CAJ | 학술논문

信息物理系统CPS（Cyber-Physic System）是融合了传感器网络、通信系统、计算机网络等多种网络和技术的新型复杂系统。其结构复杂性决定了网络中会产生大量数据，如何减小网络负载，提高能量利用率是CPS面临的一项重要挑战。数据融合通过将信息综合化，减少在网络内传输的数据量，能够有效地降低网络负载，从而达到节省能量开支的效果。同时通过将各类原始数据融合成具有意义的概括数据，大大提高了网络的数据分析和决策能力。本文将数据融合技术应用于信息物理系统，基于数据的时空和属性特点，提出了同时包含时间融合、空间融合、属性融合的STAC(Spatial-Temporal-Attribute Correlation)算法。该算法基于时空相关性对原始数据进行首次融合，进而在加权贝叶斯理论的基础上，根据不同类型数据的属性相关性对数据进行二次融合，形成有综合意义的概括数据。基于NS2平台的仿真结果证明，该算法能够有效减少数据包传输量，并且能够准确预测CPS网络事件类型，实现CPS的智能闭环控制。
신식물리계통CPS（Cyber-Physic System）시융합료전감기망락、통신계통、계산궤망락등다충망락화기술적신형복잡계통。기결구복잡성결정료망락중회산생대량수거，여하감소망락부재，제고능량이용솔시CPS면림적일항중요도전。수거융합통과장신식종합화，감소재망락내전수적수거량，능구유효지강저망락부재，종이체도절성능량개지적효과。동시통과장각류원시수거융합성구유의의적개괄수거，대대제고료망락적수거분석화결책능력。본문장수거융합기술응용우신식물리계통，기우수거적시공화속성특점，제출료동시포함시간융합、공간융합、속성융합적STAC(Spatial-Temporal-Attribute Correlation)산법。해산법기우시공상관성대원시수거진행수차융합，진이재가권패협사이론적기출상，근거불동류형수거적속성상관성대수거진행이차융합，형성유종합의의적개괄수거。기우NS2평태적방진결과증명，해산법능구유효감소수거포전수량，병차능구준학예측CPS망락사건류형，실현CPS적지능폐배공제。
Cyber-Physic System is a new type of complex system which combines sensor network, communication system, computer network etc. The complicated structure of CPS will result in producing large amounts of data, so how to reduce the network load and to improve the energy utilization is a challenge. Data aggregation is a promising technique which can reduce the amount of data transmitted in network by converting the raw data to summary data, leading to the purpose of saving energy. Meanwhile, since the aggregated data contains more comprehensive information, it can greatly improve the analysis and decision-making ability in the network. In this paper, we ifrst applied data aggregation technology to the context of CPS, and proposed a new algorithm called STAC (Spatial-Temporal-Attribute Correlation) algorithm, combining spatial aggregation, temporal aggregation, and the most important attribute aggregation. STAC turn the large amounts of raw data into GI (general information) based on the spatial-temporal correlation at ifrst. Then according to the weighted Bayesian theory, the algorithm generated second aggregated data CI (comprehensive information) based attribute correlation of different types of data. The simulation results on NS2 show that the algorithm can effectively reduce packet transmission capacity, predict network event type accurately, and achieve intelligent closed-loop control in CPS.