摘要
情境感知系统通常以情境信息直接驱动上层服务或应用。普适计算环境的情境信息在来源及种类上存在较大差异,情境信息的多样性导致由情境直接驱动的情境感知系统的扩展性、稳定性不足。对此提出一种场景驱动的情境感知计算框架。该框架以场景为基础,屏蔽了原始情境信息的异构性和多样性;以场景识别为核心,由系统中的基本情境信息识别当前用户的场景信息,由场景信息驱动相关的应用。该框架简化了系统设计的复杂度,有助于提高系统的扩展性和稳定性;同时场景识别采用神经网络的算法,避免了因采用推理所带来的知识库暴涨问题。所开发的原型系统验证了框架的有效性。
Context-aware systems generally use contexts to drive higher applications. Because context information is diverse in types and values, the systems developed are poor in extensibility and stability. This paper proposed a situation-driven model for context-awareness computing. The framework is based on situation information, and it can shield the heterogeneity and diversity of original context information. Situation-recognition is the kernel of this framework. Thesituation is recognized from basic context information,and is used to drive related applications. This framework simplifies the processing on designing a system, and is helpful in improving the systems＇ extensibility and stability. The mech-anism of situation recognition based on neural network avoids the problem of knowledge explosion when it uses the mechanism of reasoning. A prototype system was implemented, and it validated the proposed infrastructure.
出处
《计算机科学》
CSCD
北大核心
2012年第3期216-221,共6页
Computer Science
基金
国家自然科学基金项目(60803044
60903125)
国家863基金项目(2009AA011903)
教育部"新世纪优秀人才支持计划"(NCET-09-0079)资助
关键词
情境感知
场景识别
人工神经网络
普适计算
Context-awareness, Situation recognition, Artificial neural network, Pervasive computing
