We study the problem of estimating time-varying occupancy and ambient air flow signals using noisy carbon dioxide and flow sensor measurements. A regularized moving horizon estimation formulation is proposed that constrains time-varying signals to smooth Fourier expansions. We demonstrate that the regularization approach makes the estimator robust to high levels of noise. In addition, it requires minimal information about the shape of the signals. Computational experiments with simulated and real data demonstrate the effectiveness of the approach. (C) 2013 Elsevier Ltd. All rights reserved.