Performance Model Checking Scenario-Aware Dataflow Bart Theelen, Marc Geilen, Jeroen Voeten.

Slides:



Advertisements
Similar presentations
An improved on-the-fly tableau construction for a real-time temporal logic Marc Geilen 12 July 2003 /e.
Advertisements

Technische universiteit eindhoven PROGRESS 11 December 2002www.ics.ele.tue.nl/~btheelen1 Performance Modelling of Complex Hardware/Software Systems B.D.
Knowledge Based Synthesis of Control for Distributed Systems Doron Peled.
An Introduction to Petri Nets
Introduction to Petri Nets Hugo Andrés López
1 Analysis of workflows : Verification, validation, and performance analysis. Wil van der Aalst Eindhoven University of Technology Faculty of Technology.
A university for the world real R © 2009, Chapter 3 Advanced Synchronization Moe Wynn Wil van der Aalst Arthur ter Hofstede.
Distributed Markov Chains P S Thiagarajan School of Computing, National University of Singapore Joint work with Madhavan Mukund, Sumit K Jha and Ratul.
The cardiac pacemaker – SystemJ versus Safety Critical Java Heejong Park, Avinash Malik, Muhammad Nadeem, and Zoran Salcic. University of Auckland, NZ.
Marc Geilen, Eindhoven University of Technology, Information and Communication Systems 1 Object-Oriented Modelling and Specification.
Formal Modelling of Reactive Agents as an aggregation of Simple Behaviours P.Kefalas Dept. of Computer Science 13 Tsimiski Str Thessaloniki Greece.
Game-theoretic approach to the simulation checking problem Peter Bulychev Vladimir Zakharov Lomonosov Moscow State University.
Timed Automata.
Requirements on the Execution of Kahn Process Networks Marc Geilen and Twan Basten 11 April 2003 /e.
Anna Philippou Department of Computer Science University of Cyprus Joint work with Mauricio Toro Department of Comp. Sc. EAFIT University Christina Kassara.
Technische universiteit eindhoven 25 September 2002www.ics.ele.tue.nl/~btheelen1 Using the SHE Method for UML-based Performance Modelling B.D. Theelen.
Overview This project applies the tagged-signal model to explain the semantics of piecewise continuous signals. Then it illustrates an operational way.
Petri net modeling of biological networks Claudine Chaouiya.
PTIDES: Programming Temporally Integrated Distributed Embedded Systems Yang Zhao, EECS, UC Berkeley Edward A. Lee, EECS, UC Berkeley Jie Liu, Microsoft.
Discrete Abstractions of Hybrid Systems Rajeev Alur, Thomas A. Henzinger, Gerardo Lafferriere and George J. Pappas.
A denotational framework for comparing models of computation Daniele Gasperini.
1 Ivan Lanese Computer Science Department University of Bologna Roberto Bruni Computer Science Department University of Pisa A mobile calculus with parametric.
1 Petri Nets H Plan: –Introduce basics of Petri Net models –Define notation and terminology used –Show examples of Petri Net models u Calaway Park model.
February 12, 2009 Center for Hybrid and Embedded Software Systems Encapsulated Model Transformation Rule A transformation.
Model-based Testing.
Design of Fault Tolerant Data Flow in Ptolemy II Mark McKelvin EE290 N, Fall 2004 Final Project.
Models of Computation for Embedded System Design Alvise Bonivento.
A Denotational Semantics For Dataflow with Firing Edward A. Lee Jike Chong Wei Zheng Paper Discussion for.
1 Ivan Lanese Computer Science Department University of Bologna Italy Concurrent and located synchronizations in π-calculus.
February 12, 2009 Center for Hybrid and Embedded Software Systems Model Transformation Using ERG Controller Thomas H. Feng.
Verifying Distributed Real-time Properties of Embedded Systems via Graph Transformations and Model Checking Gabor Madl
System-Level Types for Component-Based Design Paper by: Edward A. Lee and Yuhong Xiong Presentation by: Dan Patterson.
Jun. Sun Singapore University of Technology and Design Songzheng Song and Yang Liu National University of Singapore.
1 Real-Time Queueing Network Theory Presented by Akramul Azim Department of Electrical and Computer Engineering University of Waterloo, Canada John P.
Lucent Technologies – Proprietary Use pursuant to company instruction Learning Sequential Models for Detecting Anomalous Protocol Usage (work in progress)
Multi-Agent Model to Multi-Process Transformation A Housing Market Case Study Gerhard Zimmermann Informatik University of Kaiserslautern.
ECE 720T5 Winter 2014 Cyber-Physical Systems Rodolfo Pellizzoni.
Voicu Groza, 2008 SITE, HARDWARE/SOFTWARE CODESIGN OF EMBEDDED SYSTEMS Hardware/Software Codesign of Embedded Systems Voicu Groza SITE Hall, Room.
Planning and Verification for Stochastic Processes with Asynchronous Events Håkan L. S. Younes Carnegie Mellon University.
Parallel Programming Models Jihad El-Sana These slides are based on the book: Introduction to Parallel Computing, Blaise Barney, Lawrence Livermore National.
- 1 - Embedded Systems - SDL Some general properties of languages 1. Synchronous vs. asynchronous languages Description of several processes in many languages.
CY2003 Computer Systems Lecture 7 Petri net. © LJMU, 2004CY2003- Week 72 Overview Petri net –concepts –Petri net representation –Firing a transition –Marks.
A Methodology for Architecture Exploration of heterogeneous Signal Processing Systems Paul Lieverse, Pieter van der Wolf, Ed Deprettere, Kees Vissers.
Page 1 Analysis of Asynchronous Systems Steven P. Miller Michael W. Whalen {spmiller, Advanced Computing Systems Rockwell.
Conformance Test Experiments for Distributed Real-Time Systems Rachel Cardell-Oliver Complex Systems Group Department of Computer Science & Software Engineering.
Petri Nets Lecturer: Roohollah Abdipour. Agenda Introduction Petri Net Modelling with Petri Net Analysis of Petri net 2.
Dana Nau: Lecture slides for Automated Planning Licensed under the Creative Commons Attribution-NonCommercial-ShareAlike License:
A Quantitative Trust Model for Negotiating Agents A Quantitative Trust Model for Negotiating Agents Jamal Bentahar, John Jules Ch. Meyer Concordia University.
CSCI1600: Embedded and Real Time Software Lecture 11: Modeling IV: Concurrency Steven Reiss, Fall 2015.
Dana Nau: Lecture slides for Automated Planning Licensed under the Creative Commons Attribution-NonCommercial-ShareAlike License:
CAP 4800/CAP 5805: Computer Simulation Concepts
Modelling and Analysis of Time-related Properties in Web Service Compositions Raman KazhamiakinParitosh K. PandyaMarco Pistore
High Performance Embedded Computing © 2007 Elsevier Chapter 1, part 3: Embedded Computing High Performance Embedded Computing Wayne Wolf.
From Natural Language to LTL: Difficulties Capturing Natural Language Specification in Formal Languages for Automatic Analysis Elsa L Gunter NJIT.
High Performance Embedded Computing © 2007 Elsevier Lecture 4: Models of Computation Embedded Computing Systems Mikko Lipasti, adapted from M. Schulte.
SystemC Semantics by Actors and Reduction Techniques in Model Checking Marjan Sirjani Formal Methods Lab, ECE Dept. University of Tehran, Iran MoCC 2008.
CS5270 Lecture 41 Timed Automata I CS 5270 Lecture 4.
Traffic Simulation L2 – Introduction to simulation Ing. Ondřej Přibyl, Ph.D.
Marilyn Wolf1 With contributions from:
Polynomial analysis algorithms for free-choice workflow nets
2. Specification and Modeling
Graph Coverage for Specifications CS 4501 / 6501 Software Testing
CAP 4800/CAP 5805: Computer Simulation Concepts
CAP 4800/CAP 5805: Computer Simulation Concepts
SS 2018 Software Verification ML, state machines
Internet of Things A Process Calculus Approach
Statistical Model-Checking of “Black-Box” Probabilistic Systems VESTA
Linköping University, IDA, ESLAB
Deadlock Detection for Distributed Process Networks
CAP 4800/CAP 5805: Computer Simulation Concepts
Presentation transcript:

Performance Model Checking Scenario-Aware Dataflow Bart Theelen, Marc Geilen, Jeroen Voeten

2 Overview Dataflow Formalisms Timed Probabilistic Systems Performance Model Checking Experimental Results Conclusions & Outlook

Dataflow Formalisms Example digital signal processing areas 3 Streaming Multi-MediaLoop-Control in Mechatronics Dataflow formalisms describe task graphs where potential parallelism is made explicit MP3 Decoder

Dataflow Formalisms: Expressivity vs Analyzability 4 Stuijk, et al. Scenario-Aware Dataflow: Modeling, Analysis and Implementation of Dynamic Applications. SAMOS’11 Synchronous Dataflow (Weighted Marked Graphs) Kahn Process Networks Scenario-Aware Dataflow

Scenario-Aware Dataflow(SADF) Scenario = operation modes with similar resource usage Detectors control processes by sending scenario-valued tokens Detectors contain automata to capture scenario occurrences Real-life: data-dependent control behaviour (normal state machine) Modelling worst/best-case only: non-deterministic state machine Modelling worst/best-case & average-case: Markov chain 5 kernel data channel detector RateIP0P0 PxPx a001 b00x c991x d101 e 0x x = {30, 40, 50,60, 70, 80, 99} MPEG-4 Decoder control channel rate tokens

Processes run in parallel according to ‘extended actor semantics’ 1.Determine scenario depending on 1.Kernels & Detectors: scenario-valued control tokens 2.Detectors: next state of Markov chain 2.Wait until sufficient tokens available 3.Perform the actual task (sample from discrete time distribution) 4.Produce and consume tokens Scenario-Aware Dataflow(SADF) 6 RateIP0P0 PxPx a001 b00x c991x d101 e 0x x = {30, 40, 50,60, 70, 80, 99} MPEG-4 Decoder

Timed Probabilistic Systems(TPS) Compositional semantic model with guarded (interactive) action transitions probabilistic transitions deterministic time transitions Alternates action/time transitions with probabilistic fan-out Pattern for generic discrete execution time distributions time advances exactly t i time units with probability p i for i=1,…,n 7 a p t t1t1 τ p1p1 pnpn 1 1 tntn

Illustrative Example 8 Rateς1ς1 ς2ς2 a20 b10 ProcessScenario Execution Time Probability Aς1ς1 198/11 511/11 572/11 B ς1ς1 55/ /19 472/19 ς2ς2 131/ /15 631/15 D ς1ς1 34/14 131/14 299/14 ς2ς2 73/4 191/4 TPS for Kernel ATPS for Detector D

Semantic Properties Model checking based on (relevant) state-space Exploit semantic properties to limit state-space explosion SADF satisfies various semantic properties Time additivity, action persistency, action urgency, action determinacy Only non-determinism between actions as a result of concurrency Policy for resolving non-determinism does not effect net behaviour 9 policy for resolving non- determinism may however effect performance result

Performance Model Checking Direct computation of quantitative results based on model checking techniques Broad variety of performance metrics Mostly complex reward-based properties Mostly time-related properties 10 MetricType Probabilities Relevant Scope Max Buffer OccupancyWorst CaseNoAll States Min/Max Response DelayBest/Worst CaseNoTransient Min/Max Inter-Firing DelayBest/Worst CaseNoAll States Response Deadline Miss ProbabilityProbabilistic ReachabilityYesTransient Expected Response DelayExpected ReachabilityYesTransient ThroughputEvent RateYesSteady State Periodic Deadline Miss Probability Sample Average / Expected Reachability YesSteady State Average Inter-Firing LatencySample AverageYesSteady State Variance in Inter-Firing LatencySample VarianceYesSteady State Average Buffer OccupancyTime-Weighted AverageYesSteady State Variance in Buffer OccupancyTime-Weighted VarianceYesSteady State Policy for resolving non- determinism only affects Max Buffer Occupancy

Model Checking Strategy - Theory 11 TPS per SADF Process Discrete Markov Chain Move Transition Labels into States |S| Deterministic TPS of Complete SADF Model Resolve Non-Determinism | S’ | TPS of Complete SADF Model Parallel Composition |S||S| p2p2 S2S2 a p1p1 S1S1 S3S3 S 1, - S2, aS2, a S 3, a p1p1 p2p2 Information on occurrence of actions and time available through reward functions on states only ≤| S | >| S’ | Reduced Discrete Markov Reward Model Remove Irrelevant States |S c | <<|S| Performance Number Compute Equilibrium

Throughput MPEG-4 throughput= average number of frames per second = average number of RC firings per time unit Let {X i | i ≥ 1} denote Markov chain with state space S Define reward c to identify firing completion action of RC If c(s) = 1 for state s, it is relevant, otherwise it is irrelevant 12 Δ is temporal reward function denoting amount of time elapsed since previous RC firing

Metric-Dependent State-Space Reduction If relevant positive recurrent state for ergodic Markov chain exists, then Reduction yields ergodic Markov chain Reduction preserves performance properties 13

Transformed TPS per SADF Process Model Checking Strategy - Practice 14 TPS per SADF Process Discrete Markov Chain Move Transition Labels into States |S| Deterministic TPS of Complete SADF Model Resolve Non-Determinism | S’ | TPS of Complete SADF Model Parallel Composition |S||S| Reduced Discrete Markov Reward Model Remove Irrelevant States |S c | Performance Number Compute Equilibrium Transformed TPS of Complete SADF Model | S’’ | Discrete Markov Chain |S| Parallel composition with on-the-fly reduction and resolving non-determinism

Experimental Results -requires more than 1.5GB of memory ¤takes more than 6 hours Reduction after resolving non-determinism 15 Case Study| S’’ ||S| MPEG-4 AVC Channel Equalizer MPEG-4 SP (PD = 1) MPEG-4 SP (PD = 2) MPEG-4 SP (PD = 3) -- MP3 (PD = 1) -- MP3 (PD = 2) -- MP3 (PD = 3) -- MP3 (4 ≤ PD ≤ 9) -- |S c |Factor ¤- Time [s]Memory [MB] ≤ > 6h¤

Statistical Model Checking as Alternative Statistical model checking supported by modelling SADF in POOSL POOSL is much more expressive than SADF but also has TPS semantics Compositional estimation of confidence intervals for long-run averages 16 Case StudyThroughputTime [s] MPEG-4 AVC ≤ Channel Equalizer · MPEG-4 SP (PD = 1) · MPEG-4 SP (PD = 2) · MPEG-4 SP (PD = 3) · MP3 (PD = 1) · MP3 (PD = 2) · MP3 (PD = 3) · MP3 (PD = 9) ¤> 6h 95% Confidence IntervalTime [s] [ , ]0.14 [ ·10 -3, ·10 -3 ]0.75 [ ·10 -3, ·10 -3 ]7.4 [ ·10 -3, ·10 -3 ]6.86 [ ·10 -3, ·10 -3 ]6.85 [ ·10 -7, ·10 -7 ]32.8 [ ·10 -7, ·10 -7 ]31.5 [ ·10 -7, ·10 -7 ]32.1 [ ·10 -7, ·10 -7 ]30.6

Conclusions & Outlook Performance model checking approach for SADF Compositional TPS semantics with discrete time distributions Exploit semantic properties Removal of metric-dependent irrelevant states On-the-fly construction of relevant state-space Broad variety of pre-defined performance metrics All expressible as temporal reward formula Statistical model-checking for long-run averages as alternative Increase flexibility to allow computing user-defined metrics Lift Markov chain reduction to bisimulation reduction on TPS Support temporal rewards as property specification language Could contemporary quantitative model checkers supporting Probabilistic Timed Automata be a suitable alternative? 17