Abstract: A cache strategy based on community nodes division and content popularity（CCNCP） was proposed to improve content availability due to the limit capacity of the routers in named-data networking（NDN）. The regions were divided according to the topology structure, and the importance of nodes was comprehensively measured to select local central nodes and global central nodes. The contents with different popularities were reasonably cached in nodes with different importance. The classic GN algorithm was used to obtain partition communities. The centrality indexes of degree centrality, closeness centrality and betweenness centrality were defined. The system model was established, and the cache strategy was provided. To evaluate the performance of the scheme, the Icarus was used to simulate CCNCP and other alternative strategies. The simulation experiment was implemented on two topologies of Zachary karate club network and GARR of real topology. The results show that CCNCP performs better in both topologies. When the parameter S is equal to 0.25, the cache hit ratio, latency and path stretch ratio of CCNCP and the second best cache strategy ProbCache are 30.9% and 26.8%, 52.9 and 56.2 ms, 0.584 and 0.617, respectively, and the former is 4.1% higher, 5.9% lower and 5.3% lower than the latter, which indicates that the proposed CCNCP has significant salient performance over the current typical strategies in all three metrics.