Sunteți pe pagina 1din 23

Formal Reliability Analysis of

Generalized Multi-state k-out-of-n Systems

Shahid Ali Murtza


(MS Student)

Proposal Defense
June, 2017
SEECS-NUST
Outline

 Introduction

 MS k-out-n :G System

 Reliability Evaluation

 Example

 Conclusion

Osman
S.A. Hasan
Murtza Progress Presentation
Multi-state NESCOM Project
System Reliability 22
10
Reliability

 The ability of a system (or component) to


perform its required functions for a specified
time

 Reliability=ℙ(no failure before a certain time)

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 32
2
10
Safety Critical Systems

More stringent reliability requirements


ISO 8402
BS 4778

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 52
2
10
Reliability Models

 Binary Model
System and components assume only two crisp
states {complete failure, perfect functioning}

Multi-state(MS) Model
 Various distinguished levels of performance or
states for system and its components
 Multiple partial failure states

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 82
2
10
Comparison

Criteria Binary Models Multistate Models

Modeling Simple to model complex to model

Realistic less realistic, thus not precise more realistic, thus more
precise
Reliability easy to evaluate complex to evaluate
Evaluation

Multistate models are preferred when accuracy and precision are required
as is the case with safety critical systems

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 92
2
10
Multistate Models

 Multistate models are of two types

 G Models :
 F Models : Failure of

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 15
2
10
2
Generalized MS k-out-of-n :G SYSTEM

– The value of k can be different for different


system states.
– Increasing MS k-out-of-n : G system
• (𝑘1 ≤ 𝑘2 ≤ ⋯ ≤ 𝑘𝑀 )
– Decreasing MS k-out-of-n : G system
• (𝑘1 ≥ 𝑘2 ≥ ⋯ ≥ 𝑘𝑀 )
– Constant MS k-out-of-n : G system
• (𝑘1 = 𝑘2 = ⋯ = 𝑘𝑀 = 𝑘)
– Mix/Generic MS k-out-of-n : G system

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 16
2
10
2
Example An Increasing MS -out-of- :G System

Given: A 3-component system with k1 = 1, k 2 = 2; k 3 = 3; Both


the system and the components can be in 1 of 4 possible states: 0,
1, 2, 3.
Because k1 < k 2 < k 3  Increasing MS -out-of- :G system.
• The system state is #3 if all 3 components are in state 3.
• The system state is #2 or above if at least 2 components
are in state 2 or above.
• The system state is #1 or above if at least 1 component is
in state 1 or above.
The system in this example has a
• 3-out-of-3:G series structure at system state #3,
• 2-out-of-3:G structure at system state #2,
• 1-out-of-3:G parallel structure at system state #1.

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 24
2
10
2
Example Solution

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 25
2
10
2
Analysis Tools

 Paper-and-Pencil Proof Methods

 Simulation

 Formal Methods

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 26
2
10
2
Paper-and-Pencil Proof Methods

 Construct a reliability model of the given system on paper


 Analytically analyze the overall reliability of the given system on
paper
 Already verified model relationships

 Error Prone
 Manual manipulation and simplification
 Missing assumptions

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 27
2
10
2
Computer Simulations

 Generate samples to model the reliabilities of the sub-modules


 from the Exponential and Weibull random variables
 Compute the overall reliability of the given system based on the
already verified model relationships

 Error Prone
 Pseudo random Numbers
 Computer arithmetic
 Numerical techniques

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 28
2
10
2
Inaccuracies in Reliability analysis

 A severe limitation in the case of safety-critical applications

like oil and gas pipelines

 May endanger human and animal life

 Lead to a significant financial loss

Osman
S.A. Hasan
Murtza Progress
Formal Presentation
Multi-state
Verification NESCOM
System Reliability
of Verilog Project
Code 29
2
10
2
Formal Methods

 Reliability models have been analyzed extensively using the

following formal techniques:

 Petri Nets

 Model Checking

 Higher-order-Logic Theorem Proving

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 30
2
10
2
Model Checking

Model
(System Requirements)
True, if Model
satisfies
Model
the given
Checker
specifications
M |= ɸ
Otherwise a
counterexample
Specification
(System Properties)

Model Checking has been used for binary system reliability analysis
(Hermanns et al.DSN-2013; Pend et al. Modeling Symp. 2013; Pervez et al.
e-HEALTHCOMM-2014)
Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 31
2
10
2
Theorem Proving

System Properties

Theorem Proving has been used


Logical Model
for binary system reliability
in HOL
analysis
(Hasan et al. TC-2010;
Hasan et al. JAL-2011;
Proof Goal Abbasi et al. WoLLIC-2014;
Ahmad et al. CICM-2014;
Ahmad et al. WiMob-2015;
Ahmad et al. IWIL-2015;
Proof Assistant
Ahmad et al. CICM-2015;
(HOL4,
Ahmad et al. CICM-2016; and
Isabelle/HOL)
Other)

Mechanized Proofs of System Properties

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 32
2
10
2
Analysis Techniques

Criteria Paper n Pencil Computer Theorem Model


Proof Simulation Proving Checking
Expressiveness    ?

Accuracy ?   

Automation   ? 

Model B/M B/M B B

 The precision of results is very important while analyzing


safety-critical domains
 Theorem Proving can guarantee more precise reliability
analysis

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 33
2
10
2
Problem Statement

 Binary systems have been analyzed accurately which are


less practical
 Multistate systems are more flexible and more precise
systems but they have not been analyzed so accurately that
they can be relied upon in many safety critical systems

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 33
2
10
2
Methodology

System Description Higher Order Logic


Probability
Formal
Structure Function
Model Reliability
Assign State Formal State Multistate Systems
Distributions Distributions
Recursive Algos
Reliability Proof
Requirements Goal

Theorem Prover

Formally Verified
Proof Done Reliability
Properties

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 34
2
10
2
Timeline
Main Tasks Time
Literature Review & Problem Statement May, 2017
Proposal Defense June, 2017
Formalization of MS Systems in HOL July, 2017

Formalization of MS Reliability Algorithm August, 2017

Verification of MS Reliability Algorithm September, 2017

Case Studies October, 2017


Thesis & Paper Write-up November, 2017

Final Defense December, 2017

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 35
2
10
2
Conclusion

– A MS system model is a more flexible tool


– Generalized MS k-out-of-n systems cover many types
of system structures
– Formal analysis (Theorem Proving) can give more
accurate reliability evaluation

Osman
S.A. Hasan
Murtza ProgressVerification
Presentation
Multi-state
Formal NESCOM
System Reliability
of Verilog Project
Code 36
2
10
2
Slide 1

Thanks!

S-ar putea să vă placă și