Which statement correctly contrasts Musa-Okumoto and Jelinski-Moranda software reliability growth models?

• A. Musa-Okumoto is a parametric model; Jelinski-Moranda is non-parametric.
• B. Musa-Okumoto uses a linear mean value function.
• C. Musa-Okumoto is deterministic; Jelinski-Moranda is stochastic.
• D. Musa-Okumoto is non-parametric with m(t) = a ln(1 + b t); Jelinski-Moranda assumes remaining latent defects control the failure rate: λ(t) = φ (N0 − m(t)).

Answer: (D)

The key idea is how each model links the observed failures to the underlying fault process. Musa-Okumoto treats failures as a stochastic process with a mean value function that grows in a logarithmic way over time: m(t) = a ln(1 + b t). This means the expected number of failures accumulates, but the rate of new failures declines over time because the derivative dm/dt = a b /(1 + b t) decreases as t increases. Jelinski-Moranda, in contrast, ties the instantaneous failure rate directly to the number of latent defects still present, so λ(t) = φ (N0 − m(t)). As you fix faults and m(t) grows, the remaining latent defects shrink, causing the hazard to drop accordingly; this produces a different trajectory for the cumulative failures (often m(t) approaches N0 as time goes on).

So option describing Musa-Okumoto with m(t) = a ln(1 + b t) and Jelinski-Moranda with λ(t) = φ (N0 − m(t)) correctly contrasts the two models. The Musa-Okumoto form is the log-based mean function, and Jelinski-Moranda uses the remaining latent defects to govern the failure rate.