A local subspace alignment algorithm for determining the relative locations of sensor nodes in large-scale and low-density wireless sensor networks is presented. Based on the pair-wise distance between each node and its neighbours within a certain communication range, the local geometry of the global structure is obtained by constructing a local subspace for each node, and those subspaces are aligned to give the internal global co-ordinates of all nodes. The localisation performance of this method is compared to the multidimensional scaling and ISOMAP methods.