This paper proposes an iterative optimal sensor placement (OSP) framework for structural identification and model updating of structural systems using a small number of mobile sensors. The model updating is performed through a Bayesian inference approach which is solved through asymptotic approximation for computational efficiency. In an iterative manner, the OSP is performed to minimize the information entropy in estimating the updating parameters of the model. Each OSP iteration is performed to find the next location of mobile sensors, where the prior probability distribution of updating parameters is assumed as the posterior probability distribution obtained from the previous iteration. This process is repeated until the uncertainties of updating parameters fall below a predetermined threshold. A forward sequential sensor placement algorithm is used to solve the OSP problem at each iteration. This algorithm provides a nearly-optimal solution and is much more efficient compared to an exhaustive search. This proposed framework is applied to a numerical case study, namely the Dowling Hall Footbridge located at Tufts University campus. Updating parameters used in this study are the added mass at different segments of the bridge deck. The purpose of using added mass is to create a realistic pseudo damage on a specific portion of the bridge. The proposed iterative system identification approach is applied for estimation of updating parameters considering different number of available sensors. This study shows that the iterative OSP approach using a small number of mobile sensors placed iteratively provides better model updating results compared to the case of using an optimal static sensor configuration involving larger number of sensors.