Multidimensional Content Modeling and Caching in D2D Edge Networks

Future Internet is going to be shaped by networked multimedia services with exploding video traffic becoming the dominant payload. That evolution requires a remedial shift from the connection-oriented architecture to a content-centric one. Another technique to address this capacity crunch is to improve spectral utilization through new networking paradigms at the wireless network edge. To this end, Device-to-Device (D2D) communications has the potential for boosting the capacity and energy efficiency for content-centric networking. To design and implement efficient content-centric D2D networks, rigorous content modeling and in-network caching mechanisms based on such models are crucial. In this work, we develop a multidimensional content model based on popularity, chunking and layering, and devise caching schemes through this model. Our main motivation is to improve the system performance via our caching strategies. The numerical analysis shows the interplay among different system parameters and performance metrics: while our schemes perform slightly poorer in terms of system goodput, they also decrease the system energy expenditure. Overall, this improvement dominates the loss in the goodput, leading to greater energy efficiency compared to the commonly-used caching technique Least Recently Used (LRU).