With the growth of location based services, indoor localization is attracting great interests as it facilitates further ubiquitous environments. In this paper, we propose FuseLoc, the first information fusion based indoor localization using multiple features extracted from Channel State Information (CSI). In FuseLoc, the localization problem is modelled as a pattern matching problem, where the location of a subject is predicted based on the similarity measure of the CSI features of the unknown location with those of the training locations. The system exploits both the amplitude and phase information of CSI over multiple antennas from Orthogonal Frequency Division Multiplexing (OFDM) system for localization. Specifically, FuseLoc implements a discriminative feature extraction from measured CSI for pattern matching, where an effective feature fusion is performed using Canonical Correlation Analysis (CCA) by maximizing the pairwise correlations across the feature sets. Finally a similarity measure is performed to find the best match to localize a subject. Experimental results show that FuseLoc can estimate location with high accuracy which outperforms other state-of-the-art approaches.