PROJECT TITLE :

Content-Aware Proactive Caching for Backhaul Offloading in Cellular Network - 2018

ABSTRACT:

Proactive caching is taken into account a value-effective methodology to address the backhaul bottleneck downside in cellular network. In this Project, we tend to propose a novel popularity predicting- caching procedure that takes raw video data as an input to see an optimal cache placement policy, that deals with both revealed and unpublished videos. To anticipate the popularity of unpublished videos of which the statistical info isn't offered, we tend to apply the content-primarily based approach by extracting and condensing video options into a high-dimensional vector. Subsequently, we have a tendency to kind G clusters of options representing the potential video classes (VCs) and map the feature vector into a G-dimensional space, where every element indicates the proportion to which the video contains the features of the corresponding VC. Finally, we train a prediction model to foresee the popularity, where the set of published videos is used as training information. Last, the prediction with expert advice methodology is employed to update the coaching set, and to gain insight into how the predictor output will deviate from the best skilled prediction, we have a tendency to address the concept of expected cumulative loss and derive the analytical expression for its upper certain. Intensive simulation results are shown to gain insight into our proposed system subject to different factors, like network size, cache capacity, and user's preference profile. In summary, we show that applying intelligence-based mostly content-aware proactive caching is an economical approach to significantly improving the operation of cellular networks in the longer term.