Verification of Delta Form Realization in Fixed-Point Digital - - PowerPoint PPT Presentation

verification of delta form realization in fixed point
SMART_READER_LITE
LIVE PREVIEW

Verification of Delta Form Realization in Fixed-Point Digital - - PowerPoint PPT Presentation

Verification of Delta Form Realization in Fixed-Point Digital Controllers Using Bounded Model Checking Iury V. Bessa , Hussama I. Ibrahim, Lucas C. Cordeiro, and Joo E. C. Filho iurybessa@ufam.edu.br Motivation Controllers DCVerifier


slide-1
SLIDE 1

Verification of Delta Form Realization in Fixed-Point Digital Controllers Using Bounded Model Checking

Iury V. Bessa, Hussama I. Ibrahim, Lucas C. Cordeiro, and João E. C. Filho iurybessa@ufam.edu.br

slide-2
SLIDE 2

Application of a Digital Controller to a Power DC-DC Converter

  • Digital controllers have become pervasive in power eletronics

applications

  • Despite several advantages, they present some limitations for these

applications

Motivation Controllers DCVerifier Evaluation Conclusions

The desired setpoint may not be a representable value due to the quantization effects

SBESC 2014 2

slide-3
SLIDE 3
  • Limit Cycle (LC) oscillations require high effort from engineers
  • Round-off errors in products or overflows in sums may cause oscillations
  • The output voltage might present an undesireble oscillation

Motivation Controllers DCVerifier Evaluation Conclusions

SBESC 2014 3

Limit cycle

  • scillations

The desired setpoint may not be a representable value due to the quantization effects

Application of a Digital Controller to a Power DC-DC Converter

slide-4
SLIDE 4
  • More energy losses and short silicon lifespan
  • LC’s are actually verified trough time-domain simulations

‒ This is an inefficient method since it is time-consuming and not conclusive

Motivation Controllers DCVerifier Evaluation Conclusions

SBESC 2014 4

Limit cycle

  • scillations

The desired setpoint may not be a representable value due to the quantization effects

Application of a Digital Controller to a Power DC-DC Converter

slide-5
SLIDE 5

Bounded Model Check (BMC)

  • Basic Idea: given a transition system M, check negation of a given

property φ up to given depth k

  • Translated into a VC ψ such that: ψ is satisfiable iff φ has

counterexample of max. depth k

  • BMC has been applied successfully to verify (embedded) software since

early 2000’s, but it has not been used to verify digital controllers . . .

M0 M1 M2 Mk-1 ¬ϕ0 ¬ϕ1 ¬ϕ2 ¬ϕk-1 ¬ϕk ∨ ∨ ∨ ∨

Counterexample trace Transition System Property Bound

Mk

SBESC 2014 5

Motivation Controllers DCVerifier Evaluation Conclusions

slide-6
SLIDE 6

Objectives of this work

  • Investigate the FWL effects in fixed-point digital controllers

implementation via a BMC tool

  • Propose a methodology for digital controllers implementation with

the aid of a BMC tool: the DCVerifier

  • Verification engine: ESBMC (Efficient SMT-based Context-

Bounded Model Checker)

  • Check the perfomance of the delta implementations
  • Verify overflows, limit cycles, time constraints, stability, and mimimum

phase in digital controllers

Motivation Controllers DCVerifier Evaluation Conclusions

SBESC 2014 6

Perform BMC of digital controllers implemented in direct and delta forms

slide-7
SLIDE 7

Digital Controllers Implementation Forms

  • Digital controllers implementation forms:

‒ Direct form ‒ Companion form ‒ Jordan form ‒ Diagonal form ‒ Ladder form ‒ Delta form

  • Direct Forms

‒ DFI ‒ DFII ‒ DTFII

SBESC 2014 7

Motivation Controllers DCVerifier Evaluation Conclusions

float controller() { float yn=0; for (int k=0; k<M; k++) { yn += *b++ * *x--; } for (int k=1; k<N; k++) { yn-= *a++ * *y--; } return yn; }

slide-8
SLIDE 8

Some advantages of the delta form

  • Delta forms:

‒ DDFI ‒ DDFII ‒ DDTFII

  • Literature indicates that

there is a close connection between digital delta form and the continuous controller

  • Better numericals

properties

  • Reduced round-off errors

SBESC 2014 8

Motivation Controllers DCVerifier Evaluation Conclusions

Output Time

slide-9
SLIDE 9

Digital Controllers Implementation Aspects

  • Reduced dynamical range
  • Quantization effects (FWL):

‒ Overflows: occurs when a sum or product exceeds the maximum representable value ‒ Limit Cycles: oscillations in output that keep a constant input due to round-offs and overflows ‒ Output errors: the response presents deviations from the expected value

  • Time constraints
  • Coefficients round-off:

‒ Poles and zeros sensitivity: dynamical behavior changes ‒ Stability issue

SBESC 2014 9

Motivation Controllers DCVerifier Evaluation Conclusions

slide-10
SLIDE 10

Digital Controllers Verification Paradigm

  • Common techniques to avoid problems:

‒ Scaling: may prevent overflows, but enhances the output error ‒ Resolution changes (number of bits): boosts the precision, reducing errors and preventing LC ‒ Linear and non-linear compesations: an aditional control loop may rectify the LCs ‒ Non-fragile Control: the deviations of FWL effects are considered in design as uncertains, and the designed controller should be robust to them

  • Digital controllers implementation validation:

‒ Based on simulations and tests ‒ Consume a lot of effort and time ‒ Cannot cover all the possibilities

SBESC 2014 10

Motivation Controllers DCVerifier Evaluation Conclusions

slide-11
SLIDE 11

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation?

Motivation Controllers DCVerifier Evaluation Conclusions

Counterexample

NO YES

SBESC 2014 11

SUCCESS

slide-12
SLIDE 12

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? Counterexample

NO YES

Using prefered tools and methods

SBESC 2014 12

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

SUCCESS

slide-13
SLIDE 13

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? Counterexample

NO YES

Fixed-point format <k,l>, k bits for the integer part and l bits for the fractional part

SBESC 2014 13

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

SUCCESS

slide-14
SLIDE 14

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? Counterexample

NO YES

DFI, DFII, DTFII, DDFI, DDFII, and DDTFII

SBESC 2014 14

Motivation Controllers DCVerifier Evaluation Conclusions

SUCCESS

slide-15
SLIDE 15

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? Counterexample

NO YES

Verification setups: maximum verification time, assertions, test case, and hardware specifications

SBESC 2014 15

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

SUCCESS

slide-16
SLIDE 16

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? Counterexample

NO YES

e.g., ESBMC

SBESC 2014 16

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

SUCCESS

slide-17
SLIDE 17

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? Counterexample

NO YES

SBESC 2014 17

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

SUCCESS

slide-18
SLIDE 18

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? Counterexample

NO YES

SBESC 2014 18

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

SUCCESS

slide-19
SLIDE 19

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? Counterexample

NO YES

Sequence of inputs and states that leads to a failure. May be reproduced in a simulation tool

SBESC 2014 19

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

SUCCESS

slide-20
SLIDE 20

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? Counterexample

NO YES

Re-choose the numeric format and/or realization form

SBESC 2014 20

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

SUCCESS

slide-21
SLIDE 21

Controller Design Define Representation Define Realization Form Configure Verifications Verify Using a BMC tool Property Violation? SUCCESS Counterexample

NO YES

Re-design the controller, in the worst case

SBESC 2014 21

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier: Digital Controllers Implementation with Bounded Model Checking

slide-22
SLIDE 22

SBESC 2014 22

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 15 bits for fractional part Define Realization Form

  • DFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP340 16

MHz Verify using a BMC Tool

  • ESBMC

Result

  • Verification

Failed

slide-23
SLIDE 23

SBESC 2014 23

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 15 bits for fractional part Define Realization Form

  • DFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP340 16

MHz Verify using a BMC Tool

  • ESBMC

Result

  • Verification

Failed

slide-24
SLIDE 24

SBESC 2014 24

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • DFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP340 16

MHz Verify using a BMC Tool

  • ESBMC

Result

  • Verification

Failed

Numeric format choosen based on impulse response sum and in the hardware limitations

slide-25
SLIDE 25

SBESC 2014 25

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • DFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP340 16

MHz Verify using a BMC Tool

  • ESBMC

Result

  • Verification

Failed

Random first trial

slide-26
SLIDE 26

SBESC 2014 26

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • DFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC

Result

  • Verification

Failed

slide-27
SLIDE 27

SBESC 2014 27

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • DFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed

slide-28
SLIDE 28

SBESC 2014 28

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • DFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed Failure due to a sum overflow (sum result = 2.0879 > 1). Input sequence: {0.9995, -0.9995, 0.9995, 1, 1, 1, 0.9995, 0.9995, 0.9995, 0.9995, 1} Redefine the implementation!

slide-29
SLIDE 29

SBESC 2014 29

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • DFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP340 16

MHz Verify using a BMC Tool

  • ESBMC

Result

  • Verification

Failed

Maintain the Representation

slide-30
SLIDE 30

SBESC 2014 30

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • TDFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP340 16

MHz Verify using a BMC Tool

  • ESBMC

Result

  • Verification

Failed

Change the Realization Form TDFII presents less sums and products

slide-31
SLIDE 31

SBESC 2014 31

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • TDFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC

Result

  • Verification

Failed

slide-32
SLIDE 32

SBESC 2014 32

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • TDFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed

slide-33
SLIDE 33

SBESC 2014 33

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • TDFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed

Repeat the test

slide-34
SLIDE 34

SBESC 2014 34

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • TDFII

Configure Verification

  • Verify overflow
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • SUCESS

The problem was solved

slide-35
SLIDE 35

SBESC 2014 35

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 12 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • TDFII

Configure Verification

  • Verify limit

cycle

  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • SUCESS

But verifing limit cycles...

slide-36
SLIDE 36

SBESC 2014 36

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,12 >:3 bits

for integer part and 15 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • TDFII

Configure Verification

  • Verify limit

cycle

  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed Appears an oscillation: {-0.002, -0.002, -0.0015, -0.0015, -0.002, -0.002, -0.0015, -0.0015, -0.002, -0.002}. Zero input sequence Redefine the implementation!

slide-37
SLIDE 37

SBESC 2014 37

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,4 >:3 bits

for integer part and 4 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • TDFII

Configure Verification

  • Verify limit

cycle

  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • SUCESS

Verifing with a different representation... There is a trade off: the oscillation is solved; however there is an accurate loss.

slide-38
SLIDE 38

SBESC 2014 38

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

. .

Define Representa tion

  • < 3,4 >:3 bits

for integer part and 4 bits for fractional part

  • Dynamical

Range: [-1,1] Define Realization Form

  • TDFII

Configure Verification

  • Verify limit

cycle

  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • SUCESS

SUCCESS

slide-39
SLIDE 39

SBESC 2014 39

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

... ..

Define Representati

  • n
  • < 3,4 >:6 bits

for integer part and 7 bits for fractional part

  • Dynamical

Range: [-5,5] Define Realization Form

  • TDFII

Configure Verification

  • Verify stability
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed UNSTABLE

slide-40
SLIDE 40

SBESC 2014 40

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

... ..

Define Representati

  • n
  • < 3,4 >:6 bits

for integer part and 7 bits for fractional part

  • Dynamical

Range: [-5,5] Define Realization Form

  • TDFII

Configure Verification

  • Verify stability
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed UNSTABLE

slide-41
SLIDE 41

SBESC 2014 41

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

... ..

Define Representati

  • n
  • < 3,4 >:6 bits

for integer part and 7 bits for fractional part

  • Dynamical

Range: [-5,5] Define Realization Form

  • TDFII

Configure Verification

  • Verify stability
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed UNSTABLE

slide-42
SLIDE 42

SBESC 2014 42

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

... ..

Define Representati

  • n
  • < 3,4 >:6 bits

for integer part and 7 bits for fractional part

  • Dynamical

Range: [-5,5] Define Realization Form

  • TDFII

Configure Verification

  • Verify stability
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed UNSTABLE

slide-43
SLIDE 43

SBESC 2014 43

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

... ..

Define Representati

  • n
  • < 3,4 >:6 bits

for integer part and 7 bits for fractional part

  • Dynamical

Range: [-5,5] Define Realization Form

  • TDFII

Configure Verification

  • Verify stability
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed UNSTABLE

slide-44
SLIDE 44

SBESC 2014 44

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

... ..

Define Representati

  • n
  • < 3,4 >:6 bits

for integer part and 7 bits for fractional part

  • Dynamical

Range: [-5,5] Define Realization Form

  • TDFII

Configure Verification

  • Verify stability
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed UNSTABLE

slide-45
SLIDE 45

SBESC 2014 45

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example

Controller Design

  • =

... ..

Define Representati

  • n
  • < 3,4 >:6 bits

for integer part and 7 bits for fractional part

  • Dynamical

Range: [-5,5] Define Realization Form

  • TDFII

Configure Verification

  • Verify stability
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • Verification

Failed UNSTABLE

slide-46
SLIDE 46

SBESC 2014 46

Motivation Controllers DCVerifier Evaluation Conclusions

DCVerifier usage example: Delta effect

Controller Design

  • =

... ..

Define Representati

  • n
  • < 3,4 >:6 bits

for integer part and 7 bits for fractional part

  • Dynamical

Range: [-5,5] Define Realization Form

  • TDFII

Configure Verification

  • Verify stability
  • Verification

time:3600

  • MSP430 16-bit

16 MHz Verify using a BMC Tool

  • ESBMC
  • Bound: 10

samples

Result

  • SUCESS

STABLE

slide-47
SLIDE 47

Experimental Objectives

  • Use BMC tools to verify digital controllers
  • Find potential bugs before the deployment
  • Evaluate the proposed methodology, in particular the DCVerifier
  • Verify overflows, limit cycles, time constraints, and stability
  • Compare the delta operator and the direct-form performance from the

verification point of view

SBESC 2014 47

Motivation Controllers DCVerifier Evaluation Conclusions

slide-48
SLIDE 48

Experiments Setup

  • Verification Enviroment

‒ Intel Core i7-2600 3.40 GHz processor, 24 GB of RAM, and Ubuntu 11.10 64- bits ‒ ESBMC v1.23 with the SMT solver Z3 v4.0

  • Hardware Considerations

‒ Verifications based on MSP340, 16 MHz clock ‒ Wordlength: 16 bits

SBESC 2014 48

Motivation Controllers DCVerifier Evaluation Conclusions

slide-49
SLIDE 49

Experimental Results

SBESC 2014 49

Motivation Controllers DCVerifier Evaluation Conclusions

The verification results are conclusive in almost 95%

  • f

the testcases

slide-50
SLIDE 50

Experimental Results

SBESC 2014 50

Motivation Controllers DCVerifier Evaluation Conclusions

Direct realization presented 40%

  • f

errors in

  • properties. Delta

decreased it to 27.5%

slide-51
SLIDE 51

Conclusions

  • BMC is a promising alternative for digital controllers verification
  • The verifications are conclusive in 94.5% of the benchmarks
  • Neither false positives nor false negatives are reported
  • The DCVerifier may reduce the design efforts

‒ Since it is automatic and reliable

  • Delta form doesn’t remove all errors, but it decreases them substantially.
  • Future work

‒ Include more properties and realization forms ‒ Include closed-loop properties verification ‒ Create an automatic design tool

Motivation Controllers DCVerifier Evaluation Conclusions

SBESC 2014 51

slide-52
SLIDE 52

Thank you for your attention!

The tool and all benchmarks are avaliable at www.esbmc.org

SBESC 2014 52