Simulative Leistungsbewertung des IEEE Systems im openwns

Transcription

1 Simulative Leistungsbewertung des IEEE Systems im openwns 19. ComNets-Workshop Mobil- und Telekommunikation Dipl.-Ing. Karsten Klagges ComNets Research Group RWTH Aachen University 11. März 2011 Karsten Klagges ComNets, RWTH Aachen University ComNets 1

2 Outline Simulator Platform Simulator Architecture Functional Units Continuous Integration Source Control System Grid/Cluster Computing Overview Wrowser WiMAC module IEEE Protocol Overview WiMAC Component Karsten Klagges ComNets, RWTH Aachen University ComNets 2

3 Simulator Platform openwns is witten in C++ and provides a rich set of common software design patterns Design patterns from Gamma et. al. Modern Software Design A. Alexandrescu openwns is flexible Functional units allow flexible configuration of protocol layer Modular composition of protocol stack Simulation model is exchangeable even non-event driven simulations are possible openwns performs data post processing On-line data concentration (Histogram, PDF, CDF) Sorting of data by context Statistical data is accessible in database Karsten Klagges ComNets, RWTH Aachen University ComNets 3

4 Simulator Platform openwns is witten in C++ and provides a rich set of common software design patterns Design patterns from Gamma et. al. Modern Software Design A. Alexandrescu openwns is flexible Functional units allow flexible configuration of protocol layer Modular composition of protocol stack Simulation model is exchangeable even non-event driven simulations are possible openwns performs data post processing On-line data concentration (Histogram, PDF, CDF) Sorting of data by context Statistical data is accessible in database Karsten Klagges ComNets, RWTH Aachen University ComNets 3

5 Simulator Platform openwns is witten in C++ and provides a rich set of common software design patterns Design patterns from Gamma et. al. Modern Software Design A. Alexandrescu openwns is flexible Functional units allow flexible configuration of protocol layer Modular composition of protocol stack Simulation model is exchangeable even non-event driven simulations are possible openwns performs data post processing On-line data concentration (Histogram, PDF, CDF) Sorting of data by context Statistical data is accessible in database Karsten Klagges ComNets, RWTH Aachen University ComNets 3

6 Simulator Core Simulation Model Database Configuration PyConfig Tree Node #1 Load generator Network (IP) Node #2 Load generator Network (IP) Statistic Tools DLL DLL Event Scheduler PHY PHY Random Number Distribution Event Logging Radio Interference Simulation Engine (RISE) Figure: Simulator Architecture Karsten Klagges ComNets, RWTH Aachen University ComNets 4

7 Functional Units Functional units have been invented 1 to provide a flexible configuration of protocol layers by reusing atomic protocol functionality in multiple layers. #1 isaccepting senddata Receptor wakeup Deliverer ondata Definition of common functional unit interface to connect functional units Data is aggregated in compounds of functional unit commands Inter functional unit flow control #2 Compound Handler Command Type #3 Specifier Connector wakeup ondata isaccepting senddata Signal flow Data flow Figure: Functional Unit Aspects 1 Schinnenburg et. al., A Framework for Reconfigurable Functions of a Multi-Mode Protocol Layer, SDR Forum 2005 Karsten Klagges ComNets, RWTH Aachen University ComNets 5

8 Outgoing direction 1. Incoming wakeup from lower 2. Forward wakeup to next higher 3. isaccepting request from higher 4. Forward isaccepting request to next lower 5. senddata request from higher 6. Handle compound and forward to next lower Incoming Direction 1. ondata call from lower 2. Handle data and send to next higher #1 isaccepting senddata wakeup ondata Receptor Deliverer #2 Compound Handler Command Type #3 Specifier Connector wakeup ondata isaccepting senddata Signal flow Data flow Figure: Functional Unit Flow Control Karsten Klagges ComNets, RWTH Aachen University ComNets 6

9 Outgoing direction 1. Incoming wakeup from lower 2. Forward wakeup to next higher 3. isaccepting request from higher 4. Forward isaccepting request to next lower 5. senddata request from higher 6. Handle compound and forward to next lower Incoming Direction 1. ondata call from lower 2. Handle data and send to next higher #1 isaccepting senddata wakeup ondata Receptor Deliverer #2 Compound Handler Command Type #3 Specifier Connector wakeup ondata isaccepting senddata Signal flow Data flow Figure: Functional Unit Flow Control Karsten Klagges ComNets, RWTH Aachen University ComNets 6

10 Outgoing direction 1. Incoming wakeup from lower 2. Forward wakeup to next higher 3. isaccepting request from higher 4. Forward isaccepting request to next lower 5. senddata request from higher 6. Handle compound and forward to next lower Incoming Direction 1. ondata call from lower 2. Handle data and send to next higher #1 isaccepting senddata wakeup ondata Receptor Deliverer #2 Compound Handler Command Type #3 Specifier Connector wakeup ondata isaccepting senddata Signal flow Data flow Figure: Functional Unit Flow Control Karsten Klagges ComNets, RWTH Aachen University ComNets 6

11 Outgoing direction 1. Incoming wakeup from lower 2. Forward wakeup to next higher 3. isaccepting request from higher 4. Forward isaccepting request to next lower 5. senddata request from higher 6. Handle compound and forward to next lower Incoming Direction 1. ondata call from lower 2. Handle data and send to next higher #1 isaccepting senddata wakeup ondata Receptor Deliverer #2 Compound Handler Command Type #3 Specifier Connector wakeup ondata isaccepting senddata Signal flow Data flow Figure: Functional Unit Flow Control Karsten Klagges ComNets, RWTH Aachen University ComNets 6

12 Outgoing direction 1. Incoming wakeup from lower 2. Forward wakeup to next higher 3. isaccepting request from higher 4. Forward isaccepting request to next lower 5. senddata request from higher 6. Handle compound and forward to next lower Incoming Direction 1. ondata call from lower 2. Handle data and send to next higher #1 isaccepting senddata wakeup ondata Receptor Deliverer #2 Compound Handler Command Type #3 Specifier Connector wakeup ondata isaccepting senddata Signal flow Data flow Figure: Functional Unit Flow Control Karsten Klagges ComNets, RWTH Aachen University ComNets 6

13 Outgoing direction 1. Incoming wakeup from lower 2. Forward wakeup to next higher 3. isaccepting request from higher 4. Forward isaccepting request to next lower 5. senddata request from higher 6. Handle compound and forward to next lower Incoming Direction 1. ondata call from lower 2. Handle data and send to next higher #1 isaccepting senddata wakeup ondata Receptor Deliverer #2 Compound Handler Command Type #3 Specifier Connector wakeup ondata isaccepting senddata Signal flow Data flow Figure: Functional Unit Flow Control Karsten Klagges ComNets, RWTH Aachen University ComNets 6

14 Outgoing direction 1. Incoming wakeup from lower 2. Forward wakeup to next higher 3. isaccepting request from higher 4. Forward isaccepting request to next lower 5. senddata request from higher 6. Handle compound and forward to next lower Incoming Direction 1. ondata call from lower 2. Handle data and send to next higher #1 isaccepting senddata wakeup ondata Receptor Deliverer #2 Compound Handler Command Type #3 Specifier Connector wakeup ondata isaccepting senddata Signal flow Data flow Figure: Functional Unit Flow Control Karsten Klagges ComNets, RWTH Aachen University ComNets 6

15 Outgoing direction 1. Incoming wakeup from lower 2. Forward wakeup to next higher 3. isaccepting request from higher 4. Forward isaccepting request to next lower 5. senddata request from higher 6. Handle compound and forward to next lower Incoming Direction 1. ondata call from lower 2. Handle data and send to next higher #1 isaccepting senddata wakeup ondata Receptor Deliverer #2 Compound Handler Command Type #3 Specifier Connector wakeup ondata isaccepting senddata Signal flow Data flow Figure: Functional Unit Flow Control Karsten Klagges ComNets, RWTH Aachen University ComNets 6

16 Simulator Platform Simulator Architecture Functional Units Continuous Integration Source Control System Grid/Cluster Computing Overview Wrowser WiMAC module IEEE Protocol Overview WiMAC Component Karsten Klagges ComNets, RWTH Aachen University ComNets 7

17 Continuous Integration with buildbot Continuous Integration is a software development practice where members of a team integrate their work frequently, usually each person integrates at least daily - leading to multiple integrations per day. Martin Fowler Development branch is always functional Unit tests on functional unit level with CppUnit testing suite System tests of full protocol stacks in selected scenarios Subsequent patches must not interfere with existing functionality Figure: Buildbot web interface Karsten Klagges ComNets, RWTH Aachen University ComNets 8

18 Source Control System Revision control provides both, collaboration of developers and archiving of achieved work. Based on Canonical s bzr also known as bazaar Everybody develops on the same codebase Small changes are developed on frequent commit base Big inventions are developed on separated branches Rev 25 Rev 24 Rev 23 Rev 22 Branches are reintegrated after successful testing Rev 21 Rev 22.2 Rev 22.1 Karsten Klagges ComNets, RWTH Aachen University ComNets 9

19 Grid/Cluster Computing Grid computing Grid computing combines distributed computer resources that are loosely coupled to compute programs in parallel. Simulation campaigns are simulated on the compute cluster, driven by the SUN gridengine Gridengine runs on Dell Power-Edge blades 16 blades with 2 Quad-Core Intel Xeon 2.5GHz up to 128 simulations run in parallel Gridengine may also run on desktop computers Karsten Klagges ComNets, RWTH Aachen University ComNets 10

20 openwns development cycle at a glance Karsten Klagges ComNets, RWTH Aachen University ComNets 11

21 Figure: Wrowser data viewer Simulation results are viewed with openwns data browser: Wrowser Aggregation of simulation results of multiple runs Statistical confidence of results shown as confidence interval Export of data as png, pdf, emf, MATLAB data and many more Karsten Klagges ComNets, RWTH Aachen University ComNets 12

22 Simulator Platform Simulator Architecture Functional Units Continuous Integration Source Control System Grid/Cluster Computing Overview Wrowser WiMAC module IEEE Protocol Overview WiMAC Component Karsten Klagges ComNets, RWTH Aachen University ComNets 13

23 Convergence Sublayer Common Part Sublayer Radio Resource Control and Management (RRCM) Medium Access Control (MAC) Data Plane Karsten Klagges ComNets, RWTH Aachen University ComNets 14

24 Karsten SFH: Klagges Super Frame Header ComNets, RWTH Aachen University ComNets 15 WiMAC module WiMAC (WIMAX MAC) is the IEEE MAC and PHY model in the openwns. WiMAC makes heavy use of the concept Radio Resource Control and Management Service flow and Connection Management Interference Management BW Request Convergence Sublayer Data Plane Convergence Classifier Buffer ARQ CRC Error Model MAC / Physical Layer SFH DL MAP UL MAP DL Scheduler Frame Multiplexer UL Scheduler Figure: WiMAC Functional Unit Network

25 Thank you Karsten Klagges ComNets, RWTH Aachen University ComNets 16