Context Recognition Framework On Smart Mobile Device

With rapid development of Information technology, the physical world becomemuch intelligent. In the network of the physical world, a large amount of digitaldevices are running for difference purposes and offer a wide variety of contextualinformation. How to effectively manage these contextual information and use theseinformation to improve the user experience has attracted more and more attention.Context recognition technique is proposed to solve the problem. Contextrecognition system solves the problem of top layer application by using contextualinformation and context inferring which solves the contradiction between the slowgrowth of the lower layer which provides contextual information and the rapidgrowth of the demand of upper layer software. Context-recognition system makescontext aware program focus on the upper design to reduce the workload of softwaredevelopment; It can also be employed as middle-ware used in various systems toimprove effectiveness and user experience of old systems.In this work, we research the design of context recognition framework on smartmobile devices and context inference algorithm, the main work is done as follows;(1) Studied the context recognition frameworks proposed by other researchersearlier and analyzed their advantages and disadvantages;(2) Designed a context recognition framework based on smart mobile device,according to the characteristics of smart mobile devices and the popular contextrecognition frameworks;(3) Personally collected data sets;4500position records which marked andcollected20-minute information on each type of activity;(4) Proposed scenario recognition and activity recognition which mainlyinclude context source selection, data preprocessing, feature extraction from datasets and use classification algorithms on data sets. And verify the real effect of thealgorithm and evaluate the results;(5) Completed building context recognition system based on contextrecognition framework.Main contribution of the paper: completed context recognition system based onthe android platform, employed Google Places API to achieve location-basedscenario recognition and proposed how to use direction data to normalize theacceleration data in activity recognition, the experiment results show encouraging.