For the renewal process whose interarrival times are uniformly distributed over (0,1), determine the expected time from t = 1 until the next renewal.

I know that $\lim_{t \to\infty}\frac{S_{N(t)+1}}{N(t)}=\mu $ and for the uniform distribution, $\mu=\frac{1}{2}$. I know I want to find $E[S_{N(1)+1}]$. I'm just now sure how to put these ideas together. Or if I'm even headed the right direction.


$$S_{N(t)+1} = \sum_{i=1}^{N(t)+1}X_i$$

where $X_i \sim \text{Uniform}(0,1)$ are iid. Since $N(t)+1$ is a stopping time and $\mathbb{E}[N(t)+1] = m(t)+ 1$ where $m(t) = \mathbb{E}[N(t)]$, if $m(t) < \infty$, we can use Wald's equation to conclude that,

$$\mathbb{E}[S_{N(t)+1}] = \mathbb{E}[X_1]\mathbb{E}[N(t)+1] = \frac{1}{2}(m(t)+1)$$

The only thing left is to compute $m(t)$ and simultaneously show that $m(t) < \infty$. It is easy to show that,

$$m(t) = \sum_{n=1}^{\infty}F_{n}(t)$$

where $F_{n}(t)$ is the $n$ convolutions of the cdf of uniform random variable. It follows from above,

$$m(t) = F(t) + (F*\sum_{k=1}^{\infty}F_{k})(t) = F(t) + (F*m)(t) = F(t) + \int_{0}^{t}m(t-x)dF(x)$$

Now, use $F(t)=t$ and $dF(x) = 1\cdot dx$ to show that,

$$m(t) = t + \int_{0}^{t}m(t-x)dx = t+\int_{0}^{t}m(x)dx \implies \frac{dm(t)}{dt} = 1+m(t) \implies m(t) = e^{t}-1$$

Therefore, $m(t) < \infty$ when $t < \infty$ and $\mathbb{E}[S_{N(t)+1}] = e^t/2$.

P.S. Please comment if something is not clear.

  • $\begingroup$ Can you please explain the step in which $m(t-x)$ is changed to $m(x)$ (in the integration part). $\endgroup$ – Harry Jul 3 at 11:31
  • $\begingroup$ Use change of variables $u \leftarrow t-x$ and $du = -dx$. $\endgroup$ – Dhruv Kohli - expiTTp1z0 Jul 3 at 15:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.