Abstract: How to better portray the composition of innovation network, so that firms can improve the innovation capability better and quickly through resource integration, is a key problem in capability catch-up process. We explore six traditional cases of Chinese firms to investigate the relationship between in-house knowledge base and the innovation network composition and its effect on innovation performance. We address 4 kinds of knowledge base depending on the dimension of knowledge breadth and depth. What’s more, we demonstrate the matching configuration of different knowledge base types with different innovation network composition depending on the dimension of knowledge heterogeneity and knowledge quality and empirically decompose the process and logics to improve innovation performance.