| Summary: | 碩士 === 國立臺灣科技大學 === 資訊工程系 === 99 === Data dissemination is an important application in DTNs. There are a large demand for data dissemination applications, such as advertisement and news publishing in urban areas or military communication in battle fields. Data-driven communication often rely on publish-subscribe paradigm which can deliver specific data to interested subscribers.
In this thesis, we study the problem of data dissemination with time-constrained through a publish-subscribe protocol in a DTN. Few researches have addressed the issue of time-constrained data dissemination in a DTN. Although a DTN is delay tolerant by definition, we consider the case in which there is an upper bound on allowable delivery latency for a publication. Our goal is to deliver as much as possible publications before they expired. However, there are few limitations for mobile nodes in a DTN, such as limited storage, limited bandwidth, and limited battery. The problem is how we choose an appropriate set of publication for exchange.
We proposed a replication–based DTN protocol for publish-subscribe which can dynamically increase or decrease the number of replica of a publication in system based on its importance. The importance is determined by the utility function. We proposed four different utility functions to rank the publications. Each utility functions is designed with different consideration in mind, such as time urgency, popularity, and fairness. We evaluate the effect of choosing different utility functions on the performance of a time-constrained publish-subscribe system in a DTN.
|
|---|
