recent$results$on$op?cal$access$networks$ · pdf file1. unbundlingopons ... • upstream:...

Rete O&ca di Accesso a Divisione di frequenza e/o di lunghezza d’onda per soluzioni Next Genera?on Network

Recent Results on Op?cal Access Networks from the PRIN Project ROAD-‐NGN

Presenter: Roberto Gaudino Politecnico di Torino, Op0cal Communica0on group

(OPTCOM www.optcom.polito.it) Corso Duca degli Abruzzi 24, 10129 Torino, Italy, E-‐mail: [email protected]

Riunione Gruppo Re?, Cavalese, 2015 January 14th

2 Riunione Gruppo Re?, Cavalese, 14 gennaio 2015

§  Project dura0on: February 2013-‐January 2016

ROAD-‐NGN

Gabriella Cinco&, Coordinator Julian Hoxha Luca Caldaroni

Pierpaolo Boffi Mario Mar?nelli, Achille PaWavina Guido Maier Lucia Marazzi, Paola Parolari

Marco Santagius?na

Antonio Mecozzi Cris?an Antonelli Francesco Matera (FUB)

Francesco Vatalaro Marco Petracca Romeo Giuliano Paolo Mancuso

Ernesto Ciaramella Fabio BoWoni Marco Presi Luca Valcarenghi Piero Castoldi

Roberto Gaudino Valter Ferrero Roberto Cigliu&

Francesco Matera


§  Scenario: Passive Op0cal Networks (PON) for FTTH solu0ons

§  Project focus: improvements compared to current PON standards in terms of:

1.  unbundling op0ons •  See my presenta0on last year for the 2014

Cor0na Mee0ng 2.  Increase in bit rate per wavelength 3.  Cost effec0ve solu0ons to handle mul0ple

wavelengths

ROAD-‐NGN project in a glance


§  Introducing the scenario: Passive Op0cal Network (PON) access architectures • The most recent ITU-‐T standard for PON: NG-‐PON2

§  The ROAD-‐NGN research goals: beyond NG-‐PON2

§  Proposed architectures for upstream and downstream transmission • Experimental results

Outline of the presenta?on


Today mostly deployed standard: GPON (ITU-‐T G.984): §  Number of user per PON tree: up to 64 §  Mul0plexing technique:

•  TDM in downstream at 2.5 Gbps, TDMA in upstream at 1.25 Gbps

Passive Op?cal Network” (PON) architecture

Central Office

Passive Optical Splitters O/E

O/E

O/E

OLT (Optical Line Terminal)

ONU (Optical Network Unit)

ODN (Optical Distribution Network)

7

ITU-T most recent standard: NG-PON2 (TWDM-PON)

•  Defined by FSAN and ITU-T in the Recommendation G.989.1 “40-Gigabit-capable passive optical networks (NG-PON2)”

•  TWDM-PON: time and wavelength division multiplexed PON

Backward compa,ble with previous PON standards (GPON, XGPON and RF-‐Video)

h"p://www.itu.int/rec/T-‐REC-‐G.989.1-‐201303-‐I/en

NG-‐PON2: 4 new wavelengths (per direc,on)

8

NG-PON2 features •  4 wavelengths per direction, 100 GHz spacing

•  Upgradeable to 8 wavelengths (50 GHz)

•  TDMA on each of the 4 wavelengths •  Each wavelength is treated as an independent XG-PON

•  Downstream: 10 Gbps •  Upstream: 2.5 Gbps

•  Traditional Splitter-based PON •  Backward compatibility with ODN loss classes

Nominal 1 (N1 class)

Nominal 2 (N2 class)

Extended 1 (E1 class)

Extended 2 (E2 class)

Minimum loss 14 dB 16 dB 18 dB 20 dB

Maximum loss 29 dB 31 dB 33 dB 35 dB


§  … besides physical layer unbundling presented last year in Cor0na

§  Target #1: Increase the bit rate per wavelength in both direc0ons • Toward 30-‐40 Gbps in both direc0ons (as symmetrically as possible)

§  Target #2: Simplify wavelength handling for the upstream, trying to solve one of the most cri0cal issues to be solved in TWDM-‐PON

ROAD-‐NGN targets


§  Target #1: higher bit rate per wavelength §  Introduce more sophis0cated modula0on formats and mul0plexing techniques •  Frequency division mul0plexing (FDMA)

→  In both direc0ons (UA and DS) →  Implemented at the electrical level on top of each wavelength

• M-‐QAM on each electrical subcarrier in the FDMA comb

§  As a result: digital signal processing (DSP) required at both the ONU and OLT •  Constraint: low DSP rate at the ONU to keep cost and power consump0on at reasonable levels

The recipes to achieve these targets


Downstream FDM signal spectrum at ONU receiver

§  We propose to use 32 electrical subcarrier, each carrying 16-‐QAM a 1 Gbps net data rate •  These FDM spectrum totally fill the available downstream electrical bandwidth over a typical direct-‐detec0on op0cal link → (7-‐8 GHz 3dB bandwidth)

FDM in downstream

Each spectral slice is dedicate to a given ONU (i.e. FTTH user)


Experimental setup

Passive fiber plant ODN emulated on a real metropolitan fiber testbed Realis,c ODN losses as requested by ITU-‐T

OLT transmiWer DSP emula,on through a Tektronix 20 Gsample/s arbitrary waveform generator

ONU receiver Complete optoelectronic receiver Offline DSP emula,on in Matlab aVer ADC using a 50 Gsample/s real ,me oscilloscope


§  Aher a careful op0miza0on of many system parameters we obtained 32 Gbps downstream at FEC threshold for the required ODN losses

Experimental results Laun

ched op,

cal

power at O

LT BER contour

plot

FEC threshold BER=2·10-3 (as required by G.795.1–I.4).


DSP Complexity at OLT

§  The DSP required at the central office is for sure significantly more complex than the current NG-‐PON2 (TWDM-‐PON) • Mostly because it must run at about 20 Gsample/s (and thus also requires extremely fast DAC)

§  But the achieved capacity per wavelength is 3 0mes bigger than NG-‐PON2

§ Moreover, in most op0cal transmission sectors it is today widely recognized that DSP is required to beat the “10 Gbps per wavelength” barrier

FP7-ICT-2011-8 Challenge 3.5 – STREP project n. 318704 – FABULOUS

FDMA Access By Using Low-cost Optical Network Units in Silicon photonics

n  At the ONU RX and after photo detection, electrical RF down-conversion is applied so that DSP can be at baseband and only on the spectral slice dedicated to each specific ONU

n  The required baseband processing for our proposal can be done using DAC and ADC working in the 500 Msample/s range

n  Low-cost chipsets are already available today to implement this electronic architecture l  UWB chipsets (for instance from Alereon)

ONU complexity (from EU STREP “FABULOUS”)

17

RFjf

Baseband digital signal processing

on I/Q components DS data out


§  Target #2: simplify wavelength handling for upstream

§  In the ITU-‐T NG-‐PON2 standards each ONU should generate its upstream wavelength with very high accuracy (100 GHz grid) by a tunable laser •  This is the key technological obstacle to be solved today •  No tunable lasers exist today with a price compa0ble with ONU

§ We propose a reflec0ve approach for the upstream that completely solve this issue

Target #2: wavelength handling


§  Generate also the upstream wavelength at the central office by a comb of CW lasers

§  Modulate them back in reflec0on at the ONU

Proposed reflec?ve solu?on for US

20

•  Key idea: upstream wavelengths are generated at the OLT, and modulated in reflection

Reflective PON with OLT centralized wavelength generation

AWG

…

OLT

RX TOF

Upstream signal

Reflec0ve TX

AWG

RX array

ONUi

TX DS array

…

…

TX US array

•  We studied in the ROAD-NGN project several variants of this architeture

•  Key point: the ONU does not need tunable lasers

ODN

Bank of CW laser, Not modulated

TOF

TOF: Tunable Op,cal Filter

21 Riunione Gruppo Re?, Cavalese, 14 gennaio 2015 21

Reflective Semiconductor Optical Amplifiers

•  The Politecnico di Milano group inside ROAD-NGN experimentally implemented this goal using reflective semiconductor optical amplifiers

•  R-SOA: Combination of a SOA and an optical mirror

•  SOA are semiconductor optical amplifier with the following characteristics •  Low cost (at least potentially, if they have mass production) •  Medium gain (10-20 dB peak gain) •  Medium-high noise figure (7-8 dB) •  Some residual polarization dependence •  The gain can be modulated by changing the injected optical current

•  Modulation speed is 1-2 GHz maximum

SOA Input fiber Op0cal Mirror

I(t) SOA gain changes with driving current


§ POLIMI, using high-‐performance RSOA developed in the EU-‐Project • FDMA in the upstream • Top performance: Up to 16 users per wavelength, each at 1 Gbps for a total upstream capacity of 16 Gbps per wavelength

• Flexible solu0on to achieve higher number of users per wavelength

POLIMI Experimental results

http://ermes-project.eu/


§ POLITO, using Silicon-‐Photonics high speed R-‐MZM developed in the EU-‐Project FABULOUS • FDMA in the upstream • Top performance: Up to 32 users per wavelength, each at 1 Gbps for a total upstream capacity of 32 Gbps per wavelength

POLITO Experimental results

http://www.fabulous-project.eu/


§  If you work in op0cal networks, please consider the FOTONICA 2015 Italian Na0onal Conference

Just a liWle of self-‐adver?sement…

§  SAVE THE DATES:


§  In ROAD-‐NGN we have demonstrated solu0ons that can be of interest for the next genera0on of ITU-‐T PON standards • NG-‐PON3 ??

§ Will all this capacity be required in fixed access? • A candidate applica0on: using NG-‐PON3 to support front-‐hauling in Cloud Radio Access Networks (C-‐RAN) for 5G mobile networks → CPRI higher data rate per remote radio unit: 10 Gbps

Conclusions


Recent Results on Op?cal Access Networks

from the PRIN Project ROAD-‐NGN



Thank you for your aWen?on!


Recent Results on Op?cal Access Networks

from the PRIN Project ROAD-‐NGN



BACK-‐UP slides

Front-haul and CRAN

explained

(one of today) mobile backhaul architecture

29

Antenna

RF TX

RF RX

Physical Layer Digital Signal Processing (LTE, LTE-Advanced, 5G)

Network layer interface

Gigabit Ethernet optical cards

Central office Optical fiber link

Antenna site hardware (base station)

Zooming on the physical layer

30

Antenna

I/Q

modulator

Physical Layer DSP

I Q

DAC DAC

I/Q

demod

I Q

ADC ADC

Very high speed rate at this interface


This layer is very computationally intensive in recent mobile standards, and it will become more and more sophisticated in the near future

RF local

oscillators

 Example  Wireless RF bandwidth: 20 MHz  Sampling frequency: >20 Msamples/s  Bit per sample: 16 bit/sample  Resulting rate for one component: 320 Mbit/s  x2 I/Q streams: 640 Mbit/s  x3 antennas for each radio site  xN due to MIMO  xM due to multi-carrier

 Future trends:

DAC and ADC required bit rate

31

The new vision: front-hauling approach

32

Antenna

I/Q

modulator

Very thin framing protocol (such as CPRI)

I Q

DAC DAC

I/Q

demod

I Q

ADC ADC

Central office

Optical fiber link (P2P or PON)

CPRI transceiver

Physical Layer DSP


Bit rates in the Gbps range at this interface

RF local

oscillators

 CPRI current data rates

Common Public Radio Interface (CPRI)

33

 CRAN Cloud Radio Access Networks Advantages: •  All functions are centralized and virtualized •  Much easier “software” reconfiguration •  Physical layer coordination of many antenna sites

Coordinated Multipoint (CoMP):  Joint processing possible if all

antennas are centrally controlled  It allows:

 coordinated scheduling  Beamforming  Crosstalk cancellation

 Distributed MIMO among antennas

Concept of CoMP in LTE Advanced

34

Central office Coordinated transmission

control

recent$results$on$op?cal$access$networks$ · pdf file1. unbundlingopons ... • upstream:...

Documents