Content

1. Overview
2. Predictive Models
   1. Example: Waterfall with Feedback
   2. Example: Sashimi Waterfall
   3. Example: Incremental Waterfall
3. Adaptive Models
   1. Example: Rapid Prototyping
   2. Example: Unified Process

Overview

• Basis of a software engineering project
  • Requirements engineering
  • Design
  • Implementation
  • Testing
  • Deployment
  • Maintenance
• Software process model: the way these activities are organized
  • Waterfall (predictive model): activity starts after the previous one finishes
  • Unified Process (adaptive model): activity conducted sequentially in a series of iterations

Predictive Models

Assumes requirements & solution clear at the start of project.

• Advantage
  • Knows exactly when the activity starts/ends
  • Customers knows what will be delivered without surprise -> mission-critical and safety-critical products
  • No waste of time & effort prototyping & experimenting
  • Less refactoring
• Disadvantage
  • If requirements wrong, it is disastrous
  • Doesn't work if customer's and developer's visions of product are different, or requirements incomplete
  • No product delivered until the end
• When will work
  • Requirements really known
  • No significant risk
  • Experienced team
  • Schedule is long enough to complete

Example: Waterfall with Feedback

• Doesn't assume each activity completed perfectly
• Working S/W still appears late
• Risks are not resolved early
• In practice, moving back is hard

Example: Sashimi Waterfall

• Allows overlapping between activities
• Allows some knowledge to be discovered in later activities to be used in earlier activities

Example: Incremental Waterfall

• Allows some degree of adaptation
• Each increment runs predictively
• Each increment long

Adaptive Models

Accepts requirements & solution cannot be known for certain, and changes are extremely likely.

• Advantages
  • Less detailed up-front planning and commitment
  • Can drive iterations by risk -> early risk mitigation
  • Better engagement with user
  • Can release at any time
  • Handles change & complexity better
• Disadvantage
  • Complexity
  • Not suitable for small, low-risk project

Example: Rapid Prototyping

Example: Unified Process