Given a compound distribution $$S:=\sum_{k=1}^{N} X_{i}$$ with
- $N$ is a discrete random variable with values in $\mathbb{N}$ with finite mean and variance.
- $X_{k}$ are non-negative iid random variables such that $\mathbb{E}\left[X_{i}\right]<\infty$ and $\sigma^{2}$:=Var($X_{i}$)<$\infty$.
- $N$ and $(X_{1},X_{2},\ldots)$ are independent.
Then $$\mathbb{E}\left[ S\right]= \mathbb{E}\left[ N\right]\mathbb{E}\left[ X_{1}\right]$$ as well as $$\text{Var}\left[ S\right]= \text{Var}\left[ N\right]\mathbb{E}\left[ X_{1}\right]^{2} + \mathbb{E}\left[ N\right]\text{Var}\left[ X_{1}\right].$$ Is there now a form of the central limit theorem applicable? In particular, are there conditions under which we can use normal approximation, i.e.
$$\frac{S-\mathbb{E}\left[S \right]}{\sqrt{\text{Var}\left[S \right]}}\approx \mathcal{N}(0,1)\; ?$$
$\mathbf{Update}$: Thanks a lot for the reference. Suppose that $\mathbb{E}\left[ X_{i}\right]=0$ and $N(t), t \geq 0$ is a family of positive, integer valued random variables, such that there is a $\theta >0$ $$ \frac{N(t)}{t}\stackrel{\mathbb{P}}{\rightarrow}\theta,\,\text{ as } t\to\infty. $$
According to Renyi's or Anscombe's Theorem, we then have \begin{align*} \frac{S_{N(t)}}{\sigma\sqrt{N(t)}} \xrightarrow[]{d} \mathcal{N}(0,1)\,\text{ as } t\to\infty \\ \frac{S_{N(t)}}{\sigma\sqrt{\theta\cdot t}} \xrightarrow[]{d} \mathcal{N}(0,1)\,\text{ as } t\to\infty, \\ \end{align*} which is different from the above normal approximation (using Wald's identity).
My question is now:
Under which assumptions is the first/above normal approximation valid?
What is the key difference between the two approximations respectively which one is preferrable? For example if $N(t)$ is a Poisson distribution?