Mobile edge computation induced caching strategy for huge online education with college teachers and students

Abstract

Via wireless network, mobile edge computation (MEC) provides computation ability to make user feel better. However, caching strategy restricts the performance of MEC. For instance, MEC requires to decide which course should be cached in huge asynchronous online education with college teachers and students. In order to solve the caching problem in asynchronous online education, we put the computation ability to network edge, then make the access points can analysis independently. In access points, we use a reduced support vector regression (rSVR) model to predict the popularity of local data. The popularity is used to improve the hit rate of the caching files. The rSVR model is learnt offline. Different SVR, rSVR is learnt on a small retained subset. Thus, learning rSVR is much faster and the rSVR's model is smaller than SVR's, which is helpful to improve the caching process. A real‐world asynchronous online learning dataset is used to verify effectiveness of our framework. The experimental results show that rSVR predicts popularity for local data is better than previous ones and our caching framework only requires near 50% storage space to maintain the same cache hit rate.