Big data and semantics management system for computer networks

摘要: We define Big Networks as those that generate big data and can benefit from management in their operations. Examples of networks include the current Internet emerging things social networks. The ever-increasing scale, complexity heterogeneity make it harder to discover emergent anomalous behavior network traffic. hypothesize endowing otherwise semantically-oblivious with memory mimicking human functionalities would help advance capability learn, conceptualize effectively efficiently store traffic behavior, more accurately predict future events. Inspired by memory, we proposed a distributed system, termed NetMem, extract utilize semantics matching prediction processes. In particular, explore Hidden Markov Models (HMM), Latent Dirichlet Allocation (LDA), simple statistical analysis-based techniques for semantic reasoning NetMem. Additionally, propose hybrid intelligence technique integrating LDA HMM based on learning patterns features syntax dependencies. also locality sensitive hashing reducing dimensionality. Our simulation study using real demonstrates benefits NetMem highlights advantages limitations aforementioned techniques.