Ensuring Software Quality

= Degree to which a system satisfies stakeholder needs (quality model = captures functionality, performance, security, maintainability, etc.)

1. Performance (response time, throughput)

2. Reliability (POFOD, MTTF, MTTBC)

3. Security (MTTSF = time until a security failure)

“software system where the correct functioning of the system depends on the results produced by the system AND the time at which these results are produced”

= Systems that monitor and control their environment

Usually embedded systems with sensors and actuators

-> Must respond within specified time!!

Logical correctness = correct result

Temporal correctness = correct timing

-> Both must be met for overall correctness!!!

software system where the correct functioning of the system depends on the results produced by the system AND the time at which these results are produced

1. Soft real-time: degraded if results are not produced according to the specified timing constraints

2. Hard real-time: incorrect if results are not produced according to the timing specification

1. Periodic: predictable intervals (e.g., 10x/sec)

2. Aperiodic: unpredictable (e.g., power failure)

-> A powerful processor may replace interrupts with polling

1. Interrupts: control is automatically transferred to a pre-determined memory location

   -> must be be short, simple and fast

2. Periodic: have period, execution time, and deadline. Managed by real-time clock -> Shorter deadline process gets priority

Monitoring Systems

Control Systems

Data Acquisition Systems

1. Monitoring: act on exceptions

2. Control: act continuously via actuators

-> Both poll sensors for environmental data

1. Collect sensor data for analysis

2. Different collection vs. processing deadlines S

1. Sensor triggers system processing

2. Response sent to actuator

Control loop = same physical parameter

1. Timing varies per sensor-actuator pair

2. Fast switching between handlers required

-> Sequential loops inadequate

1. Safety = absence of danger

2. Reliability = failure-free operation (often through redundancy)

Fault → Error → Failure

• Fault = defect (may lie dormant)

• Error = incorrect internal state

• Failure = externally visible misbehavior

Not every fault causes failure

Systematic = design/build mistakes

Random = wear or damage

• Transient = disappears (retry works)

• Persistent = stays until intervention

  
  

Real-Time (Design) Patterns = RT-pattern is a solution to recurring software design problems in the area of real-time systems

• Focus here: safety and reliability patterns

1. Data flows through sequential filters

2. Parallelism improves performance & safety

3. Redundant channels improve reliability

-> Also known as Pipe & Filter pattern

Two identical channels

Switch to backup on fault

-> Protection against random faults, not assuming a fail-safe state

3 identical channels

Majority vote determines output

No validation needed

-> Protection against random faults without a failsafe state

Same function, different implementations

Protects against random AND systematic faults

No fail-safe state required

-> Protection a ainst random and systematic fa lts

One channel with error detection

Requires fail-safe state

Handles transient faults (e.g., retry) -> Inexpensive safety improvement without redundancy

Monitoring channel supervises actuator

Requires independent actuator sensor

Detects random & systematic faults -> Needs fail-safe state

Lightweight form of Monitor-Actuator

Only approximate validation

Needs fail-safe state -> Simpler, less expensive, less precise

Detects time-based faults or deadlocks

Heartbeat mechanism monitors activity

No actuator supervision -> Lightweight, fail-safe state required

For complex systems

Coordinates watchdogs, policies, checks

Centralized fail-safe logic -> Handles complicated failure scenarios

“Conscious reflection and evaluation of the potential consequences of using the developed software”

Ethos = Ethical Awareness

• “Mindset shaped by moral values”

1. Software impacts grow with digitization

2. Some areas (e.g., weapons) demand higher awareness

3. Emerging fields raise new ethical questions -> SEs must evaluate consequences of software use

Limits of Regulation

• Software often precedes regulation

• Laws are vague, slow, and hard to enforce

• Surveillance would be required for enforcement

-> Regulation helps, but cannot fully solve the problem

Because laws are not in place and regulation will never be a complete solution:

We must see it as the responsibility of the individual

Critically evaluate the consequences of one’s own actions