Most of research efforts have elaborated on k-anonymity for location privacy. The general architecture for implementing k-anonymity is that there is one trusted server (referred to as location anonymizer) responsible for cloaking at least k users' locations for protecting location privacy. A location anonymizer will generate cloaked regions in which there are at least k users for query processing. Prior works only explore grid shape cloaked regions. However, grid shape cloaked regions result in a considerable amount of query results, thereby increasing the overhead of filtering unwanted query results. In this paper, we propose a cloaking algorithm in which cloaked regions are generated according to the features of spatial networks. By exploring the features of spatial networks, the cloaked regions are very efficient for reducing query results and improving cache utilization of mobile devices. Furthermore, an index structure for spatial networks is built and in light of the proposed index structure, we develop a Spatial-Temporal Connective Cloaking algorithm(abbreviated as STCC). A simulator is implemented and extensive experiments are conducted. Experimental results show that our proposed algorithm outperforms prior cloaking algorithms in terms of the candidate query results and the cache utilization.