Distributed expectation-based spatio-temporal cluster detection for pocket switched networks

Distributed cluster detection has been widely applied to the problem of efficient data delivery in Pocket Switched Networks (PSNs). However, previously proposed methods can become less efficient over long periods of time as more and more transient, pair-wise connections are used to form clusters which monotonically increase in size. Monotonically increasing clusters, also known as aggregated clusters, suffer an inevitable loss of temporal information. Thus cluster relevance to individual movement patterns at particular times is diminished. This paper applies expectation-based spatio-temporal clustering to the problem of data delivery in PSNs. We will show that spatio-temporal clusters can be detected distributively and can improve the data delivery efficiency by 7% and deliver 6% more packets compared to aggregated methods. Furthermore, 3 data delivery strategies which utilise spatio-temporal clusters are proposed.