Abstract
Crowdsensing is a popular paradigm that outsources sensory/multimedia tasks to mobile workers. In the crowdsensing systems, workers perform diverse tasks such as detecting sensory data and taking pictures by employing (using) their smartphones, which are equipped with sensing and multimedia functions. We provide a crowdsensing system to efficiently delegate sensory/multimedia tasks to mobile workers, and we focus on spatial-temporal tasks that must be conducted at specific locations and time. Each worker supplies his/her destination with a deadline to our system and does not mind taking detour paths to maximize profits. Once workers submit their profiles to our system, they will receive detour paths, which consist of tasks in particular orders. Workers execute tasks by following their detour paths and receive maximal profits. We formulate this problem as a detour planning problem, and the advanced problem is multi-users detour planning problem. The difference between these two problems is that the detour planning problem just considers a worker at a time. In this thesis, we develop a detour planning algorithm (DP) and a multi-users detour planning algorithm (MDP) to solve problems respectively. We simulate the extensive trace-driven scenarios and demonstrate the effectiveness and efficiency of our algorithms. Developing a working prototype on Android OS and addressing other challenging aspects of the considered systems are our future tasks.