Research On Task Crowdsourcing Based Group Construction In Participatory Sensing System

Participatory sensing which is appeared in recent years, called urban sensing as well, is a people-centric sensing perceptual technology. Participatory sensing can use many existing data collection equipments to collect information on the status of the city, including images, sounds, and other types of sensor datas. Participatory sensing endows mobile phones with more functions. Not just as a communication media among individuals, smartphones can also be used as tools of interactive or autonomous data collection, data classification and transporting all kinds of state information. Then using these data make a smart decision and provide service for human life and social activities. Recently some scholars have proposed in participatory sensing systems adopting the way of crowdsourcing to obtain perceptual data or complete the complex tasks. Crowdsourcing in participatory sensing systems task is done jointly by a group of participatory sensing nodes, so in order to support participatory sensing crowdsourcing applications with high efficiency and high quality. It is a very key scientific question to create a group composed of appropriate participatory sensing nodes dynamicly.According to the above problems, this thesis firstly analyzes the application of participatory sensing systems based on task crowdsourcing and relevant features of the model. Then a crowdsourcing model formalization description method is put forward. This thesis is on the basis of study on ant clustering algorithm, and then Combined with ant clustering algorithm, propose two kinds of group construction algorithms. One is based on negative correlation factors adjusting, and the other is based on the optimal selecting solution algorithm. The formation of the group are discussed under different constraint conditions using the two algorithms.The research results show that the task model of crowdsourcing in the participatory sensing system can be effectively dealt with the crowdsourcing applications. The model can do with a lot of crowdsourcing services preprocessing and task decomposition. At the same time, in view of the ant clustering we begin data difference expansion algorithm research, which expanding the use of the ant clustering algorithm. The data difference expansion algorithm can be effectively dealt with grouping problem based on the importance weight assignation. The group construction algorithms based on crowdsourcing task have good problems solving skills and good extensibility. To sum up, the studies in this thesis has important theoretical significance and practical significance in carrying out the crowdsourcing applications in participatory sensing system in the future.