Design And Implementation Of Ubiquitous Search System With Sparse Sensing Data

In recent years,with the frequent occurrence of missing persons and missing items,the target search and tracking in a large range has become a hot research topic.The traditional single point positioning methods,such as GPS and WiFi,gradually decline after the arising of "ubiquitous network",it means to achieve seamless integrate in informative and physical space,and it enables people and people,people and things,things and things to communicate smoothly.The search system based on"ubiquitous network" named "ubiquitous search" system holds on the idea of the collaboration between intelligent mobile terminals and wearable devices,aiming to search for lost people as a specific application scenario.With low power Bluetooth wearable devices combined with multiple mobile terminals,a high positioning precision,energy saving and easy to use participatory sensory searching system is built by our work.In this paper,the building of the "ubiquitous search"system is firstly introduced,in the aspect of the design of system architecture,this paper gives full consideration to user privacy,equipment energy efficiency and other factors,and then proposes a layered system structure suitable for participatory perception service platformIn the "ubiquitous search" participatory sensing system,the design of an optimized participant selection algorithm is the key to ensure both the the quality of the sensing data and high positioning accuracy.However,the sparsity of sensing data is the intrinsic data characteristic of"ubiquitous search" system.Therefore,a set of participant selection algorithm in ubiquitous search scenarios combined with localization computation is proposed in this paper.Different from the traditional algorithms,the algorithm considers not only the participants bidding amount,reputation value,energy consumption and other factors,but also subtly take the importance of the position of participants as the key factor for evaluation,moreover the mathematical model and evaluation method for the importance of the position is also given in this paper.In addition,the algorithm needs to advance on sparse sensing data using a data recovery method,a method based on Compressed Sensing is proposed to recovery the RSSI missing data,which ensures the quality of the positioning calculation data.Finally,this work was validated with experiments on two open access datasets in both indoor,outdoor and a mixed environment.Results demonstrate that our proposed mechanism outperforms the other participant selection algorithms both in quality of uploaded sensing data and positioning accuracy.