Framework and Software Reliability Dolores R. Wallace SRS Technologies Software Assurance Technology Center http:

2539 days ago, 983 views
PowerPoint PPT Presentation
Framework and Programming Dependability Dolores R. Wallace SRS Innovations Programming Certification Innovation Center William H. Farr, Dr. John R. Crigler Maritime Surface Fighting Center Dahlgren Division. NASA OSMA SAS '03. Review of the Issue.

Presentation Transcript

Slide 1

Framework and Software Reliability Dolores R. Wallace SRS Technologies Software Assurance Technology Center William H. Farr, Dr. John R. Crigler Naval Surface Warfare Center Dahlgren Division NASA OSMA SAS '03 SAS 03/GSFC/SATC-NSWC-DD

Slide 2

Overview of the Problem Reliability Measurement is a basic target for NASA frameworks Systems are surveyed from the product/equipment/frameworks point of view Methodologies for equipment unwavering quality appraisal have been created and used in the course of recent decades Methodologies for programming dependability evaluation have been produced since the 70's and have been used in the course of the most recent a quarter century for framework dependability appraisal have just been tended to throughout the most recent 10 years with little application encounter Need for a device that coordinates all parts of dependability information (programming, equipment, and frameworks viewpoints) SAS 03/GSFC/SATC-NSWC-DD

Slide 3

Project Objectives Enhance the ability for NASA to survey programming dependability by recognizing and consolidating late models into the apparatus S tatistical M odeling and E stimation of R eliability F unctions for S ystems ( SMERFS^3 ) First Year Initiative Perform a definite writing look (1990 and past) Enhance the capacity for NASA to survey framework unwavering quality by refreshing SMERFS^3 Second Year Initiative Identify framework models for joining Apply the distinguished strategies to venture informational collections inside the NASA/DoD conditions SAS 03/GSFC/SATC-NSWC-DD

Slide 4

FY03 Research Plan Literature seek Selection of new models Build new models into SMERFS^3 Test new models with Goddard extend information Make most recent rendition of SMERFS^3 accessible SAS 03/GSFC/SATC-NSWC-DD

Slide 5

Articles from 1990 forward Journals - test IEEE TSE IEEE Reliability Software Testing, Verification, and Reliability IEEE Software IEEE Computer Conferences ISSRE ICSE Reliability & Maintainability High-Assurance Systems Eng. Different others Model determination criteria Model suppositions Fit inside ebb and flow SMERFS^3 Type of framework Data accessibility Domain Experts Literature Search SAS 03/GSFC/SATC-NSWC-DD

Slide 6

Software Real-time Large-scale Time-basic Embedded Maybe overwhelming COTS Distributed System Safety-basic parts Heterogeneous Fault tolerant Costly to grow Long lifetime, developmental Characteristics of the Software Based Systems SAS 03/GSFC/SATC-NSWC-DD

Slide 7

SMERFS^3 Current Version highlights: 6 programming unwavering quality models 2D, 3D plots of info information, fit into each model Various dependability gauges User inquiries for expectations Updates limitations: Employ information from joining, framework test, or operational stage Use existing representation of SMERFS^3 Integrate with existing UIs, decency of-fit tests, and forecast abilities SAS 03/GSFC/SATC-NSWC-DD

Slide 8

Available Data Large GSFC extend, however privacy required GSFC individual precious in clarifying the framework and the information Several subsystems Data level records – much exertion into spreadsheet/database Operational disappointments just Remove particular blames and sort others Apply IntervalCounter Bottom line: arranging information required considerable exertion – limited if extend individual arranged the information SAS 03/GSFC/SATC-NSWC-DD

Slide 9

Identified Models Hypergeometric Schneidewind (upgrades) Log-strategic Extended Execution Time (EET) The initial two models require mistake tally disappointment information; the last two require time-between-disappointment information Only blunder include information has been caught the GSFC extend database accessible for investigation Hence, programming unwavering quality increments to SMERFS^3 in this errand will be restricted to the hypergeometric demonstrate and the measurements improvements to the Schneidewind display SAS 03/GSFC/SATC-NSWC-DD

Slide 10

Hypergeometric Model Assumptions Test example , t(i): A gathering of info test information. N : Total number of starting shortcomings in the product. Issues identified by a test occasion are evacuated before the following test case is practiced No new blame is embedded into the product in the expulsion of the distinguished blame. A test occasion t(i) faculties w(i) introductory shortcomings. w(i) may change with the state of test examples over i. It is now and then alluded to in the creators' papers as a "sensitivity" figure. The underlying deficiencies really detected by t(i) rely on t(i) itself. The w(i) beginning shortcomings are taken haphazardly from the N introductory flaws. SAS 03/GSFC/SATC-NSWC-DD

Slide 11

Hypergeometric Model Meets a significant number of our choice criteria: Data sort Fits inside the system of the SMERFS^3 programming Research demonstrates that it seems to perform well against different models Allows for testing force consider (for instance: number of experiments, number of testing work force, troubleshoot time ) Scheduled for usage in the last quarter of FY03 SAS 03/GSFC/SATC-NSWC-DD

Slide 12

Schneidewind Model There are three adaptations: Model 1 : All of the blame means each testing period are dealt with the same. Demonstrate 2 : Ignore the principal s-1 testing periods and their related blame numbers. Just utilize the information from s to n. Show 3 : Combine the blame numbers of the interims 1 to s-1 into the primary information point. Along these lines there are s+1 information focuses. SAS 03/GSFC/SATC-NSWC-DD

Slide 13

Schneidewind Assumptions The quantity of deficiencies identified in each of the particular interims are autonomous. The blame redress rate is relative to the quantity of shortcomings to be remedied. The interims over which the product is tried are altogether taken to be of a similar length. The combined number of issues by time t, M(t), takes after a Poisson procedure with mean esteem work μ(t). The mean esteem capacity is to such an extent that the normal number of blame events for whenever period is relative to the normal number of undetected issues around then. The disappointment power work, λ(t), is thought to be an exponentially diminishing capacity of time; that is, λ(t)=αexp(- βt) for some α, β > 0. SAS 03/GSFC/SATC-NSWC-DD

Slide 14

Schneidewind Model Enhancements Meets a significant number of our choice criteria: Data sort Basic model as of now in the SMERFS^3 programming It has been appeared to perform well against different models Allows expectation to absorb information impact Updates are being executed this quarter Risk measures Operational quality at time t Risk basis metric for the rest of the issues at time t Risk foundation metric for the opportunity to next disappointment at time t Confidence interims SAS 03/GSFC/SATC-NSWC-DD

Slide 16

Data Analysis of NASA Three Month Fault Counts

Slide 17

Proposed Next Steps FY03 – Focused on programming Complete usage and testing Prepare paper depicting the examination and model determination, usage, conclusions Apply the improvements on the Goddard informational index Prepare SMERFS^3 for conveyance FY04 Conduct comparable research exertion for System Reliability University of Connecticut will take part Enhance and approve framework models SAS 03/GSFC/SATC-NSWC-DD