It has been common knowledge that the assumption of constant volatility in the Black-Scholes model is notsupported by the actual data. Various alternative models have instead been proposed, including stochastic ones. One popular model is due to Hull and White, which extends the Black-Scholes model by letting the volatility itself be modelled as a diffusion process. One immediate question that arises is: can we estimate the volatility (preferably online) as we gather the stock price data?. The purpose of this talk is attempting to answer this estimation problem in two dimensional setting using nonlinear filtering technique.