Caching em Redes Tolerantes a Atrasos com Dados Nomeados

Abstract

Neste projeto de investigação, abordar-se-á o desempenho de vários algoritmos de caching em Redes Tolerantes a Atrasos com Dados Nomeados. Os algoritmos de caching são fundamentais para o armazenamento e distribuição dos conteúdos nos nós. É utilizado como ponto de partida os algoritmos de caching das Redes Tolerantes a Atrasos e Redes de Dados Nomeados, e igualmente nas redes IP tradicionais e web proxies. Com base nesses conhecimentos, serão introduzidas alterações no sentido de os adaptar ao contexto proposto dos mecanismos de Redes de Dados Nomeados em Redes Tolerantes a Atrasos. Em termos experimentais será usado o ambiente de simulação ICONE (Information Centric Opportunistic Networking Environment) desenvolvida com base no ONE (Opportunistic Networking Environment), um simulador para Redes Tolerantes a Atrasos. O ICONE introduz extensões de suporte a Redes de Dados Nomeados no ONE. Os algoritmos serão avaliados pela sua probabilidade de interesses satisfeitos, latência de interesses satisfeitos, interesses encaminhados, sobrecarga da rede, entre outros. Serão usados parâmetros como espaço de armazenamento para a cache, tempo de simulação, ritmo de geração de interesses entre outros. São também importantes aspetos como a segurança, confiabilidade e escalabilidade dados.

The importance of communication in our actual civilization causes a constant search for solutions able to solve existing problems. Areas where communications are intermittent, constitute one of these problems. Through a Delay Tolerant Network it is possible to communicate more efficiently and reliably. Their ability to data caching can store the contents on all nodes through which they travel. This enables better resilience to failures and delays. The ability to run other technologies on NDN, makes possible the coexistence with the Named Data Networking. These were developed in order to replace the traditional IP networks. NDN claims that an architecture based on host-to-host communications like the current Internet, it is not suitable for network user patterns. Instead, NDN use names for each existing content on the network and allows interest generation on a particular content that the network will send back. Each node is capable of data caching and this allows it to respond with the corresponding content, when a cache hit occurs. Due to the good coexistence between DTNs and NDNs, there are several network architectures and joining these two technologies. They use the data caching as a way to increase performance and network resiliency. For this reason, the study of caching is a very important issue. The goal of this thesis is the study of several algorithms and mechanisms for caching on Named Data Delay Tolerant Networks. Then, using the ICONE testbed, these algorithms were simulated and evaluted using several metrics namely percentages of satisfied interests, cache hits and computational efforts to complete the process. Results show significant improvements in the percentage of satisfied interest when using caching algorithms, in particular with the Network Interest algorithm. Results improve with the usage of repository components (Repo) architectures.