Computer Engineering is coming soon

We're still writing and reviewing the notes for every subject, and we'll open it when every subject is ready. Meanwhile you can practise aptitude, ask Aria about any subject, or see the branches that are ready.

Quality attributes of embedded systems

Quality attributes of embedded systems describe non‑functional properties such as reliability, availability, safety, performance, power and maintainability that an engineer must design, measure and trade off.

Drafted with Aria, reviewed by the AiCanCode.org team · Updated . How we write and check our notes. Spotted an error? Use Give Feedback at the bottom of the page.

Why it matters

Embedded engineers use quality attributes to decide design trade‑offs (for example, how much redundancy, how large a battery, or which RTOS scheduling policy) so the final product meets required behaviour in the field. Good understanding of these attributes is necessary for system design, testing, acceptance and certification.

Key ideas

  • Diagram to draw (exam): draw three component boxes in series (label A, B, C) and note “series reliability = product of component reliabilities” under the diagram.

  • Availability — fraction of time the system is functioning correctly (sometimes called operational availability). For the simple repair model with mean time between failures MTBF and mean time to repair MTTR, availability is exactly A = MTBF / (MTBF + MTTR). Here MTBF denotes mean uptime between failures (i.e., MTTF); if your convention defines MTBF = MTTF + MTTR, use A = MTTF / (MTTF + MTTR) instead. Assumptions: alternating cycles of up and down time, steady state, MTBF and MTTR well defined.

  • Safety — freedom from unacceptable risk of physical injury or damage. In engineering practice safety is managed by hazard analysis, fail‑safe design and, for some systems, following standards. Safety is a separate attribute from reliability: a reliable system can still be unsafe if failures cause hazards.

  • Security — protection of the system and its data against malicious actions. Security attributes include confidentiality, integrity and availability (the CIA triad). Security often trades off with performance and cost.

  • Real‑time behaviour — concerns timeliness and determinism. Define hard real‑time as systems where missing a deadline is catastrophic; soft real‑time where late results degrade performance but are acceptable. Deadline, latency (time between event and response) and jitter (variation in latency) are key terms.

  • Performance — includes latency (seconds, milliseconds, microseconds), throughput (items per second), and CPU utilisation. Use Amdahl's law to reason about acceleration limits when part of the system is sped up or parallelised.

  • Power and energy — power is the rate of energy use (watts), energy is power integrated over time (joules). For battery‑powered systems, energy efficiency and average power determine battery life. Power trade‑offs often affect thermal design and reliability.

  • Maintainability — ease of performing repairs and updates (time to diagnose, replace, or reprogram components). Closely linked to testability and modularity.

  • Testability — ability to observe internal behaviour and detect faults using tests. A design with built‑in test (BIST) or self‑diagnostics increases testability and may improve availability.

  • Redundancy — duplication of components to improve reliability or availability. Two basic configurations: series (system fails if any component fails) and parallel (redundant) (system fails only if all redundant components fail). Exact formulas below.

When deriving reliability or availability, always list assumptions: independence of component failures (if you use product rules), constant failure rate for exponential models, and negligible common‑cause failures unless explicitly modelled.

Formulas

R(t) = P(T > t) — Reliability at time t; T is time‑to‑failure; unitless probability; applies generally.

R(t) = e^{-λt} — Reliability for exponential failure distribution; λ is failure rate (s^{-1}); t in seconds; assumes constant λ and memoryless failures.

MTTF = 1/λ — Mean time to failure (s) for exponential distribution; λ in s^{-1}. For repairable systems with exponential failure and repair, MTBF = MTTF + MTTR.

A = MTBF / (MTBF + MTTR) — Availability (unitless fraction, often expressed %); MTBF and MTTR in same time units; steady‑state alternating up/down assumption. Here MTBF denotes mean uptime between failures (i.e., MTTF); if your convention defines MTBF = MTTF + MTTR, use A = MTTF / (MTTF + MTTR).

R_series = ∏_{i=1..n} R_i — Reliability of n components in series; each R_i is component reliability at the mission time; unitless; assumes independent failures.

R_parallel = 1 - ∏_{i=1..n} (1 - R_i) — Reliability of n parallel (redundant) components; unitless; assumes independent failures.

P = V · I — Electrical power (W = volt × ampere); V in volts, I in amperes; applies to DC circuits or instantaneous values.

E = P · t — Energy (J = W × s); P in watts, t in seconds.

t_battery (hours) = Capacity (Ah) / I_load (A) — Battery life in hours using nominal capacity (assumes constant current draw and full usable capacity).

Speedup = 1 / ((1 - P) + P / S) — Amdahl's law for parallelisation; P is parallelisable fraction (0..1), S is speedup of the parallel part; unitless. Assumes fixed problem size and no overhead for parallelisation.

Worked examples

  1. Availability from MTBF/MTTR

    • Given: MTBF = 5000 h, MTTR = 2 h.
    • Formula: A = MTBF / (MTBF + MTTR).
    • Numbers: A = 5000 / (5000 + 2) = 5000 / 5002 = 0.99960016.
    • Convert to percent: 0.99960016 × 100 = 99.960016%.
    • Answer: Availability = 99.960% (approx).
  2. Battery life and energy

    • Given: Li‑ion cell capacity 2000 mAh at 3.7 V. Device draws 50 mA at 3.7 V continuously.
    • Convert capacity: 2000 mAh = 2.0 Ah.
    • Battery life (hours): t = Capacity (Ah) / I_load (A) = 2.0 / 0.05 = 40 h.
    • Energy stored: E = V × Capacity (Ah) = 3.7 V × 2.0 Ah = 7.4 Wh = 7.4 × 3600 = 26640 J.
    • Answer: Approx 40 hours of operation; stored energy 7.4 Wh (26640 J).
    • Notes: Real battery life is lower because of cut‑off voltage, discharge rate, temperature and converter losses.
  3. Reliability of small systems (series and parallel)

    • Series: Two components with reliabilities R1=0.995 and R2=0.997 for the mission time.
      • R_series = R1 × R2 = 0.995 × 0.997 = 0.992015.
      • Answer: Series system reliability = 0.992015 (99.2015%).
    • Parallel (redundant sensors): Three identical sensors each R=0.98 (probability each survives mission).
      • R_parallel = 1 - (1 - 0.98)^3 = 1 - (0.02)^3 = 1 - 8e-6 = 0.999992.
      • Answer: Parallel system reliability ≈ 0.999992 (99.9992%).
    • Comment: These assume independent failures and no common‑mode faults.

Common mistakes

  • Treating reliability and availability as the same: reliability is probability of no failure in a time interval; availability includes repair and is a time fraction.
  • Using Amdahl's law without accounting for parallel overhead or communication cost; reporting unrealistic speedups.
  • Overlooking common‑cause failures when claiming benefit from redundancy (e.g., same firmware bug in all redundant units).
  • Ignoring battery derating factors (temperature, age, cut‑off voltage) when calculating life.

For your exam

Kinds of questions you will see:

  • Definitions: give precise definitions for reliability, availability, safety, latency, jitter. A full‑marks answer gives the definition, units, and a short example.
  • Short notes: explain differences between reliability and availability, or between hard and soft real‑time. Include practical design implications (e.g., redundancy, watchdogs, RMA and MTTR reduction). (RMA = Rate Monotonic Analysis.)
  • Derivations: derive R(t) = e^{-λt} from constant failure rate assumption, or derive availability formula from MTBF/MTTR. Full marks need step‑by‑step assumptions and a final formula.
  • Numericals: compute availability, battery life, or system reliability for series/parallel components. Show unit conversions and steps; state assumptions.
  • Diagrams: draw series and parallel reliability diagrams or a simple block showing MTBF and MTTR cycles. Label components and annotate formulas used.

What a full‑marks answer includes:

  • Start by defining terms used.
  • State assumptions explicitly (e.g., independent failures, constant λ).
  • Show each algebraic step and include units for numerical answers.
  • Mention practical caveats (common‑mode failures, temperature effects, measurement limits).

Quick check

  1. What is the reliability R(t) for an exponential failure process with failure rate λ = 2×10^{-6} s^{-1} after 1e6 seconds?
  2. Convert 2500 mAh battery to ampere‑hours and calculate life if the device draws 125 mA.
  3. Two identical components in parallel each have reliability 0.90. What is the parallel system reliability?

Answers: R(1e6) = e^{-2e-6 × 1e6} = e^{-2} ≈ 0.1353; battery 2500 mAh = 2.5 Ah so life = 2.5 / 0.125 = 20 h; parallel reliability = 1 - (1 - 0.9)^2 = 1 - 0.01 = 0.99.

Finished this topic? Mark it so your progress, study plan and readiness keep up.

Stuck on something here?