In this paper, the deployment of wireless network is analyzed in accordance with the manifold distribution. Next, the previous Locality Preserving Canonical Correlation Analysis (LPCCA) algorithm is used to build a mapping from the signal-vector space to the physical location space and develop a location algorithm, Location Estimation-Locality Preserving Canonical Correlation Analysis (LE-LPCCA, for short), which sufficiently takes the local structure of the network topology into account. Finally, experimental results on benchmark show that this method achieves a higher accuracy and stability than other location algorithms.