The application of Internet-of-vehicles technologies in automobiles drives the modern automobiles developing towards electronization, networking and integration. The rapid development of Internet-of-vehicle technologies causes the problem of rapidly increasing data volume and constrained bandwidth for in-vehicle CAN (controller area network) networks. To solve these problems, this research addresses the signal packing problem in the design of CAN network. First, the signal set is divided into signal clusters according to signals' period. Next, the clusters are sorted in order of increasing period. Then, combined with two presented bandwidth slack evaluation metrics, a signal clustering-based signal packing (CSP) algorithm is proposed to realize the optimization of bandwidth utilization. Finally, by comparing with the state-of-art algorithms, the optimism of CSP in improving the bandwidth utilization is verified. Comparing with the algorithms proposed in reference 7, 8 and 11, the obtained average and maximal optimization ratio in bandwidth utilization for CSP is between[0.5%, 6.4%] and[2.4%, 22.65%], respectively.