OCARI : Optimization of Ad-Hoc Communications for Industrial

OCARI : O ptimization of Ad-hoc Communications for Industrial Networks

http://ocari.org

ETSI M2M Workshop 19-20 October 2010

Projet No. ANR-06-TCOM-025 © OCARI Consortium, 2007-2010 Consortium, OCARI © Outline 1. Context and objectives Industrial requirements and challenges Technologies and market Objectives OCARI partners 2. Description of the OCARI stack OCARI network Estimation of node residual energy MACARI: medium access EOLSR: energy-efficient routing SERENA: node activity scheduling 3. Demonstrators MACARI+SERENA+EOLSR GMOCARI: global middleware 4. Conclusion & perspectives © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 2 1. Context & objectives :

industrial needs,

challenges © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 3 Segmentation of industrial requirements

Sending frequency

Continuous monitoring Quasi Measures in hostile continous environment

Instrum. mobile check 1 s Teledosimetry Sensors « pilotage » Challenge of low power Sur chantier Gamma ray monitoring

Déplacement 10 s Inter -chantier RP Beacons « Performances » Domain "mobility": • High frequency sensors 1 min • Mobility in the plant OCARI • Fast configuration Building monitoring Valves

1 h Cadenas Transmetteurs fixes électronique sur fins de course Electrical alimentation Domain « static sensors" Mobility Fast configuration Targeted autonomy

1 week 80 days 18 months N x 18 months © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 4 Example 1: supervision of radioprotection

Teledosimeter Radiameter Phone DECT personal & mobile

Controled area

Coordinator of Coordinator of an OCARI star an OCARI star Coordinator of (battery) Electromagnetic (battery) an OCARI star Coordinator of (battery) an OCARI star (battery) Coordinateur of an OCARI Coordinateur network of an OCARI • ~ 50 sensors ( mobile & network fixed ) distributed in an area of 40m diameter • 1 sample./5s, few bytes /sensor • Time constrained and Supervision room for radioprotection delivery guarantee © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 5 Example 2: predictive maintenance of a ship

Up to 400 parameters per compartment & 4 measure points per square meter Vibration analysis Measure of pression, temperature & throughput Analysis of oil…

Integration in a constrained environnement: Temperature : Turbine gaz – CEI 60068 Electromagnetic compatibility : Radar, warfare devices– STANAG 4436, 4435, 4437 Electromagnetic discretion: TEMPEST rules Metalic channels with fluids © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 6 Challenges & choice of a mesh topology

• Efficiency in time and space

• Scalability and mobility

• Delay tolerant & asynchronous delivery for a destination temporarily out of coverage area

• Time contrainted vs « duty cycle » & energy consumption

• Network lifetime maximization with battery - operated routers Choice of a mesh • Dimensioning and deployment tools topology because: Time & space efficiency of the frequency spectrum Transport capacity (bits/m/s) proportional to N © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 7 1. Context & objectives :

technology & market

Slide 8 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 No solution on the market for the identified needs

Many proprietary technologies: EnOcean, Z-Wave, SensiNet, HomeRider, SmartDust, MeshScape, Xanadu-Wireless/Green Peak… But no global support Existing standards: of the three needs:

Mobility Determinism Low power

(Source : www.isa.org) © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 9 1. Context & objectives :

objectives © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 10 Objectives :

Technical feasibility of industrial wireless sensor networks providing :

Time-constrained medium access (soft real-time), Micro mobility of some nodes, Scalability & self-healing Energy and spectrum efficiency

To contribute to the emergence of an open standard designed for industrial environments

with higher performances than the market offer

for wireless sensor networks © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 11 OCARI Partners

Routing Sector: Energy and continuous process

Modeling of energy constraints

Medium access

Sector: navy defense

Provider of RF devices & ZigBee stack developer

Medium access © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 12 2. Description of the

OCARI stack

29 juin 2010 Optimisation des Communications Ad-hoc pour les Réseaux Industriels Slide 13 2007-2010 Consortium, OCARI © OCARI network © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 14 OCARI Stack

Application Layer Projet OCARI ZigBee API for network Application Framework “application Management management (LQI, 4 R A object “ - S I N Device RSSI, residual energy 20 2 / A Object m 3 O L level…) O Energy + MDO A 2 Public Interfaces G

APSDE - APSDE - APSDE - APSDE - APSDE - SAP SAP SAP SAP SAP

Application Support (APS) Layer SAP SAP APSME- NDE-SAP Energy efficient NwCARI

routing SAP SAP

Unconstrained User traffic Control Traffic ESPN- Energy supports node Unicast routing Broadcast routing Service according to according to MPRs SERENA EOLSR Provider mobility & avoids SAP SAP

EOLSR table NME- interferences

MDE-SAP MME-SAP SAP SAP SAP SAP MaCARIME- ESPM- MaCARI Constrained User Traffic Network Creation Estimator of Association Address Tree Relaying Control Allocation node residual energy PDE-SAP PME-SAP IEEE 802.15.4 Physical (PHY) Layer 2.4 GHz Radio Radio Controls the medium access High Layers High Layers Interface OCARI Layers OCARI interface supporting time-constrained Application Profiles traffic & initializes the network © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 15 Estimation of node residual energy

Application Framework MDO + Management Energy 4 R A Public N Interfaces - S I Device A 20 2 / Object m 3 O L A 2 O G

APSDE-SAP APSDE-SAP APSDE-SAP APSDE-SAP APSDE-SAP

APSME-SAP Application Support (APS) Layer

NDE-SAP NwCARI ESPN- SAP Unconstrained User traffic Control Traffic Energy Unicast routing according to Broadcast routing according to MPRs Service EOLSR table SERENA EOLSREOLSR NME-SAP Provider

MaCARIME- SAP

MDE-SAP MME-SAP ESPM- SAP

MaCARI Constrained User Traffic Network Creation Association Control Address Allocation Tree Relaying

PDE-SAP PME-SAP

Physical (PHY) Layer

2.4 GHz Radio

High Layers High Layers Interface OCARI Layers OCARI interface Application Profiles

Slide 16 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Goals of the energy estimator

Modeling of : Energy consumed by a transmission (sending, receiving and overhearing)

Battery of a sensor node to estimate its residual energy Estimation of residual energy by the battery voltage

inadequate for alkaline batteries

Battery modeling [Rakhmatov 03, Rong 03] too complex +∞ t t 2 2 σ ( t ) = i (τ )d τ + 2 i (τ )e − β m ( t − τ ) d τ for implementation ∫0 ∑ ∫0 m = 1 in a sensor node © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 17 How to adapt the model to sensors

Reduction of resources needed by the model

Computation is simplified: • limited development Reduction of memory size: = ⋅∆ • time intervals Ln n Recursive model n  n −1  σ = δ + λ σ − δ + + δ (L n ) ∑ I k k  (Ln −1 ) ∑ I k k  2 I n A(Ln , Ln −1 n , Ln −1 ) k =1  k =1 

Takes into account the temperature factor © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 18 Accuracy of the battery model Simulation Comparison with the DUALFOIL simulator Current, mA DUALFOIL Linear model E % Recursive model E % 20 18156 20140 10,92 19116 5,28 40 9249 10291 11,26 9751 5,42 60 6203 6911 11,41 6537 5,38 80 4664 5203 11,55 4912 5,31 100 3737 4171 11,62 3932 5,21 Real discharges Comparison with a prototype Current x Cycle Measured Linear model E % Recursive model E % lifetime 100 mA x 50 % 3255 3600 10,59 3541,2 8,7 100 mA x 10 % 10332 18000 74 9751 14 © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 19 MaCARI: medium access control in OCARI

Application Framework MDO + Management Energy 4 R A Public N Interfaces - S I Device A 20 2 / Object m 3 O L A 2 O G

APSDE-SAP APSDE-SAP APSDE-SAP APSDE-SAP APSDE-SAP

APSME- Application Support (APS) Layer SAP

NDE-SAP NwCARI ESPN- SAP Unconstrained User traffic Control Traffic Energy Service Provider Unicast routing according to Broadcast routing according to MPRs EOLSR table SERENA EOLSR NME-SAP

MaCARIME- SAP

MDE-SAP MME-SAP ESPM- SAP

MaCARI Constrained User Traffic Network Creation Association Control Address Allocation Tree Relaying

PDE-SAP PME-SAP

Physical (PHY) Layer

2.4 GHz Radio

High Layers High Layers Interface OCARI Layers OCARI interface Application Profiles © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 20 Objectives of MaCARI

Provides a MAC layer

Supporting two types of traffic Time-constrained Time-unconstrained

Ensuring a deterministic medium access

Saving energy of all entities (sensors as well as coordinators)

Supporting hundred nodes per island

Reusing the PHY layer at 2.4GHz of IEEE 802.15.4 © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 21 Principles of MaCARI

Tree topology with a CPAN : A

B C

synchro scheduled activities unscheduled activities inactivity synchro T0 T1 T2 T3 T0 A synchro B Cascad of sequential beacons C

Sequential activation of A scheduled activities B stars : data gathering & C parent-child relaying Simultaneous unscheduled activities A, B, C activation of coordinators © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 22 MaCARI : Sequential activition of stars in [T1, T2]

Stars are activated sequentially At the end of the activation period, data are forwarded to the parent in the tree Intra-star traffic : Two QoS are offered: 1. Unconstrained traffic (best effort mode ) : slotted CSMA/CA

– Traffic between coordinators in [T 2-T3] (SERENA + EOLSR) 2. Time-constrained traffic (deterministic access ) : GTS

– Traffic between coordinators in [T 1-T2] Exchanges between a parent and its children GTS – Several options of GTS slotted CSMA/CA QoS #1 QoS #2 • Definition of a reservation level adapted Exchanges to the appli needs : intra-star traffic GTSn + PDS Parent-Child

T0 T1 T2 T3 © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 23 MaCARI: Results

Simulations Optimization of synchronization Dimensionning Simulations show that MaCARI outperforms slotted CSMA/CA

Prototyping General functioning: synchronization Implementation of GTS, GTSn, PDS Allows us to select representative simulation parameters

Deployment in a mine © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 24 Energy efficient routing & node activity scheduling

Application Framework MDO + Management Energy 4 R A Public N Interfaces - S I Device A 20 2 / Object m 3 O L A 2 O G

APSDE-SAP APSDE-SAP APSDE-SAP APSDE-SAP APSDE-SAP

APSME- Application Support (APS) Layer SAP

NDE-SAP NwCARI ESPN- SAP Unconstrained User traffic Control Traffic Energy Service Provider Unicast routing according to Broadcast routing according to MPRs NME-SAP EOLSR table SERENA EOLSREOLSR

MaCARIME- SAP

MDE-SAP MME-SAP ESPM- SAP

MaCARI Constrained User Traffic Network Creation Association Control Address Allocation Tree Relaying

PDE-SAP PME-SAP

Physical (PHY) Layer

2.4 GHz Radio

High Layers High Layers Interface OCARI Layers OCARI interface Application Profiles

OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 25 2007-2010 Consortium, OCARI © Objectives of the network layer

Network layer Provides a multi-hop communication while maximizing network lifetime Adaptive and energy efficient routing : EOLSR Node activity scheduling: SERENA Supports two types of traffic time-contrained traffic: relaying at level 2 uncontrained traffic: routing at level 3 Supports the mobility of some nodes

State Power value (W) 802.11 802.15.4 Transmit 1.3 0.1404 Receive 0.9 0.1404 Idle 0.74 0.0018 Sleep 0.047 0.000018 © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 26 Energy efficiency

Energy efficient routing Node activity scheduling EOLSR SERENA

Energy efficiency

Transfer optimization Topology control EOLSR+SERENA

OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 27 2007-2010 Consortium, OCARI © EOLSR : an energy-efficient routing

EOLSR : the energy efficient extension of the OLSR routing protocol

Neighborhood discovery Topology dissemination

Goals:

Minimize the energy consumed by the end -to -end transmission of a packet Balance the residual energy of nodes Avoid nodes with a low residual energy Reduce the routing overhead © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 28 EOLSR: an energy-efficient routing

EOLSR consists of 4 modules 1. Energy cost of a transmission

Cost (transmission by i) = E trans + n * E rcv

2. Selection of EMPRs : Energy efficient MultiPointRelays 2 3 1 4 Intermediate nodes in the computed routes

3. Routing algorithm Selection of routes according to the energy consumed by an end-to-end transmission

4. Optimized broadcasts Network broadcasts consume many resources the number of retransmissions must be minimized © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 29 EOLSR: an energy efficient routing

How to reduce the overhead of EOLSR Cross-layering with the application Use the information coming from the application and the lower layers to optimize the resource utilization and protocols performances Definition of application profiles general mode

strategic mode for data gathering appli. – maintain only useful routes (routes toward the strategic nodes) © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 30 SERENA: node activity scheduling

SERENA based on node coloring Two nodes with the same color transmit simultaneously without interfering

Assignation of time slot according to the color of the node

Cross-layering with the application general mode data gathering mode collecte transmission types © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 31 SERENA +EOLSR

Optimize the utilization of network resources Bandwidth: spatial reuse

Gain Time – Reduction of the activity period Increase of the supported traffic Activité Inactivité

– Minimization of data gathering delay – Better time consistency of collected data

Energy – Reduction of consumed energy – Increase of network lifetime © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 32 3. OCARI demonstrators

29 juin 2010 Optimisation des Communications Ad-hoc pour les Réseaux Industriels Slide 33 2007-2010 Consortium, OCARI © Demonstrator of EOLSR + SERENA +MaCARI

Application Framework MDO + Management Energy 4 R A Public N Interfaces - S I Device A 20 2 / Object m 3 O L A 2 O G

APSDE-SAP APSDE-SAP APSDE-SAP APSDE-SAP APSDE-SAP

APSME- Application Support (APS) Layer SAP

NDE-SAP NwCARI ESPN- SAP Unconstrained User traffic Control Traffic Energy Service Provider Unicast routing according to Broadcast routing according to MPRs NME-SAP EOLSR table SERENASERENA EOLSREOLSR

MaCARIME- SAP

MDE-SAP MME-SAP ESPM- SAP

MaCARI Constrained User Traffic Network Creation Association Control Address Allocation Tree Relaying

PDE-SAP PME-SAP

Physical (PHY) Layer

2.4 GHz Radio

High Layers High Layers Interface OCARI Layers OCARI interface Application Profiles

TELIT © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 34 Demonstrator of SERENA + MaCARI

Goals of the demonstrator Check that two nodes with the same color can transmit simultaneously without interfering

Configuration 6 nodes using 5 colors. Color 4 is used by nodes 2 and 18 Slot of color 4 appears 830ms after the last beacon

Interpretation Nodes 2 and 18 transmit in the same slot. © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 35 Simulation of EOLSR + SERENA + MaCARI

Objectives

Check that in OCARI, two nodes with the same color can transmit simultaneously without interfering Evaluate performance Study the impact of cross-layering (data gathering profile) Study the impact of unidirectional links

Color conflict Two nodes of the same color prevent the destination from receiving correctly a message destinated to it because of a collision

If only links meeting the following inequation are used received power > reception threshold + capture threshold, no color conflict is detected in all tested configurations © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 36 Simulations of EOLSR + SERENA + MaCARI

Example of simulated configuration for data gathering by node 0

Network - 49 nodes - density 8

Data gathering delays obtained for a network of 49 nodes, density 8, cycle duration = 4s, 1 packet/5s/node

Coloring Coloring general mode data gathering

High improvement brought by cross-layering © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 38 Demonstrator of EOLSR + SERENA + MaCARI

• Implementation integrating MaCARI + SERENA + EOLSR

Liaison RF IEEE-802.15.4

Carte Telit B2400ZB-Tiny Carte Telit B2400ZB-Tiny

MaCARI MaCARI Liaison RS232

Liaison RS232

SERENA+EOLSR+IHM SERENA+EOLSR+IHM

• Implementation of EOLSR on Telit card ZE50-2.4

Objective Check the implementation feasibility of the EOLSR protocol on a limited support (128K Flash, 8K RAM, 16MHz)

Method Implementation of the EOLSR stack on the basis of a ZigBee stack, replacing the AODV routing of ZigBee by the EOLSR routing.

Tests Mobility tests show that packets can be delivered without loss when a mobile node is moving along several router nodes.

Constatations Reliability can be improved by taking into account link quality in the EMPR selection.

Slide 40 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 GMOCARI : global middleware of OCARI

Slide 41 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 GMOCARI : global middleware of OCARI

Open middleware for a standardized integration of wireless sensor networks in SCADA applications

Slide 42 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Test Plateform : loop EVEREST at EDF R&D

• 100 IEEE-802.15.4 nodes • distributed in 3 OCARI networks and 2 buildings © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 43 4. Conclusion &

Perspectives

29 juin 2010 Optimisation des Communications Ad-hoc pour les Réseaux Industriels Slide 44 2007-2010 Consortium, OCARI © Technical realizations & scientific results

Technical realizations : Three PhD theses defended:

A prototype of MaCARI built by LATTIS Joseph RAHMÉ, “ Constraints Modeling and LIMOS on the B2400ZB-Tiny platform and Energy Management in Multi-Hop of Telit RF. Wireless Networks ”, Ph.D Thesis of the UNIVERSITY OF PARIS-SUD XI. An implementation of EOLSR done by Telit RF on its ZE50-2.4 platform. Gérard CHALHOUB, “ MaCARI : Une An implementation of SERENA + EOLSR méthode d’accès déterministe et done by INRIA on PC. It communicates économe en énergie pour les réseaux de with MaCARI via an RS232 link. capteurs sans fil ”, Ph.D Thesis of the lUNIVERSITY BLAISE PASCAL. An implementation of the integration middleware GMOCARI (Global Saoucène MAHFOUDH, “ Energy Middleware for OCARI) done by EDF efficiency in wireless ad hoc and sensor networks: routing, node activity … scheduling and cross-layering ”, Ph.D Thesis of the UNIVERSITYof PARIS 6 PIERRE ET MARIE CURIE. © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 45 Dissemination

Publications Publications multipartners monoparteners Journals 4 2 (IEEE Transaction on IE, Journal of (Journal on mobile Interconnection Networks et WSN information systems, Journal, Future Internet) Future Internet) International Conferences 4 (1 Best Paper Award) 16 (invited paper - IFIP Wireless Days 2008, IEEE International Conference on Industrial Informatics - July 2008, ", 2 x IEEE WCNC 2010 - April 2010) Vulgarization Journal du Club Automation et Wikipedia Conferences ETSI Wireless Factory Workshop 2008 ISA-FRANCE FORUM 2008 : The Actions of promise of wireless diffusion EXERA Wireless Sensors Day 2007 Colloque de Metrologie 2009 , Club Automation : réseau capteurs 2009 Other FP7 Wireless sensor days 2008 Salon Mesure Expo IERE Workshop 2008 2009 © OCARI Consortium, 2007-2010 Consortium, OCARI © OCARI – ETSI M2M Workshop – 19-20 October 2010 Slide 46 Perspectives

Economic positioning

OCARI networks will bring a unique solution for a wide range of needs: mobility, time-constrained and maximized lifetime …

Solutions like ZigBee, WirelessHART, ISA100.11a are not satisfying.

Community of industrial users: EDF, DCNS, AREVA, Hydro-Québec & EXERA (40 large industrial entreprises)

Standardization

Interested in standardization activities: ETSI , IEEE 802.15.4e (LCIM), Wireless group of IAEA (International Atomic Energy Agency).

EDF LIMOS

Tuan Dang, Eric Perrier, Reinald Kutschera, Michel Misson, Alexandre Guitton, Catherine Devic, Antoine Druihle Gérard Chalhoub, Michel Fernandez, Antonio Freitas DCNS LRI Stéphane Brochard, Maurice Sellin, Jean-Baptiste Fievre Khaldoun Al Agha, Joseph Rahmé

INRIA TELIT

Pascale Minet, Saoucène Mahfoudh, Marc-Henri Berthin, Mathieu Pouillot, Ichrak Amdouni, Cédric Adjih Bennani Bensalem LATTIS

Thierry Val, Adrien Van den Bossche, Erwan Livolant, Nicolas Fourty, Réjane Dalcé

Questions ?