Abstract: To solve the problem of drastically changing network fluctuation and long tail, an adaptive bit rate algorithm model based on deep learning is proposed. The algorithm optimizes video bit rate, bit rate switching frequency, and video pause time, and it is robust to the randomness of network fluctuations. Compared with other models, the proposed model has better performance in the scenario of drastic fluctuations in wireless networks. Under the worst network conditions, the comparison model causes video playback lag with a probability as high as 16% , while the proposed model has a lag probability of only 1% , and the average quality of experience index is 30% higher than that of the comparative model.

Key words: reinforcement learning, mobile wireless network, adaptive bit rate