Resource Efficient Multi-dimensional Cache Management Strategies in Content-Centric Networks
CCN promises to solve many problems related to traditional network architecture with its receiver-driven, secure and simplistic model. Transparent and ubiquitous in-network caching, mobility management are two active research topics within CCN domain, and we intend to address both in this thesis. This document mainly consists of our work on improving in-network cache management in static and dynamic network environments.
In-network cache management in Content-Centric Networking (CCN) has received significant interest from research community in recent years. On the positive aspect, it helps in increasing content availability and quality-of-experience (QoE) by reducing end-to-end delays, and reduces server load. On the other hand, opting the default approach, that is, store a content at every node on the delivery path is not resource-efficient as it introduces high cache redundancy. A multitude of schemes are proposed to increase efficiency of the network, which aim to reduce redundancy by selecting a small fraction of nodes on returning path for storing contents. This selection is generally based on taking into account knowledge about the network or the content itself. However, in many schemes node’s utility is generally determined by analyzing only one concept at a time. For example, some schemes store contents based on popularity, while others select nodes employing only the information about network topology etc. Considering only a single aspect while taking a decision may limit the scope of the cache management scheme to be deployed in diverse scenarios, and might result in sub-optimal performance when the environment is changed. Hence, there is a need to devise an efficient caching mechanism that can dynamically adjust and adapt itself to any environment. With this in mind, we propose an adaptive caching strategy, named as Multi-Attribute Caching Strategy (MACS) based on multi-parameters for CCN. MACS attempts to overcome inefficient cache utilization by intelligently selecting caching locations along the content delivery path. Simulation results show that MACS reduces cache load at each node while providing comparable delay and better cache hit rate using both synthetic and real network topologies.
In second part of the thesis, we analyze the aspects of content-centric caching in Mobile Ad-hoc Networks (MANET). We propose a caching framework that dynamically adapt the caching decision of each content and relocate the replica if the old cached node moves to another location. We envision that such cache relocation mechanism to stimulate cooperation among nodes have potential to help with practical deployments in real scenarios.