Docstoc

Network Drivers Network IP UDP APIs Sockets Resiliency Toolbox

Document Sample
Network Drivers Network IP UDP APIs Sockets Resiliency Toolbox Powered By Docstoc
					Distributed Systems Lecture 4



    Subject of today:
    Example from the real world – the HIDENETS project
Distributed Systems Lecture 4
 About me:
   Erling Vestergaard Matthiesen
   Master graduate June 2005
   Ph.D student since January 1st 2006
   Email: evm@kom.aau.dk
   Office in Aalborg:
      Niels Jernes Vej 12, room A5-207
       9220 Aalborg Øst
    Working on HIDENETS
About HIDENETS
   HIghly DEpendable ip-NETworks and Services
   Funded by the EU
   Focus on dependability within car to car networks
   Web: http://hidenets.aau.dk/
HIDENETS Goals
   Develop and analyze end-to-end resilience solutions
     for scalable distributed applications and mobility aware
       services
     in ubiquitous communication scenarios
           Example use-case: car2car communication with server-based
            infrastructure
       assuming highly dynamic, unreliable communication
        infrastructures
HIDENETS Goals
 Planned results are
   architectural and design solutions, tools for
     development and analysis
   for end-to-end system level resilience and
     dependability
   based on standard off-the-shelf components
   in wireless communication networks and
     infrastructures
 Measures of success
   proof-of-concept prototype, analytic and
     simulation models
   training and background material
   contributions to standardization organizations
 HIDENETS Scenarios
    Challenges/properties of HIDENETS service
                                                                             S1
     provisioning
         Dynamically changing communication
          characteristics in ad-hoc domain and in
          connection to infra-structure services                                  response
                                                             Ad hoc domain                        request
         Off-the-shelf, standard systems and
          components in both domains                                             S2
         Services with high dependability and                                               S3
          scalability requirements
    Selected use-cases of ad-hoc car-to-car
                                                                        request
     communication with connectivity to infra-
     structure services                                                               response


Note: HIDENETS solutions also applicable in other, related
     scenarios, including Personal Area Networks and
     cellular networks with ad-hoc coverage extension.
                                                                Infrastructure
                                                                domain
    Consortium
–    Total nine partners from eight different countries
       Industry:
        Carmeq (GER), FSC (GER),
        WMC (NL), Telenor (NO)
       Academia/research institutions:
         AAU (DK), BME (HU), LAAS (FR),
         Uni-Fi (IT), FCUL (PT)
–    Covering the three areas
        Distributed Systems
        Communications
        Applications and Servers
–    Additional industrial involvement
     via advisory board
    Consortium: Advisory Board
Advisory Council
–     Role:
         Feedback on project results and research directions  quality assurance
         Exploitation and dissemination
         Input/discussions for fine-tuning & adjustment of future project directions
–     Members:
        Mobile Equipment Vendors:
            Francis Tam, Nokia, Finland
            Robert Seidl, Siemens Communications, Germany
        Mobile Operators:
            Chidung Lac, France Telecom, France
        Automotive Industry
            Paolo Coppola, Intecs, Italy
            Gerard Segarra, Renault, France
            Hans-Jörg Vögel, Car-2-car communication consortium & BMW, Germany
                    Structuring of Work
Project Overview and Workpackages
                                 WP1
                      Use cases & Reference model




                                                & middleware
                                                 architecture
                                                  Resilient
                Application




                                                                                                                                   Experimental prototype
                                          WP2




                                                                      Quantitative evaluation


                                                                                                      Design methodologies
                Middleware




                                                                WP4




                                                                                                                             WP6
                                                                                                WP5
         L5: Session Control: SIP, .
         L4: Transport: TCP, UDP,




                                                communicatio
                 RTP/UDP




                                                  Resilient
                                          WP3
           L3: Network Layer: IP




                                                     n
               L2: MAC/LLC

             L1: Physical layer


                                    WP7
                               Dissemination

                                    WP8
                                  Management
Node Architecture
                                         Applications




                                                                            User Space
 Simplified view:           Middleware Solutions
                                      (User space)


                          Resiliency Toolbox
                                  (user space)




                                                                           Operating System HW
                                      APIs       /   Sockets
                          Resiliency                      Middleware
                          OS Support                      OS Support
                           (kernel space)                 (kernel space)

                                             TCP                 UDP
                     Resiliency                      Network / IP
                      kernel
                                                 Network Drivers

                     Resiliency HW
                     (GPS, Ctrl, …)    Network Interface Cards (COTS)
Node architecture
 The full picture:
Software architecture of a node
        Application                      Application                                                            Distributed                                                                                                               Configuration manager
        process 1                        process 2                                                              Black Box

                                                                                                                                                                                                                                                      Service middleware
                SAForum AMF interface                                                                           SAForum interfaces: HPI, CKPT, LCK, NTF, MSG, CLM, EVT                                                                Hidenets interfaces: …




                                                                                                                                                                                                                                                                               HA middleware mgmt
                  Fault tolerance manager
           QoS Adaptation          Reconfiguration &                                                                                    Cooperative Data                Mobile agent




                                                                                                                                                                                                                                                                                  SAForum IMM
             Manager              Maintenance manager                                                                                    Backup (LAAS)                 manager (AAU2)




                                                                                                                                                                                                                                                                                    interface:
              (FCUL)
                                Diagnostic manager                                                                                                                     Inconsistency estimation        NW context          Group communication         Server
         Replication
                                      (UNIFI)                                                                                                                              module (AAU8)            Repository (AAU1)            support              & Client ?
       manager (AAU6)
             (?)
  Lower layer
  middleware




                  Timeliness and
                  trustworthiness                                                                                                                   Proximity Map     Session          GW/network       Profile         Resource/Servic     Naming       QoS/differentiation
                                                                                                                                                       (LAAS)         control           selection     Management          e Discovery       Service       Manager + CAC
                      oracles
                                                                                                                 Communication adaptation manager
                                                        Performance monitoring (FCUL)
                                            3 Layer 4




                  Violation Detectors                                                                                                                                                                Transport layer functions (reliable data transfer,
                                                                                        Error Detection (AAU)




                        Trust and                                                                                                                                                                    congestion control, flow control, sequence, …)
                       Cooperation
                      Oracle (LAAS)
                                                                                                                             (AAU4)?




                       Freshness
                     Detector (UNIFI)                                                                                                                     Infrastructure Mobility
                                             Layer




                                                                                                                                                                                                                     GW Agent
                      Timely Timing
                                                                                                                                                           Support - client part                                   (ad-hoc mode)
                     Failure Detector                                                                                                                             (AAU9)                                                                      IP Routing
                                                                                                                                                                                               IP Forwarding and
                          (FCUL)                                                                                                                                                                                             (incl routing topology and routing tables)
                                                                                                                                                                                                Route Resilience
                      Distributed
                                                                                                                                                                                           (incl QoS / differentiation)
                     measurements                                                                                                                          Broadcast/Multicast/                                                  Ad-hoc topology
                         (FCUL)                                                                                                                             GeoCast (AAU3)                                                        control (AAU5)         Multi-channel/
                                                                                                                                                                                                                                                          Multi-radio
                     Authentication                                                                                                                                                               Traffic Failure                Neighbour                 Routing
                         (FCUL)                                                                                                                                                                     Detection                     discovery
                                            2




                    Reliable and Self-
                                              Layer




                      Aware Clock                                                                                                                                                                                                                             Multi-channel/
                         (UNIFI)
                                                                                                                                                                                                                                                               Multi-radio
                                                                                                                                                                                                                                                              Management
                                                                                                                                                                                                    Data Link and
                                                                                                                                                                                             Radio Resource Management                                 MAC / LLC
                Car control/             Positioning equipment +
                                                                                                                                                                                                     UMTS/GPRS                                            WLAN
                 actuators                 Internal car sensors
HIDENETS Applications
 In   no particular order:
      Floating Car Data
      Traffic Sign Extension
      Unusual Driver Behavior Warning
      Hazard Warning of Own Vehicle
      Platooning
      Distributed Black Box
      Streaming Data
Aplications
 Floating car data
   Floating car data (FCD) describes the process of
     collecting traffic flow information and calculating
     up-to-date information about the current traffic
     flow on roads.
   Impact alert
   Adaptive cruise control
Applications
 Unusual Driver Behaviour Warning
   Unusual driver behaviour warning comprises a
    number of applications that have different levels
    of relevance and safety impact. The warning is
    directly related to the behaviour of a car’s
    driver. The warning is broadcast from the car
    whose driver shows unusual driver behaviour
Applications
 Traffic Sign Extension
   Traffic signs constitute one of the means for
     traffic regulation, which is independent of the
     kind of road and traffic characteristics. This
     application consists of extending traffic signs by
      a) allowing centralized control of the information
         indicated by each sign (e.g. continuous control
         and adjustment of speed limits) and
      b) allowing one-way direct communication between
         signs and (some) near-by cars (those to which
         information is concerned and should be
         disseminated).
Applications
 Hazard Warning of the Own Vehicle
   This application is based on information received
    by a car from sources in its environment which
    are processed by the in-car equipment for the
    purpose of detecting potential hazards. Note
    that the hazard information is not provided by
    external sources directly but only indirectly by
    the car’s own system by evaluation of available
    information.
Applications
 Platooning
   This application provides both positional and
     velocity control of vehicles in order to operate
     safely as a platoon on a highway. A platoon is
     formed by two or more vehicles following each
     other closely, controlled by the vehicle at the
     head of the platoon.
Applications
 Distributed Black Box
   The “classical” black box can record
     informational data, such as: engine / vehicle
     speed (typically 5 seconds before impact), brake
     status (again here, 5 seconds before impact),
     throttle position(s), and even the state of the
     driver's seat belt switch (on/off). The
     combination of this information along with other
     engineering factors is indeed very valuable for
     motor vehicle accident investigation.
Applications
 Streaming Data
   Streaming data is transmission of information
     that is continuously updated like for instance
     positioning information or a patient’s pulse or
     blood pressure. Even though such an application
     has high quality and performance requirements,
     it does not depend on the correct reception of
     every single message. Relevant QoS attributes
     are throughput and data correctness. Data
     transmission can be both from the fixed network
     or originating from a vehicle.
HIDENETS seen from my perspective
 My interests:
   Cluster forming
   Replica candidate selection
   Distributed redundancy through state sharing
HIDENETS seen from my perspective
 Recent work on a cost metric to make replica
  candidate selection more efficient
   Reliable service provisioning in ad-hoc networks
     between cars is difficult
   Service replication helps improve availability
       Which car should be used for replication
    New cost metric based on position and speed of
     cars
 Why is it important to select good candidates?
   Easier management by less frequent changes in
    cluster membership
   Less signaling overhead if clusters are more
    stable
 Why not just let everyone share their state?
   All to all replication does not scale!
 Why dont you just send your state updates to a
  server in the internet?
   This is not always possible

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:1
posted:12/25/2012
language:Unknown
pages:24