RF circuitry as well as A/D and D/A converters driv-ing/receiving from the transport medium. The main benefit from this architecture is to allow the computational resources (and hence power consumption) to be optimally matched to the spatio-temporal statistics of traffic in a fairly large geographic area. Traditional cell sites (macro or small-cell) are dimensioned with sufficient computation resources to handle a fully-loaded cell, even if the cell is only partially loaded at any given time, thus wasting computation resources and consequently power. Moreover, since signal processing is centralized it allows for more sophisticated spatio-temporal processing of radio signals which could dramatically increase spectral-efficiency (i.e. through distributed or network MIMO which is also known as cooperative multipoint – CoMP transmission and reception). 2.1.3 Main Results Achieved in the Reporting Period and planned activities One of the research topics addressed witih NEWCOM# is the instantiation of software-based base band units (BBUs) on virtual machines of data-centres, following the cloud paradigm [1]. This raises several challenges, as the tight real-time performance requirements of base stations that must be respected in a virtualised environment. The LTE base station emulator OpenAirInterface is used to evaluate the feasibility of such challenges, when instantiated on virtual machines of a public cloud-provider. The performance of virtual machines is evaluated when a base station is configured to provide specific radio resources and loaded with typical traffic usage profiles. Its behaviour and stability is also evaluated, when instantiated in a shared infrastructure of a public cloud-provider. It is concluded that BBU processing is de-pendent on many factors. The larger the system bandwidth (from 5 to 20 MHz), or the modu-lation coding scheme (from QPSK to 64QAM) or the amount of used resource blocks (from


Security: Public

one to all), the heavier are the processing needs of the BBU. Reception is more processing consuming than transmission, being decoding at PHY layer the heaviest one. At the MAC layer, processing increases with the number of users. Still, one of the issues is the fact that OAI’s PHY layer is not multi-threaded. On the other side, running on a public cloud of shared resources results in large processing fluctuations, which is undesirable for this kind of sys-tem. Another ongoing research, that aims to extend the previous one is the ethernet transport of IQ samples. Such an interface is under development for the existing lte-softmodem software from CNRS/Eurecom. It aims to enable the transport via ethernet of IQ samples from a base station running on a specific machine to another machine where na antenna or users are available. This raises several challenges of synchronisation, besides the high needed speeds to respect the intrinsic LTE system time constrains. This will enable to evaluate the perfor-mance of a real BBU instantiated on a virtual environment, a proof of concept tha still raises several issues in terms of the realtime processing needs these systems require. To be noticed further is that the cooperative research of INOV and CNRS/Eurecom around OpenAirInterface lead to the adoption of this tool in the FP7 MobileCloud Networking (MCN) project where INOV is partner. This tool has been integrated and explored by a large re-search consortium involving actively University of Bern, Orange France, OneSource and Cloudsigma in the development of new features and evaluation of its performance in chal-lenging cloud environments. This strong cooperation also lead to the recent invitation by MCN to CNRS/Eurecom join the consortium. This evidences the strong impact that NEW-COM# may have in building and extending research communities within Europe. Regarding the second objective of this JRA, hardware accelerators for the decoding of Polar codes have been developed in a joint effort between BILKENT and CNIT-POLITO. The de-velopment includes a hardware oriented C model has been derived, which closely follows the planned decoder architecture. The new model includes quantization, storage and arithmetic details that allow for direct comparison of software and hardware simulation results. In paral-lel, a VHDL model has been produced describing a fully parallel decoder architecture, target-ing ASIC implementation; the model incorporates several parameters, which enable the hardware tuning according to the results of simulations from C code. The synthesis of the fully parallel architecture has been performed by means of Synopsys Design Compiler, and the obtained circuit has been characterized in terms of throughput, occupied Silicon area, working frequency and power consumption. The results achieved are promising (overcoming prefixed target) so this architecture have been chosen as basic structure for the following developments. Subsequently, the work followed two parallel branches. The first one aims at mapping a properly refined and optimized version of the obtained VHDL design onto an FPGA board. The second branch has to do with the design of a reduced-complexity architecture, with some further architectural refinements and able to save implementation complexity with lim-ited effects on communication performance. The final achieved results will be compared against published implementations available in the literature. The designed decoder and the collected comparisons will form the core of a new joint paper that will be submitted to an in-ternational journal for review. 2.1.4 Publications

[1] Dimitrova, Desislava; Alfayawi, Islam; Ferreira, Lucio S.; Gomes, André; Nikaein, Navid; Georgiev, Alexander; Pizzinat , Anna, “Challenges ahead of RAN virtualization in LTE,” EUCNC 2014, European Conference on Networks and Communication, Workshop 5 Mobile Cloud Infrastructures and Services, June 23-26, 2014, Bologna, Italy


Security: Public

[2] Andrea Biroli, and Guido Masera, "A VLSI Implementation of the Belief Propagation Algorithm Applied to the Decoding of Polar Codes", EuCNC2014, European Confer-ence on Networks and Communications, Bologna 26th June 2014

2.2 JRA #2 Cellular Broadcasting CNRS/Supelec, CNRS/Eurecom, UCAM, KCL

2.2.1 Description of Activity Multimedia Broadcast Multicast Services (MBMS) is a point-to-multipoint interface specifica-tion for existing and upcoming 3GPP cellular networks, which is designed to provide efficient delivery of broadcast and multicast services. The topic is clearly multidisciplinary, but N# knowledge can have a large impact on the long term evolution of the topic. In WP23 we are mainly interested in the experimental aspects of such scenarios and have thus initiated a collaboration with the ICT-ACROPOLIS NoE. Under the lead of Kings College London, they have submitted an application to the English regulator OFCOM to use TVWS spectrum to trial different TVWS devices and services. Among them is also CNRS/Eurecom’s OpenAirInterface platform, which will be used to trial LTE eMBMS. For more detail see the Appendix 2.2.2 Relevance with the identified fundamental open issues Cellular broadcasting is one of the key technologies for 4G and 5G. It provides scalable ac-cess for any number of users, which is interesting for very dense scenarios like stadion events or similar. The main services will be video streaming, but are not limited to it. Other services might include software updates, news, and others. 2.2.3 Main Results Achieved in the Reporting Period and planned activities CNRS/Eurecom participated in the OFCOM TVWS trials by providing their experimental equipment. This equipment includes an ExpressMIMO2 card with an additional RF frontent to provide the required performance. At the time of writing, the trails are still ongoing and final results will be published next year. 2.2.4 Publications

[1] B. Zayen, F. Kaltenberger, and R. Knopp, Opportunistic Spectrum Sharing and White Spa-‐ce Access: The Practical Reality. Wiley, 2015, ch. OpenAirInterface and ExpressMIMO2 for spectrally agile communication.

2.3 JRA 3 4G/5G coexistence using spectrum overlay TUD, CNRS/Eurecom

2.3.1 Description of Activity 5G networks will very likely include a new waveform that will more easily enable dynamic spectrum access (DSA) by exploiting spectrum holes. At the same time, the coexistence with existing 4G networks needs to be guaranteed. TUD has developed a new waveform, called Generalized Frequency Division Multiplexing (GFDM) that has favourable properties com-pared to OFDMA and SC-FDMA waveforms of 4G systems. In this JRA, TUD and CNRS/Eurecom conducted joint experiments to evaluate the impact of the new GFDM waveform on an existing 4G system. The 4G system was based on CNRS/Eurecom’s OpenAirInterface for the eNB and a commercial UE. The 5G system was emulated using the PXI Platform with corresponding RF adapter modules from National In-struments and TUD’s GFDM implementation. The detailed experimental setup is described in the Appendix 4.3.


Security: Public

2.3.2 Relevance with the identified fundamental open issues In order to alleviate the spectrum scarcity due to the high bandwidth demands from new de-vices and applications, cellular network operators may deploy a Dynamic Spectrum Access (DSA) overlay. In such a scenario, an operator opportunistically uses spectrum as a second-ary user to augment its spectrum holding. While DSA is an important direction to increase the spectral efficiency, research for new waveforms is carried out for 5G systems. These wave-forms address drawbacks for OFDM used in 4G LTE-based systems, such as OFDM’s need for frequency orthogonality and time synchronicity. One of the approaches for new wave-forms is Generalized Frequency Division Multiplexing (GFDM) a digital multi-carrier trans-ceiver concept that employs pulse shaping filters to provide control over the transmitted sig-nal’s spectral properties, a cyclic prefix that enables an efficient FFT-based frequency do-main equalization scheme, as well as tail biting as a way to make the prefix independent of the filter length. One specific GFDM feature is the ultra-low out-of band radiation due adjust-able Tx-filtering. This feature makes GFDM attractive to be used by the secondary user in a DSA overlay network, where the ultra-low out-of band radiation minimizes the impact of the secondary system on the primary system. As such, DSA with the new waveform enables coexistence between 4G and 5G systems. 2.3.3 Main Results Achieved in the Reporting Period and planned activities Joint experiment between TUD and CNRS/Eurecom performed in October 2014. Results will be submitted to a conference very soon. 2.3.4 Publications n/a

2.4 JRA 4 OAI Lab setup for joint teaching activities CNRS/Eurecom, IASA

2.4.1 Description of Activity This JRA has been created for CNRS/Eurecom to support IASA in the setup of a lab based on the OpenAirInterface and the ExpressMIMO2 platform. IASA is affiliated with the Physics Department of University of Athens, where the lab is been host. The main objective of this activity is to enhance the teaching activities in the University of Athens in the area of wireless communications systems and networks, by providing a tool for experimentation and educa-tion. The joint work of the two institutions comprise the setup of the lab, with all the related supporting actions, the creation of educational material, and the establishment of some basic experimental activities for the education of IASA’s personnel to the new tool. 2.4.2 Relevance with the identified fundamental open issues One of the fundamental objectives of Newcom# are the transfer of knowledge and teaching. The benefit is twofold:

• IASA personnel will be trained on using OpenAirInterface, an open/source soft-ware/hardware platform designed for innovation in the area of digital communications. This will be the first hardware platform owned by IASA and it is our belief that it will contribute substantially to increase its competiveness.

• New engineers and scientists will be educated and trained on a quite promising area.

2.4.3 Main Results Achieved in the Reporting Period and planned activities


Security: Public

IASA has acquired two ExpressMIMO2 cards from CNRS/Eurecom, which have been deliv-ered in spring 2014. During the summer, with some remote support from CNRS/Eurecom, IASA setup the necessary equipment in their lab and installed the ExpressMIMO2 cards. Up to now some basic experimental tests were performed to verify the functionality of the cards. IASA is also in the process of implementing some needed software tools in order to be able to use seamlessly the platform with Matlab. The next steps will be the following

• IASA with the remote help of CNRS/Eurecom will be trained to use the most im-portant tools of the OpenAirInterface platform.

• IASA with the remote help of CNRS/Eurecom will setup some basic experimental ac-tivities for IASA’s personnel training.

• CNRS/Eurecom and IASA will setup a lab program suitable for the students of the communication sector within the Physics Department as well as for the students of the master program in radio communications.

2.4.4 Publications n/a

2.5 JRA 5 Multihop coding for networks CNRS/Eurecom, IASA

2.5.1 Description of Activity The purpose of this planned activity is to exploit the existing OAI collaborative communica-tions software development left over from the LOLA/SHARING/CONECT projects to derive experiments pertinent to ground communication networks. In particular we will aim to exper-iment with so-called “barrage networking” concepts, in particular to provide collaborative mul-ticasting services. 2.5.2 Relevance with the identified fundamental open issues Mesh networks are particulary interesting for public safety, especially in disaster recovery scenarios. 2.5.3 Main Results Achieved in the Reporting Period and planned activities This JRA has been dormant for this reporting period and might be removed next year. 2.5.4 Publications n/a

2.6 JRA 6 Localization with distributed antennas IASA, CNRS/Eurecom

2.6.1 Description of Activity Single source localization, is one problem with extremely large literature due to their applications and practical use. We focus at the localization based to Received Signal Strength (RSS) from a sensors network. First, jointly IASA and CNRS/Eurecom put some effort to find theoretically, if we can improve our localization results by using past measure-ments from known position. The claim is true due to the spatial correlation of Shadow fading. So we focus on analysing the performance of RSS based localization, in a correlated log-


Security: Public

normal environment with use of past measurements. The main goal is to answer the follow-ing questions: a) what is the density (sensors/m2) of the real-time measurement network for a target localization performance? b) How much is this density reduced when utilizing high density-past measurements? c) How these two densities (real time and past measurements) are related given a propagation environment parameterization? At next, we made an experimental campaign based on CNRS/Eurecom’s OpenAirInterface system to verify experimentally our results. So with OpenAirInterface platform, we measure the RSS of one Transmitter at a tough indoor environment in order to assess the gains of the spatial correlation. After the collection of the measurements we use two different approaches to model the correlation (Voronoy and Interpolation) of the shadow fading and we compare the results with the uncorrelated one. We use Maximum Likelihood (ML) estimator in all cas-es. 2.6.2 Relevance with the identified fundamental open issues Localization is an important feature of 4G and 5G networks with many different applications. Indoor localization is a non-trivial problem, since coverage of Global Navigation Satellite Sys-tem (GNSS) is usually not given. This JRA explores methods for single and multi-source in-door localization using simple RSS measruements and evaluates them experimentally. 2.6.3 Main Results Achieved in the Reporting Period and planned activities The main results of the experimental procedure are displayed in Table 1. It is clear, that the use of past measurements improves the performance. The constant interpolation, although better than the uncorrelated case, does not give the best performance, possibly due to un-derfitting. The simple Voronoi approach, provides a fairly good performance in comparison with the two other interpolation methods (Linear and Quadratic). Lastly, we see that the Quadratic model gives worst results than the Linear, something that could be accounted to overfitting

Table 1: Mean Error for all models Uncorrelated

log-normal Constant interpolation

Linear inter-polation

Quadratic interpolation

Voronoi

All points 4.16 2.95 2.02 2.55 2.15 Excluding 1 worst

3.6 2.68 1.8 2 2.06

Excluding 2 worst

3.1 2.48 1.64 1.88 1.97

These results have also been presented orally [1] and will be submitted to a conference soon. As further activities we have to use the extensive past measurement campaigns that exist in this field. Study them under a common spatial correlation model. 2.6.4 Publications [1] George Arvanitakis, Ioannis Dagres, Florian Kaltenberger, Andreas Polydoros, Adrian

Kliks “RSS based localization: Theory and experimentation”, EUCNC 2014, European Conference on Networks and Communication, Special Session, From Theory to Practice: Experimental Research Activities in NEWCOM#’s EuWIn Labs, June 23-26, 2014, Bolo-gna, Italy


Security: Public

2.7 JRA 7 Exploiting Channel Reciprocity in MIMO TDD channels CNRS/Eurecom, VUT, Linköping University (Associate Partner Type II)

2.7.1 Description of Activity The performance of MIMO systems relies to a great extent on the available channel state information at the transmitter (CSIT). In FDD systems, this CSIT is obtained by feedback and is therefore subject to low resolution and delay. In TDD systems, channel reciprocity can be exploited to infer CSIT from the uplink channel. Especially multi-user MIMO and massive MIMO systems rely to a great extent on the exploitation of channel reciprocity to gain CSIT. However, while the physical radio channel is reciprocal, the effects of the radio frequency circuits is not and must be calibrated. In this JRA several avenues of reciprocity are investigated. In the first activity, which is most-ly joint work between CNRS/Eurecom and VUT, we investigate the estimation of channel reciprocity parameters in the presence of frequency. In the second activity, which is mostly joint work between CNRS/Eurecom and Linköping university, we investigate the effect of mutual coupling and cross-talk on the estimation of channel reciprocity compensation matrix. For this work a new simplified methodology to acquire bidirectional channel measurements with a multiple ExpressMIMO2 cards is used. The cards are perfectly synchronized and thus we avoid the problematic of frequency offsets. 2.7.2 Relevance with the identified fundamental open issues One of the key technologies for 5G systems is massive MIMO. These systems need to ex-ploit channel reciprocity to function properly. However, practical implementations that exploit this reciprocity are not yet fully developed and require new ideas and experiments. 2.7.3 Main Results Achieved in the Reporting Period and planned activities CNRS/Eurecom started to build a demonstration platform for massive MIMO TDD in the be-ginning of 2014. This plaftform can house up to 16 ExpressMIMO cards, providing up to 64 independent RF chains. Today we are able to drive 4 cards (16 chains) and we plan extend to the full scale until the end of 2015. A more detailed description of the platform is given in the appendix. Some initial results have also already been achieved and are described in the publication [1]. 2.7.4 Publications [1] Jiang, X.; Cirkic, M.; Kaltenberger, F.; Larsson, E. G.; Deneire, L. & Knopp, R., “MIMO-

TDD Reciprocity and Hardware Imbalances: Experimental Results”, ICC 2015, submitted


Security: Public

3. General Conclusions and Prospects In general, the workpackage is very active with 6 active JRAs (and one dormant JRAs). Sev-eral joint papers were produced and some papers are in the pipeline. Moreover, the OpenAir-Interface platform, which is the core product of the EuWIn@Euracom lab, is gaining signifi-cant attention, both within the network and outside (academic and industry).

OpenAirInterface currently provides a standard-compliant implementation under a GNU GPLv3 license of a subset of Release 10 LTE for UE, eNB, MME, HSS, SGw and PGw on standard Linux-based computing equipment (Intel x86 PC architectures). It can be used in conjunction with standard RF laboratory equipment available in many labs (i.e. National In-struments/Ettus USRP and PXIe platforms) in addition to custom RF hardware provided by CNRS/Eurecom to implement these functions to a sufficient degree to allow for real-time in-teroperation with commercial devices. Some industrial users have working OpenAirInter-face-based systems integrated with commercially-deployable remote radio-head equipment and have provided demonstrations at major industrial tradeshows (Mobile World Congress Asia 2014, Mobile World Congress Barcelona in 2013, IMIC 2013). The current major indus-trial users of OpenAirInterface for collaborative projects are Agilent, China Mobile, IBM, Al-catel-Lucent, Thales, National Instruments and Orange. The primary future objective is to provide an open-source reference implementationthat follows the 3GPP standardization pro-cess starting from Rel-12 and the evolutionary path towards 5G and that is freely available for experimentation on commodity laboratory equipment.

The output of this WP2.3 will help extend OpenAirInterface towards the definition of 5G sys-tems. In particular the open-source policy will hopefully help to drive innovation in 5G by fol-lowing the standard as it is being drafted and to leverage the crowdsourcing effect both from industrial and academic users. To this end, the JRAs #1 (cloud RAN), #3 (4G/5G coexist-ence), and #7 (exploiting TDD reciprocity) are a first step in this direction. The resulting de-velopment can be used in both publicly funded collaborative projects as well as industry-driven initiatives aiming to demonstrate 5G features at the earliest possible stage. Moreover, the results can be replicated in several locations independently through the combination of open-source and commodity hardware. This then becomes a truly distributed experimental facility with a very large number of potential contributors.


Security: Public

4. Annex I: Detailed Description of Main technical WP Achievements Detailed descrirption of what mentioned in Section 1, one by one

4.1 Achievement 1: Exploiting TV white spaces with OpenAir4G1 TV White Spaces technology is a means of allowing wireless devices to opportunistically use locally-available TV channels (TV White Spaces - TVWS), as long as the devices are certi-fied as communicating directly with a geolocation database, implementing the channel/power usage instructions sent from the geolocation database, take into account security considera-tions, and comply with requirements such as achieving their stated or given spectrum mask, among others. In this section we explore how OpenAirInterface can be used to as an experimental platform for communication in TVWS. In Section Errore. L'origine riferimento non è stata trovata. e review the requirements on TVWS devices in terms of the spectral mask, which is one of the most difficult challenge in TVWS communications. In Section Errore. L'origine rifer-imento non è stata trovata. we promote the use of LTE as an air interface for TVWS com-munications. Finally in Section Errore. L'origine riferimento non è stata trovata. we show ow the OpenAirInterface testbed has been adapted to be used as a TVWS communication device. Challenges in RF Design The primary issue for a TVWS transmitter is the emission mask which is an indication of the power emitted in the channel of interest as well as the leakage in adjacent channels. This is imposed by the regulator in each country or region. For TVWS operation, we provide two examples which are quite different, the regulations of the United States’ FCC (Federal Com-munications Commission) and the United Kingdom’s OFCOM (Office of Communications). FCC requirements The FCC specifies absolute channel power limits and adjacent channel leakage ratio (ACLR) for fixed and portable WSD’s. The requirements are quite stringent. Fixed WSD’s are limited to 1 Watt/30 dBm (6dBi antenna gain) with adjacent channel power limited to -42.8 dBm (in all adjacent channels) which results in an ACLR of 72.8 dB. This is extremely challenging from an RF design perspective if channel filters are not used at the output of the power am-plifier. For portable WSD’s, transmit powers are limited to 100 mW/ 20dBm when no adja-cent channel is present and 40 mW / 16 dBm with an adjacent channel. Adjacent channel powers are limited to -56.8 dBm which results in ACLRs of -76.8 dB and -72.8 dB. For a low-power/low-cost device these leakage ratios are very difficult to achieve with current technol-ogy. To put this in the perspective of high-end terminal electronics, an ACLR of 36dB (-25 dBm per 1MHz in the adjacent channel) with a 200 mW/23 dBm transmit power is the re-quirement of a 4G smartphone. Similarly for a basestation, the ACLR is required to be at least 45 dB in the adjacent channel. OFCOM Requirements 1 The following text has also been accepted for publication in “Opportunistic Spectrum Sharing and White Space Access: The Practical Reality”, Wiley, 2015


Security: Public

OFCOM does not specify absolute powers but rather classes with different adjacent channel leakage. Adjacent channel leakage is defined based on measurement bandwidths of 100kHz. The relationship between device classes and adjacent channel frequency leakage ratio (AFLR) is shown in Table 2, where powers are read according to the device class and the adjacent channel index.

Adjacent channel

Class 1 Class 2 Class 3 Class 4

n+1 74 dB 74 dB 64 dB 54 dB n+2 74 dB 74 dB 74 dB 64 dB n+3 84 dB 74 dB 84 dB 74 dB

Table 2: OFCOM Adjacent Channel Frequency Leakage Ratios (AFLR) The WSD must be able to guarantee that power in every 100 kHz measurement band of the adjacent channels falls below the in-band power indicated by the database minus the AFLR from Table 2. We can convert the table entries to an adjacent channel leakage ratio (ACLR) measured over the whole 8MHz channel by removing 10 log!" 8MHz 100kHz = 19dB from each entry. This yields 35 dB ACLR for Class 4 and 45dB for Class 3. These are practically equivalent to a 4G terminal and basestation respectively. For Class 1-2 we require a 55 dB ACLR. We note that, overall, these requirements are significantly less difficult to achieve than the FCC requirements. LTE for TVWS As shown before, one of the most difficult issues with TVWS transmission at reasonably high-power is the cost that comes with dynamic spectrum access and out-of-band emission requirements. For this reason, it seems reasonable to assume that mobile transmitting de-vices in TVWS would be less attractive than receive-only devices. Moreover, it may be sim-pler to use TVWS spectrum as it is used for television broadcast, namely a small number of high-power transmitters. In this case, we would use a broadcast or multicast transmission and make use of applications which exploit this mode of transmission. For instance, if end devices are smartphones, specifically designed applications which automatically store con-tent in a local cache memory based on user-defined criteria and allow users to retrieve the information from the cache instead of the network. This is essentially applicable to delay-tolerant content retrieval. Transmitters can also be co-located with content servers whose information caters to the preferences of the local population. They could use a combination of broadcast carriers and DL-only component carriers which are controlled by a primary car-rier in licensed bands. A potential air-interface technology that supports the mentioned scenarios is 3GPP LTE and, in particular, Rel-10/11 LTE-Advanced which allows for dynamic spectrum allocation. The master WSD’s would be equivalent to small-cell basestations (eNodeB) and the slave WSD’s would be terminals (UEs) equipped with TVWS radios. LTE-Advanced provides a sophisti-cated protocol (in comparison to 802.11af) for managing spectrum. Up to 5 carriers arbitrari-ly allocated in frequency are controlled by the so-called Primary-Component Carrier. The latter carries all layer-3 signalling for the UEs in the cell. The Secondary-Component Carriers can be TDD or FDD or downlink-only, or even multicast/broadcast only. Layer 2 signaling for scheduling can be transmitted either on the primary or secondary carrier arbitrarily (using so-called cross-carrier scheduling). The multicast/broadcast transmission format, or enhanced Multimedia Broadcast Multicast Service (eMBMS), is yet to be commercially-deployed anywhere in the world, although trials have taken place in most markets and by many operators. The two main United States’ op-erators, AT&T and Verizon, have recently committed to deploying eMBMS in 700 MHz spec-


Security: Public

trum. This is important since the deployment scenarios and applications could be similar to those envisaged for TVWS transmission. Their primary application is offloading video deliv-ery to multicast carriers using eMBMS. Carrier aggregation could be used to make optimal use of TV channels as shown in Table 3. However, in order to fulfil the requirements on spectral masks described above, it is also fea-sible to use less than that. For example, the TVWS solution based on OpenAirInterface (see below) uses a single 5MHz carrier in a single TV channel.

Channel Configuration Carrier Configura-tion

Bandwidth

Single TV channel 5 MHz + 2.5 MHz 6.75 MHz 2 Contiguous TV channels 15 MHz 13.5 MHz 3 Contiguous TV Channels 20MHz + 5 MHz 22.5 MHz 4 Contiguous TV Channels 20MHz + 10 MHz 27 MHz

Table 3:LTE Channel configurations for TVWS operation Interconnection with the geo-location database would be possible in the resource manage-ment entity connected to the eNodeB which feeds the basic signalling of the cell on the pri-mary component carrier. If used in conjunction with licensed primary carriers the dynamic switching of a secondary (TVWS) carrier would be normal functionality in LTE-Advanced. Moreover, all the signalling mechanisms for requesting UEs to perform signal strength measurements in adjacent bands are also normal functionality. OpenAirInterface TVWS Solution For the reasons mentioned above the OpenAirInterface TVWS solution uses LTE as an air-interface technology together with eMBMS. The target architecture of the fixed network infra-structure is shown in Figure 1. It comprises a single-PC for the eNB which houses an Ex-pressMIMO2 and which can drive both a single-TV channel in TVWS with eMBMS and po-tentially a second carrier with TD-LTE for a mock licensed-band2. Initially, we will only con-sider the eMBMS/TVWS carrier. Further we use a single carrier of 5MHz in a 8MHz TV channel to ease the requirements on the spectral mask. The PC is interconnected with a TVWS database to acquire channel and power information.

Figure 1: Network Infrastructure Elements

2 At the time of writing the carrier aggregation functionality is located in a branch of openair4G but integration with the main trunk is planned.


Security: Public

The eNB PC is interconnected with a second PC housing a small-scale Enhanced Packet Core (EPC) which is part of the OpenAirInterface.org development. Finally, an application server is connected to the EPC to provide traffic to the TVWS carrier. At the terminal side either ExpressMIMO2 with OpenAirInterface software modem or com-mercial radios (i.e., LTE dongles) can be used. In terms of equipment, this scenario requires deployment of a high-power master WSD with an interconnection to a white-space database. We have proposed to use an OAI eNB with ExpressMIMO2 coupled with a high-power amplifier. The high-power amplifier is part of a dual-antenna TD-LTE front-end module shown in Figure 2. The module contains a transmit power amplification chain which can provide two antenna ports at 43 dBm (20 Watts) with a 3GPP-LTE compliant spectral mask. In order to be granted approval by OFCOM to deploy the equipment, we must guarantee compliance with the regulatory constraints.

Figure 2: Internal View of Front-end Module To this end, consider the power spectrum measurements of the front-end module shown in Figure 3 and Figure 4. These were generated using an LTE signal generator and the front-end module. The module is capable of supporting OFCOM class 3 with a 1 watt output and OFCOM class 1-2 with 100mW output. In addition to the adjacent channel leakage due to the power amplifier, spurious emissions from the baseband generation circuits can also generate undesirable tones in the TVWS band. This is the case of the RF ASICs that are used on the ExpressMIMO2 board whose phase-locked-loops (PLL) generate tones throughout the band. Although the quality of the signal is sufficient for broadband signal generation according to LTE requirements at UHF frequencies, it is insufficient in many TV channels. As a result we have opted for using Ex-pressMIMO2 to generate the TVWS signal at a higher, fixed frequency (1.5 GHz), which is appropriately filtered using an RF SAW bandpass filter, prior to being downconverted to the TVWS channel that is used for transmission. This guarantees spurious-free emissions in the TVWS bands prior to amplification by the front-end module.


Security: Public

Figure 3: Front-end output in TVWS (30 dBm, OFCOM Class 3)

Figure 4: Front-end output in TVWS (21 dBm, OFCOM Class 1-2)

1 RMAVG

A

*

Offset 17 dB

LVL

**

3DB

RBW 50 kHzVBW 1 MHzSWT 20 msRef 30 dBm Att 20 dB

Center 731.5 MHz Span 50 MHz5 MHz/

OVTRC

-60

-50

-40

-30

-20

-10

0

10

20

Tx Channel E-UTRA/LTE Square/RRC Bandwidth 4.5 MHz Power 30.69 dBm

Adjacent Channel Bandwidth 4.5 MHz Lower -50.78 dB Spacing 5.5 MHz Upper -51.16 dB

Alternate Channel Bandwidth 4.5 MHz Lower -70.90 dB Spacing 11 MHz Upper -71.04 dB

Date: 7.JAN.2014 12:17:40

Class 3 Limit (-37 dBm/50 kHz)

1 RMAVG

A

*

Offset 17 dB

LVL

Ref 20 dBm Att 10 dB

Center 731.5 MHz Span 50 MHz5 MHz/

**

3DB

RBW 100 kHzVBW 1 MHzSWT 20 ms

OVTRC

-70

-60

-50

-40

-30

-20

-10

0

10

Tx Channel E-UTRA/LTE Square/RRC Bandwidth 4.5 MHz Power 21.11 dBm

Adjacent Channel Bandwidth 4.5 MHz Lower -57.55 dB Spacing 5.5 MHz Upper -57.54 dB

Alternate Channel Bandwidth 4.5 MHz Lower -69.73 dB Spacing 11 MHz Upper -69.73 dB

Date: 7.JAN.2014 12:18:31

Class 1-2 Limit (-56 dBm/50 kHz)


Security: Public

CONCLUSIONS AND OUTLOOK In this section we have described a solution for TVWS communication based on OpenAir-Interface. At the time of writing, this equipment is being used as part of the OFCOM TVWS trials Errore. L'origine riferimento non è stata trovata.. The trials will allow us to do re-search and experimentation on several aspects of TVWS communications. Firstly, we will study the aggregation of resources/links, e.g., TVWS with licensed and other unlicensed such as ISM, and within TVWS. This is also the subject of the ongoing SOLDER project Er-rore. L'origine riferimento non è stata trovata. (see also Chapter 9). Another interesting research avenue is the study of secondary coexistence, e.g., LTE and 802.11af in TV white space. Last but not least we would like to study and survey the performance that can be achieved, e.g., in terms of interference to primary, secondary user performance through ob-jective user opinion polling.

4.2 Achievement 2: Design of HW decoders for Polar Codes. In May 2014, a PhD student from CNIT/Polito (Dr. Andrea Biroli) started a 6 month period of collaborative work at Bilkent University. During the whole activity, periodic skype meetings are arranged among Prof. Guido Masera, Prod. Erdal Arikan and Dr. Biroli, in order to dis-cuss intermediate results and to plan following steps in the research activity. The final objec-tive of the joint effort between the two partners is to design an innovative hardware decoder for Polar Codes. Initially, the key elements to be pursued in the development of advanced channel coding decoder architectures have been deeply investigated. Starting from known results in the sci-entific literature and from current market requests in terms of throughput and energy dissipa-tion, feasible solutions have been pointed out. It was agreed that an investigation on belief propagation (BP) polar code decoders will be the research activity. The main motivations behind this choice is the capability of the BP algo-rithm to output a complete codeword after a small number of iterations and its high degree of potential parallelism: because of these known characteristics, the BP approach seems able to overcome throughput limits of successive cancellation (SC) decoders and to reduce de-coding latency. The most important target performance for the planned decoder was identi-fied as a throughput as high as 1Gb/s, which is a relevant improvement with respect existing architectures, able to reach 100Mb/s. As a second step, the polar code BP algorithm has been deeply studied. Starting from previ-ous software models available at Bilkent University and Polito, an improved C model has been developed for a system including BP polar encoder, transmission channel and decoder. Simulations allowed to validate the model and to generate test vectors to be used for the assessment of the hardware model. As soon as the functional C model was ready, tested and debugged, a refined, hardware oriented C model has been derived, which closely follows the planned decoder architecture. The new model includes quantization, storage and arithmetic details that allow for direct comparison of software and hardware simulation results. In parallel, a VHDL model has been produced describing a fully parallel decoder architecture, targeting ASIC implementation; the model incorporates several parameters, which enable the hardware tuning according to the results of simulations from C code. In this phase of the work, the possible use of automatic tools for the generation of VHDL models from high level C language descriptions has been considered and evaluated; how-ever, this approach has been abandoned because of the strict performance constraints. Testbenches were written in order to prove the correctness of the VHDL code in compliance with expected results from the C simulations. The synthesis of the fully parallel architecture has been performed by means of Synopsys Design Compiler, and the obtained circuit has been characterized in terms of throughput, occupied Silicon area, working frequency and power consumption. The results achieved are


Security: Public

promising (overcoming prefixed target) so this architecture have been chosen as basic struc-ture for the following developments. Subsequently, the work followed two parallel branches. The first one aims at mapping a properly refined and optimized version of the obtained VHDL design onto an FPGA board. The second branch has to do with the design of a reduced complexity architecture, with some further architectural refinements and able to save implementation complexity with lim-ited effects on communication performance. To fulfill this task all VHDL writing and testing has been done and also new simulative C code for the novel architecture has been produced. The new model will be synthesized and validated using the same approach and tools de-scribed for the initial implementation. The final achieved results will be compared against published implementations available in the literature. The designed decoder and the collected comparisons will form the core of a new joint paper that will be submitted to an international journal for review. 4.3 Achievement 3: 4G/5G coexistence studies Figure 5 depicts the joint TUD-CNRS/Eurecom experimental setup to test the coexistence of 4G and 5G systems. The primary system is based on CNRS/Eurecom’s OpenAirInterface LTE system representing 4G. The secondary system uses TUD’s testbed based on National Instrument’s PXI platform, RF adapter module and TUD’s Labview implementation of a 5G waveform (GFDM). A spectrum analyzer is employed to measure the power levels. The pri-mary system has been configured in FDD 5 MHz and the scheduler in the eNB makes sure it leaves a hole in the UL resources for the secondary system. The frequency spectrum is illus-trated in Figure 6, where the 5G secondary system uses the GFDM waveform to transmit data in the UL of the primary 4G LTE system. Table 4 gives the relevant list of parameters for the primary system and the three optional waveforms for the secondary system, i.e. GFDM in two different configuration and OFDM as the reference waveform. The objective of the experiments is to measure performance degradation in the uplink of the primary system when a secondary system transmits in the holes of the UL spectrum of the primary system. From theoretical results it is expected that the performance is less degraded than with OFDM as a secondary system as GFDM is designed to have a lower out-of-band leakage (see also Section 2.3).


Security: Public

Figure 5: 5G spectrum overlay

Figure 6: Secondary GFDM system (red) transmitting in the UL of the primary LTE sys-tem. Table 4: Paramters of primary and secondary system

Primary system LTE configuration FDD 25PRB (5MHz) DL frequency 2.68 GHz UL frequency 2.56 GHz Scheduler Not scheduling PRBs 21-23 DL link adaptation Based on CQI feedback


Security: Public

UL link adaptation Fixed, mcs 8 UL power control Closed loop eNB TX power -24dBm/PRB UE TX power Check with measurements Secondary system Frequency 2.56 Ghz Maxmium TX power: 7.5 dBm Bandwidth 7.68 MHz Secondary waveform GFDM1 K (FFT Size) 512 Allocated subcarrier 36 Center subcarrier 121 First subcarrier 102 Blocks 10 Filter Raised-cosine GFDM2 Blocks 20 OFDM Blocks 6 Filter Rectangular The measured key performance indicators at primary system are:

• UL RX power • UL Interference power • UL BLER • UL throughput (measured with iperf application)

All KPIs were measured as a function of the transmit power of the secondary system, which was controlled using the variable attenuator. As an example we show some throughput re-sults in Figure 7.

Figure 7: Throughout results

0

100

200

300

400

500

600

700

800

900

20 15 13 11 10 9 8 7 6 5 4 3

Throughput [kbps]

Attenuation

GFDM1

GFDM2

OFDM


Security: Public

4.4 Appendix: Massive MIMO Experimental Testbed CNRS/Eurecom is currently developing a testbed for massive MIMO base on its Express-MIMO2 software radio board. The testbed will support up to 64 active antenna elements and allow real-time two-way experimentation. OpenAirInterface OpenAir-Interface.org is an open-source platform (GNU GPL v3) for experimentation in wire-less systems primarily targeting cellular technologies (LTE/LTE-Advanced and beyond) and rapidly-deployable mesh/ad-hoc networks. It has been funded predominantly by collaborative projects under the 6th and 7th EU framework programmes and to a lesser extent by the French ANR programmes. The platform comprises both hardware and software components and can be used for simulation/emulation as well as real-time experimentation on both cus-tom RF equipment and standard equipment from National Instruments/Ettus. It comprises the entire protocol stack from the physical to the networking layer starting from Rel-8 LTE. The current software can interoperate with commercial LTE terminals and can be interconnected with closed-source EPC (Enhanced Packet Core) solutions from third-parties. The objective of this platform is to provide methods for protocol validation, performance evaluation and pre-deployment system test. OAI is currently hosted and maintained by CNRS/Eurecom. It will become a foundation (French fonds de dotation) in the fall of 2014 and will seek funding from members, in the spir-it of foundations offering support for open-source software such as OpenStack, APACHE and ANDROID. The current major industrial users of OAI for collaborative projects are Agilent, China Mobile, IBM, Alcatel-Lucent, Thales, and Orange. One of the main objectives is to pro-vide an open-source reference implementation which follows the 3GPP standardization pro-cess starting from Rel-12 and the evolutionary path towards 5G.


Security: Public

Massive MIMO Testbed Architecture

Figure 8: MASSIF Prototype Architecture (BS) The architecture of the OpenAirInterface Massice MIMO testbed is shown in Figure 8. Exist-ing components are shown with a red outline. Each component is described in more detail in the following.

• One ExpressMIMO2 board for every four active antenna elements. ExpressMIMO2 can drive and process up to 4 parallel RF chains with up to 20 MHz bandwidth (4x5 MHz, 2x10 MHz, 1x20 MHz) per chain in the range of 350-3800 MHz. It intercon-nects with a baseband computing engine using Gen1 1-way PCIe (2.5 Gbit/s peak full-duplex bi-directional throughput). The latest version of the ExpressMIMO2 board (revision 2) also has built-in amplifiers, LNAs, and switches for TDD operation.

• One PCIe backplane for multiplexing 16 Gen1 1-way PCIe (40 GBit/s peak) into a single 16-way Gen2 PCIe link (80 Gbit/s). Note that the bandwidth requirements of the above architecture are significantly less (50 %) than the sustainable throughput of a PCIe backplane. For LTE sampling rates, the required throughput of the overall sys-tem is less than 16 Gbit/s (assuming 32-bit complex IQ data samples and either 64x5 MHz baseband channels, 32x10 MHz baseband channels, or 16x20 MHz baseband channels)

• One high-end 20-core Xeon server (10-core dual-processor 3 GHz) with AVX2 in-structions running a real-time Linux variant (PREEMT RT or RTAI) and OpenAir4G LTE baseband processing. OpenAir4G currently requires 2 cores for TDD 2-Antenna

16-‐way

PCIe

Backplane

ExpressMIMO2

16x 2.5Gbit/s =40 Gbit/s peak

16-‐way PCIe Gen 2

(80 Gbit/s peak)

High -‐End Intel Xeon

Computing Engine

3GHz Dual -‐Proc AVX2

20 Parallel Cores

64 element antenna array

4

4

4

4

4

4

4

4

4

4

4

4

OpenAir4G

RT -‐Linux

MODEMs

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

ExpressMIMO2

4

4

4

4

PCI Express


Security: Public

20MHz BS transceivers, which is used mainly for BS receiver processing (and we note that 80% of this is for turbo-decoding at full uplink throughput). The latter makes use only of SSE4 instructions and will directly benefit from a factor 2 improvement with AVX2 instructions for DSP-intensive algorithms. Therefore, we envisage 1 Xeon core per pair of antennas for 20 MHz channels. 20-core machines should be largely sufficient for 16 baseband streams of 20 MHz bandwidth, 32 baseband streams of 10 MHz bandwidth and 64 baseband streams of 5 MHz bandwidth.

• Massive antenna array technology. This part still needs to be built and several op-tions are available. The first option is to use several off-the-shelf antennas and mounting them on a single board. The second option is to subcontract the design of a massive MIMO array to one of CNRS/Eurecom’s partners IABG. Their solution could provide a 64-element radiating square-grid antenna array etching directly on the PCB. The RF configuration of the IABG array will be fabricated with 100mW output power per array element in order to achieve radiated powers of 6W. This is comparable to current small-cell BSs used to cover hot-spot zones. The target frequency band will be in the 3.5GHz unpaired spectrum (TDD) which is envisaged for future high-data rate services.

• The UEs will use a single ExpressMIMO2 board coupled with existing RF front-ends (up to 4 antenna elements) already in use at CNRS/Erecom with off-the-shelf anten-nas. The terminals can be mounted in specially-conceived vehicles owned by CNRS/Eurecom in order to perform outdoor experimentation. EURE will request ac-cess to 3.5 GHz spectrum in order to deploy a test-cell around its premises.


Security: Public

Comments and suggestions for the improvement of this document are most welcome and should be sent to:

[email protected]

________________________________________________________________________

http://www.newcom-project.eu

newcom# - cordis · 2017. 4. 22. · the main asset of the euwin@cnrs/eurecom lab is the...

Documents