The invention belongs to the field of mobile edge computing, and provides a streaming media cache distribution device based on a dynamic programming algorithm and a multi-cell user mobility and a working method thereof, and the device can collect context perception information of an edge server, and the context perception information comprises a movement track of a user in a cell, a request mode of a terminal, a current network load and a distribution state of server resources. According to the method, a space-time locality principle is considered to carry out abstract analysis on a multi-celloverlapping switching probability of a user, and a popularity cache model and a motion sensing prefetching model are established. On the premise of meeting the constraints of storage, bandwidth and computing resources, an optimization target of maximizing cache benefits is established. An optimal cache placement and replacement decision is made for the server through dynamic programming, and theoverall cache benefit is improved from three aspects of popularity statistics, mobility perception and cost management. According to the device and the method, the average access delay of the terminaland the congestion cost of resources can be effectively reduced, and the cache performance of an edge computing system is improved.