A D2D caching strategy based on file segmentation and user leaving trend probability
Xu Guang Huang, Hui Song, Daru Pan, Xian Zhou · 2021
In the present study on Device-to-Device (D2D) cache, the files in the file library are cached in the users’ space and a complete way. However, they all disregard the fact that most users do not watch the video in full, especially in the traffic accounting network, it is more obvious. Hence, this undoubtedly causes waste of storage space, and then segmenting video files can solve this problem well. Most existing articles about video file partitioning rely on helpers which the system deliberately selects to assist in caching file blocks, but here we increase the overall hit probability of the system by changing the cache probabilities of different blocks instead of helpers. In this paper, we propose a new caching strategy based on user leaving probability and video file library segmentation, and compare it with other traditional caching algorithms. Simulation and analytical results show that the proposed method HA is feasible, and are capable of offering better performance than other benchmarks in terms of hitting probability.