THEORY EXAMINATION (SEM–VI) 2016-17 SOFTWARE RELIABILITY

SOFTWARE RELIABILITY (EIT063)

Section-wise Solved Answers & Notes

SECTION – A (10 × 2 = 20 Marks)

Very short & direct answers

(a) Software vs Hardware Reliability

(b) Defect rate and reliability

• Defect rate: Number of defects per unit size (KLOC / FP).
• Reliability: Probability that software performs failure-free for a specified time.

(c) Tools of software reliability techniques

• Failure data analysis tools                                                  • Curve-fitting tools
• Reliability modeling tools (SARA, CASRE)                          • Statistical analysis tools

(d) Two matrices to define software reliability

• Defect density                                                                  • Mean Time To Failure (MTTF)

(e) Curve fitting in software reliability

Curve fitting is the process of matching collected failure data with a mathematical reliability model to predict future failures.

(f) Defect, fault and failure                                     • Defect: Error in code or design
• Fault: Incorrect internal state                              • Failure: Incorrect external behavior

(g) Accuracy of estimating project schedule

Accuracy is generally low in early phases and improves as project progresses; early estimates may vary by ±25–50%.

(h) Software reliability model

A mathematical model that predicts software failure behavior over time based on historical failure data.

(i) Parametric reliability growth model

A model that assumes specific parameters to describe failure intensity reduction over time (e.g., exponential, logarithmic models).

(j) Need of documents and matrices

They help in:
• Tracking defects                                                      • Measuring quality
• Predicting reliability                                                • Improving decision-making

SECTION – B (Attempt Any Five) (5 × 10 = 50 Marks)

(a) Software metrics for analysis and design models

Metrics used:
• Size metrics: LOC, Function Points                        • Complexity metrics: Cyclomatic complexity
• Design metrics: Coupling, cohesion

They help evaluate maintainability, reliability, and testability early.

(b) Software Quality Assessment Models

These models evaluate quality using metrics and attributes.

Examples:                                                                   • McCall’s Quality Model
• Boehm’s Quality Model

(Block diagram: Requirements → Metrics → Analysis → Quality rating)

(c) Size & structure in measuring product attributes

• Size: LOC, FP → indicates effort & defects
• Structure: Coupling, cohesion → indicates maintainability

Product quality is ascertained using defect density, reliability, and maintainability metrics.

(d) Evolution of SQA & major issues

Evolution:                         • Testing-oriented → Process-oriented → Quality management

Major SQA issues:
• Incomplete requirements                     • Poor testing
• Schedule pressure                                • Inadequate metrics

(e) Predicting reliability techniques (example)

Techniques include:                                • Failure rate modeling
• Reliability growth models                    • Statistical estimation

Example: Predicting remaining failures using exponential model after test phase.

(f) Hierarchical model of software quality assessment

Quality factors → Quality criteria → Metrics

Example:   Reliability → Failure rate → MTTF

This structure provides clear traceability from metrics to quality goals.

(g) Zero defect software & reliability attributes

Zero defect software:           Aim to prevent defects rather than fix later.

Reliability attributes:
• Failure rate                             • MTTF
• Availability                              • Recoverability

(h) Fault & failure data collection & phase-based defect removal

• Defects are collected phase-wise (requirements, design, code, testing).
• Early removal reduces cost.

Pattern:
Maximum defects introduced in coding, ideally removed in reviews/testing.

SECTION – C (Attempt Any Two) (2 × 15 = 30 Marks)

Q3 (a) Logarithmic Poisson Execution Time Model

Failure intensity:

λ(μ)=λ0e−θμ\lambda(\mu) = \lambda_0 e^{-\theta \mu}λ(μ)=λ0​e−θμ

Given:
Decay parameter θ = 0.25/failure

Resources required include:
• Testing time                                          • Skilled testers
• Debugging effort                                  • Tools and infrastructure

As failures are removed, failure intensity decreases logarithmically.

Q3 (b) Error seeding, failure rate & curve fitting

• Error seeding: Artificial defects inserted to estimate remaining defects
• Failure rate: Failures per unit time
• Curve fitting: Matching real failure data with reliability model

Q4 (a) Software reliability technique (block diagram)

Process:
Requirements → Design → Coding → Testing → Failure data collection → Reliability modeling → Prediction

Purpose: Estimate and improve reliability before release.

Q4 (b) Static code metric

Static metrics measure code without execution.

Examples:                                              • Cyclomatic complexity
• Lines of code                                      • Comment density

Used to predict error-prone modules.

Q5 Short Notes (Any Three)

(a) Rayleigh model

Used when failure rate increases initially, then decreases; suitable for large projects.

(b) SARA tool

SARA (Software Automated Reliability Analysis) supports:
• Failure data analysis                                  • Reliability prediction
• Model fitting

(c) Metrics for software maintenance

• MTTR                                                         • Change request rate
• Defect backlog

(d) Major SQA activities

• Reviews & audits                                       • Testing
• Metrics collection                                      • Process improvement