Large deviations for sample paths of Gaussian processes quadratic variations

We show a functional large deviations principle for the family of random functions
$$ \left\{V_n(x)=\sum_{k=1}^{[nx]}(Z_{k/n}-Z_{k-1/n})^2,\ x\in[0,1]\right\}, $$
where $\{Z_t,\,t\in[0,1]\}$ is a real valued centered Gaussian process.