In mobility-assisted, opportunistic networks, data is disseminated in a store-and-forward manner by means of spontaneously connecting mobile devices. Therefore, mobility itself moves in the center of investigation. Knowledge about movement characteristics of single devices can be used to add realism to random mobility models and to understand the likelihood of communication options. This paper contributes to the field of observing movement characteristics of single devices for opportunistic networks by describing movement features and investigating how these features can contribute to human movement activity estimation. Activity descriptions are useful for characterizing the purpose of movement. Additionally, in case movement patterns are uncertain or fragmentary, knowledge about activities may help to estimate average movement characteristics faster. We use activity estimation based on the Naive Bayes classifier applied to a multi-variate feature set consisting of commonly considered movement features. We investigate the classification success rate experimentally when using all features and when using only a subset of features. Therefore, we conducted a user study collecting real-trip GPS traces labeled by the users. We selected four most frequent urban movement use case activities for classification and achieved a success rate of 80.65%.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below