Let $\{B_i\}$ be iid Bernoulli$(\frac{1}{2})$ random variables and consider the number in $[0,1]$ whose binary expansion corresponds to the infinite sequence $\{B_i\}_{i=1}^{\infty}$, i.e. the number \begin{align*} \sum_{n=1}^{\infty} 2^{-n} B_n. \end{align*} Let $X_n$ be the $n^{th}$ partial sum, i.e. \begin{align*} \sum_{k=1}^{n} 2^{-k}B_k. \end{align*} Show that $X_n$ converges in distribution to a continuous Uniform$[0,1]$ random variable.
To show convergence in distribution, we can show that the CDF's of the $\{X_n\}$, which we can denote $\{F_n\}$, converge to the CDF of Uniform$[0,1]$, which is $F(x) = x$.
I tried to compute $F_n(x)$, which is \begin{align*} P(X_n \leq x) &= P(\sum_{k=1}^n 2^{-k}B_k \leq x) \\ &= P(\sum_{k=1}^n 2^{n-k} B_k \leq 2^n x) \end{align*} I don't know how to simplify this expression.
Another approach I took was to compute the characteristic function of $X_n$ and show that it converges to the characteristic function of the uniform$[0,1]$ random variable. After some calculations, this amounted to showing that \begin{align*} 2^{-n} \prod_{k=1}^{n} (1 + e^{2^{-k}it}) \to \frac{e^{it} - 1}{it} \end{align*} for each $t \in \mathbb{R}$. Again I am stuck here. Can anyone find a slick way to prove this intuitive result?