sd-seminar-5-11-2008

Document Sample
sd-seminar-5-11-2008 Powered By Docstoc
					   An Enriched Finite State Machine
  Model-based Formalism for Layer-5
   Internet Protocols Modelling – An
Investigation on Protocol Performance
            Alexandru Catalin Ionescu
                   05.10.2008
                                   Content
• Surf the MSc Thesis
    •   Background, Problems and Methods
    •   (Mobile) Internet Communications
    •   Packet Data Access in Mobile Networks
    •   Layered Network Architectures
    •   Protocol Modelling
    •   Protocol Modelling for Performance Measurements
    •   Conclusions
• Personal Experience (How I wrote my MSc. Thesis)
• Surf the MSc Thesis
    • Background, Problems and Methods
    •   (Mobile) Internet Communications
    •   Packet Data Access in Mobile Networks
    •   Layered Network Architectures
    •   Protocol Modelling
    •   Protocol Modelling for Performance Measurements
    •   Conclusions
• Personal Experience (How I wrote my MSc. Thesis)
                              Background
             Mobile Services and Applications
•   New Mobile Networks providing High Data Rates
•   Fixed Internet Services are coming into the Mobile Environment
•   It is all about instant access to information
•   Ensure the Quality of Service (QoS)
     • Different applications need different QoS
     • In Mobile Networks resources are never enough 
• The transition From GSM to UMTS
• New Internet Engineering Task Force Multimedia Architecture


    The need to provide a model that allows a developer to first
    analyze and eventually optimize the application protocol.
           IETF Multimedia Architecture

Video Equipment

                   RTP / RTCP
                   Encapsulation
Audio Equipment

                                              Packet
                                   TCP / IP   Network
  User Data                        UPD / IP
  Applications
                     Security
                  SIP
System Control             SDP
                  SDP
 User Interface       RTCP
                      RSVP
               Problems and Methods

Question 1: How could one formally model the communication protocols
according to a specific use case?

Question 2: How to estimate the generated network traffic for a specific use
case when using a certain communication protocol?

Question 3: How to improve a protocol in order to decrease the generated
network traffic?
• Surf the MSc Thesis
    • Background, Problems and Methods
    • (Mobile) Internet Communications
    • Packet Data Access in Mobile Networks
    • Layered Network Architectures
    • Protocol Modelling
    • Protocol Modelling for Performance Measurements
    • Conclusions
• Personal Experience (How I wrote my MSc. Thesis)
           Mobile Internet Communications
• Mobile Communication - A way we see the world
• Demand for Complete Connectivity
• Service Evolution
    •   First was the voice
    •   Then came the basic messaging (SMS)
    •   Data transfer followed (FAX)
    •   Internet Applications in Mobile handsets
          Performance in Mobile Networks
• Performance (efficiency) is hard to define
    formally, analyze and not easy to accept
•   Performance = Service Efficiency = Service       Effective
    Usability
•   Mobile Communications transmission Media
    is faulty – Reliable protocols
     • Error detection
     • Error correction
• Topological Failures in Mobile Networks are
    common – Resilient protocols
     • Ability to test the Link                Resilient   Reliable
• Effectives is defined as Easy to implement
    and Use
     • Easy and Use depend on the Context.
             Resilience in Mobile Networks

• Communication between End-points goes over various different physical
    media
•   Sending data - the first leg is the radio network
•   Receiving Data - the last leg of physical communication is again the radio
    wave
•   Level-5 Protocols cannot solve link loss in Mobile Networks
•   Resilience of Level-5 protocols translates to the ability of the protocol to test
    the availability of the link
    Example: A Voice over IP application uses SIP in order to perform the so
    called Registration. The Client registers its “location” to a Registrar Server.
    The server then knows how to contact the Client when there is an incoming
    call. In a GSM / UMTS environment the SIP protocol offers the possibility to
    both Client and Server to query the availability of each other. In case the
    Server notices a connection break it can reject all the incoming calls. At the
    same time a Client can try to re-Register.
             Reliability in Mobile Networks
• Radio resources in Mobile Networks are the most likely to create connectivity
    problems (e.g. bad signal or signal loss)
•   Bad signal quality – lower level protocols solve the reliability problem
     • Correction
     • Retransmission
     • All these at Level-4 and lower
• Signal loss
     • No ability to handle correction and retransmissions
     • Level-5 needs to cope with these case
• One could observe that Resilience and Reliability have the similar meaning in
    case of Level-5 protocols in mobile networks.
        Effectiveness in Mobile Networks
• Monetary vs. Non-monetary aspects
• An effective Level-5 protocol is network friendly
    • Light
    • Not to case “too much” traffic
• Monetary aspect
    • Direct – the amount of data sent and received is to expensive –
    • Indirect – the amount of data is too high and consumes resources that will otherwise
      be useful to other services
• Non-monetary aspects need to be considered as well.
    • Mobile device resources are not costly but limited (e.g. processing power, battery)
• Surf the MSc Thesis
    • Background, Problems and Methods
    • (Mobile) Internet Communications
    • Packet Data Access in Mobile Networks
    • Layered Network Architectures
    • Protocol Modelling
    • Protocol Modelling for Performance Measurements
    • Conclusions
• Personal Experience (How I wrote my MSc. Thesis)
 Resource Allocation in Mobile Networks
• QoS classes
    •   Conversational – Conversational pattern
    •   Interactive – preserves the time relation between the informational elements
    •   Streaming – Request / Response pattern
    •   Background – Data is not expected with strict time limits
• Protocol performance within a QoS class
    • Performance of a protocol needs to be considered in the context of the QoS class
        (e.g. A protocol can offer good performance when used by applications running
        under the Conversational Class but could be the wrong tool for the job within the
        Interactive Class)
• Mobile Networks allocate the communication resources differently, depending
  on the QoS class.
    • In practice, means that an application running within the Conversational Class has
        more resources to spare compared to an application running within the Interactive
        Class
    •   Protocols that have already been defined and perform well are not necessarily the
        best choice for the new applications being deployed.
                 Communication Channels
• Common Channels
   • Caring the signalling within the network but In some cases they are also used for
       user data
   •   Low setup time - Fast to establish in order to send and receive data and then tear-
       down
   •   Cannot handle the so called soft handover – no cell to cell hoping
   •   The network will not allocate them when the data amounts to be sent or received is
       high - work with small individual packets
• Dedicated Channels
   •   Exact opposite of the Common Channels
   •   Long time to set up
   •   High bit rates and the bit rate can be changed during the transmission
   •   Improved radio performance
   •   Any protocol can be used on these channels
   •    Issues can arise only when the entities involved in the communication expect
       responses to their request within a certain time interval
• Shared Channels
          Communication Channels (cont)
• Shared Channels
   •   Share a channel in time between different users
   •   Bit rate is lower in comparison with the achieved rates on the dedicated channels
   •   Good for applications that generate bursty traffic
   •   Capacity of the air interface is shared among many users at the same time
• Surf the MSc Thesis
    • Background, Problems and Methods
    • (Mobile) Internet Communications
    • Packet Data Access in Mobile Networks
    • Layered Network Architectures
    • Protocol Modelling
    • Protocol Modelling for Performance Measurements
    • Conclusions
• Personal Experience (How I wrote my MSc. Thesis)
                       Layered Model


          Application           Peer Protocol        Application

          Presentation                               Presentation
 Upper
Network
            Session                                    Session
Layers
           Transport                                  Transport

           Network                                    Network

Lower      DataLink                                   DataLink
Network
Layers      Physical                                   Physical


                            Physical Media for OSI
               Relay Open System

Application            Peer Protocol        Application

Presentation                                Presentation

  Session                                     Session

 Transport                                   Transport

 Netw ork                 Netw ork           Netw ork

 DataLink                 DataLink           DataLink

  Physical                 Physical           Physical




                   Physical Media for OSI
               IP Suite Stack Host to Host
              communication over Internet

Application                                                                  Application
                                PEER TO PEER Communication
Transport                                                                    Transport


 Network                   Network                     Network                Network


Data Link                 Data Link                   Data Link              Data Link



              Internet                    Internet                Internet


                         Physical Layer
Internet Reference Model in Mobile
 Mobile Device Networks   Application Server




 Application                                          Application
                    PEER TO PEER Communication
  Transport                                           Transport


  Network                    Network                   Network

                GPRS                      Ethernet
  Data Link                 Data Link                 Data Link


                Radio
                                           Internet
               Network
                          Gateway GPRS
                           Support Node
• Surf the MSc Thesis
    •   Background, Problems and Methods
    •   (Mobile) Internet Communications
    •   Packet Data Access in Mobile Networks
    •   Layered Network Architectures
    • Protocol Modelling
    • Protocol Modelling for Performance Measurements
    • Conclusions
• Personal Experience (How I wrote my MSc. Thesis)
                        Protocol Modelling
• Informal methods
    • Successfully used
    • Do not handle well high complexity
• Formal methods
    • Analyze rigorously and efficiently
    • Several papers in the literature
    • Extremely valuable for Validation and Verification
• Performance evaluation has not been addressed much
    • Done in an informal manner
    • Acknowledge issues but do not offer clear fact and figures
    • Lack of formalism makes comparison between two set of results impossible
            Protocol Modelling Techniques
• High-level Programming Languages
    •   Convenient tool – Communication party represented as a program
    •   Can represent data, variables, counters, etc…
    •   Cannot represent complex structures – general metrics
    •   Do not serve the purpose of calculating traffic
• Finite State Machines
    • Use in the early stages of Protocol definition
    • Single-machine model vs. Multi-machine approach
    • The model is applicable to any protocol having a finite number parties involved in
        communication
• Petri Nets
    • Petri Nets can model a protocol that has an infinite number of states
    • We can develop a model that allows messages to have size using the so called
        Coloured Petri Nets extension
    •   The size of the graph can become a problem
         • Even in the case of simple protocols this can result in growing to complexities that are not
           easy to manage
                     Finite State Machine Model

                 Process 1                                 Process 2
                  (user)                                    (server)


                 Not Logged-in                                 IDLE
- LOGIN                                          + LOGIN
                        + ACK_LOGOUT                             - ACK_LOGOUT

                           + ACK_LOGIN                                   + LOGOUT


          Wait                   Logged-in   Service_Request       Authenticated

                       - LOGOUT                            - ACK_LOGIN
                                  Petri Nets Model
            Client                        Network                    Server
Login                                            Msg1                           Idle


                     Send Login
                                                        Login Received
                 Wait1
                                                         Service Login Request
Logged-in
                                          Ack1
                                                         Login Ack
                                                                         Authenticated

                 Ack Login Received

                                                 Msg2


               Send Logout                                Logout Received
Logout
                Wait2                                      Service Logout Request


                                          Ack2
                                                                           Logged Out
                                                        Logout Ack
                Ack Logout Received
 Logged Out
• Surf the MSc Thesis
    •   Background, Problems and Methods
    •   (Mobile) Internet Communications
    •   Packet Data Access in Mobile Networks
    •   Layered Network Architectures
    •   Protocol Modelling
    • Protocol Modelling for Performance Measurements
    • Conclusions
• Personal Experience (How I wrote my MSc. Thesis)
Protocol Measurement System



  Context
                        The System

   Entity A             Network             Entity B




              Measure the traffic flowing
                through the network
               Handling an external event in the
                  measuring environment
                                                    The System

                 Entity A                               Network                        Entity B


    External Event
        (data)
                       The protocol implementation
                       transforms the external event in a
                       protocol primitive / message
                       packing the data carried by the
                       event

                                Message 1
                                                                    Relay Message 1

                                                              Entity B Processes the
                                                              Message and creates
                                                              response

                     Relay Response Message 1                     Response Message 1
External Response
      (data)                                                                                External Event
                                                                                                  (data)
        A User – Client – Server Model


                            Access to information


User

                                   The System

              Entity A                                 Entity B
Use client                         Network
              (client)                                 (server)
Application



                         Measure the traffic flowing
                           through the network
Connect / Subscribe / Notify / Disconnect


                         Connect

User

                          The System

              Entity A                 Notify   Entity B
Use client               Network
              (client)                          (server)
Application



                           Disconnect
                          Use Case Execution
                                    The System

                       Entity A                        Entity B

       (1) Start
                       (client)                        (server)
User   use case
                   stable state 0                    stable state 0
                                     (2) message 1
                            (3)                              (3)

                   client state x                    server state x
                                     (4) message 2
                              (5)                             (5)

                   client state y    (6) message 3   server state y

                                                                  (7)
                              (7)
                                                     server state z
                   client state z    (8) message 4

                              (9)                              (9)

                   stable state 1
                                                     stable state 1
A Graphical Representation of FSP
        Process 1                                  Process 1
Time       (user)                                   (server)
       Not Logged-in                                  IDLE
                        + LOGIN     - LOGIN


           Wait                                  Service_Request
                       - ACK_LOGIN + ACK_LOGIN


        Logged-in                                 Authenticated
                        + LOGOUT   - LOGOUT


           Wait                                  Service_Request
                                  +
                       - ACK_LOGOUT ACK_LOGOUT



       Not Logged-in
                                                      IDLE
Abstract Time-line Mode for Presence
       Presence Client                            Presence Server


           IDLE                                           IDLE
Time
                            Connect to Service


          State 1                                        State 1
                      Publish Presence Information


          State 2                                        State 2
                    Subscribe for Presence Information

          State 3
                                                         State 3

               State 1 – Connected
               State 2 – Connected & Published
               State 3 - Connected & Published & Subscribed
SIMPLE Connection to Presence Service
         Presence Client                           Presence Server
                            Connect to Service
             IDLE                                       IDLE
  Time
                            SIP Register

                            SIP 401 Unauthorized

                            SIP Register

                            SIP 200 OK


            State 1                                     State 1

                 State 1 – Connected
                 State 2 – Connected & Published
                 State 3 - Connected & Published & Subscribed
       SIMPLE Presence Publication
       Presence Client                            Presence Server
                                    …
           IDLE                                         IDLE
Time

                     Publish Presence Information
          State 1                                      State 1
                                SIP Publish
                                SIP 200 OK
                         Publication expires or
                    user changes the Information
                                SIP Publish
                                SIP 200 OK
               State 1 – Connected
               State 2 – Connected & Published
               State 3 - Connected & Published & Subscribed
       SIMPLE Presence Subscription
       Presence Client                              Presence Server
                                    …
            IDLE                                           IDLE
Time
                                    …
           State 1                                        State 1

           State 1    Publish Presence Information
                                                          State 1
                                SIP Subscribe
                                SIP 200 OK
                                SIP Notify
                                SIP 200 OK

                          Subscription expires or
                     Connection is lost the Information

                State 1 – Connected
                State 2 – Connected & Published
                State 3 - Connected & Published & Subscribed
                                    First Results
                UC 1            UC2            UC 3       UC 4       UC 5

Subscription          86400     62208000    (1 62208000   62208000   62208000
   LifeTime           (1 day)      month)

  Refresh       7200            7200           43200      43200      43200
  Interval
  (hours)
Presence        1               1              1          1          2
   Changes
   (per hour)
Presentities per 10             10             10         20         20
   watcher
Number     of 588               17292          16092      30492      60732
  Messages
Amount of data 1                31.65          30.11      56.87      113.43
  (Mb)
                                   Improvements
                         UC 1          UC2         UC 3       UC 4       UC 5
Subscription                 86400     62208000 (1 62208000   62208000   62208000
   LifeTime                  (1 day)      month)


Refresh Interval         7200          7200        43200      43200      43200
(hours)

Presence    Changes 1                  1           1          1          2
   (per hour)

Presentities        per 10             10          10         20         20
   watcher
Number of Messages       564           16572       15792      39372      60612

Amount         of   data 0.41          11.9        11.5       21.64      43.19
  (Mb)
• Surf the MSc Thesis
    •   Background, Problems and Methods
    •   (Mobile) Internet Communications
    •   Packet Data Access in Mobile Networks
    •   Layered Network Architectures
    •   Protocol Modelling
    •   Protocol Modelling for Performance Measurements
    • Conclusions
• Personal Experience (How I wrote my MSc. Thesis)
                                Conclusions
                                         General
• New directions in Mobile Communications are shaped by the business
    requirements optimization needs for profitable investments
•   Increase complexity – introduce fixed-Internet technologies to the Mobile
    Domain
•   IETF protocols and especially the level-5 ones are becoming the de facto
    protocols for mobile services
•   Solve the performance or efficiency evaluation
•   Need to understand how a certain technology works
     • Improves the performance
     • Quality of the Services
     • Impact on the operational costs
                                Conclusions
                             Protocol Modelling
• Finite State Machines and Petri Nets - in their original form - did not suite all
  our needs
    • Good for validation and verification
    • Not good for measurements due to lack of metric
• Colored Petri Nets
    • Good for measurements
    • Hard to use
    • Potentially will alienate some of the stakeholders
• Improvement of the Finite State Machines - METRICS
                               Future Work
• The performance measurements need to address two aspects
    • Number of messages sent over the network
    • Amount of data exchanged in the communication
• FSP model proposed by this work is able to handle well the first aspect
• The amount of data aspect is not handled in an elegant way
• An improved model to deal with this aspect is needed
    • Coloured Petri Nets are a good candidate for this particular aspect
• Extended versions of the general automata, such as X-machines, provide
  computability and therefore direct implementability and testing.
• Surf the MSc Thesis
    •   Background, Problems and Methods
    •   (Mobile) Internet Communications
    •   Packet Data Access in Mobile Networks
    •   Layered Network Architectures
    •   Protocol Modelling
    •   Protocol Modelling for Performance Measurements

• Personal Experience (How I wrote my MSc. Thesis)
                             The Subject
•   Choose the subject early enough
•   Stick with it – do not give up when things get tough
•   Get your references well in advance
•   Read all your references and the some
•   Attend the Scientific Writing in Human-Computer Interaction (Kari-Jouko Räihä
    OR Howell Istance)
                            Work and Study
• A real problem
• How to get the work done and how to get the work done 
• Difficulties
    • Share time in the office / time at the University / time at home / travelling
    • Confidentiality management
    • Relationship with the management
• Time management
    • How many hours do you allocate per week to your studies
    • Try to find a day when you spend most of those hours – preferably a working day
    • Save your evening for reading
• The feeling after    

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:0
posted:5/17/2012
language:English
pages:46
fanzhongqing fanzhongqing http://
About