The invention relates to a video pre-fetching and cache replacement decision-making method for mobile perception in mobile edge calculation, designs a mobile perception utility function, and providesa novel trade-off method between a mobile edge data pre-fetching and cache replacement strategy. And not only is the allocation of edge server cache resources among different videos optimized, but also storage resources are allocated for different code rates of the same video. The invention proves that the proposed cache replacement problem is a knapsack constrained NP (NP-Hard) problem. In orderto solve the problem, the invention provides a mobile perception video prefetching cache and replacement (MAVPCR) algorithm, which is a dynamic programming (DP) algorithm. According to the invention,the trade-off of data prefetching and cache replacement strategies can be better realized. Experiments prove that the optimization problem is converted into integer linear programming with knapsack constraints, the hit rate can be increased, the cost is effectively controlled, and meanwhile, the video distortion rate is reduced.