Routing and Name Resolution in Information-Centric Networks

Abstract

Information-centric networking (ICN) has recently attracted research attention and several architecture designs have been proposed. However there is lack of theoretical foundations and quantitative models to evaluate different design choices and compare their respective advantages. In this paper, we take an initial step to build a quantitative framework to enable characterization and comparison of two popular classes of ICN name-oriented routing and resolution techniques, flooding and distributed hash table (DHT). Our results indicate that the interaction of several factors such as network size, content location dynamics and content popularity contributes to the performance of the routing and resolution mechanisms. Especially we obtain a quantitative expression to determine under what conditions one approach yields superior performance over the other. It also suggests that a hybrid DHT-flooding mechanism may perform better although the actual design of such a hybrid protocol is out of the scope of this paper. Furthermore, we analyze the impact of variable-length names and obtain an expression for the optimal name length assigned to a content object. Finally, we present another application of our model, determining when the name aggregation in content location publishing can reduce control overhead and improve scalability. Our modeling and analysis results reveal valuable insights regarding design tradeoffs and provide design guidelines.