Situation-awareness and sensor stream mining for sustainable society

Criminal activities are causing a huge amount of loss both in term of financial and human lives. Due to these acts, business and social sectors are striving. This paper is aimed to develop an online sensor stream mining system, able to analyze situational behavior of all persons in some specific vicinity and proposes real-time alert system to take countermeasures. This system is designed to gather different information from heterogeneous sensors and fuse that information to generate realtime alerts to minimize chances of disaster. These alerts and alarms assist security personnel to take necessary decisions in real-time scenarios. The novelty of this approach comprises context-awareness with online diagnoses to take countermeasures in real-time which will in turn reduce losses of lives, society and economy. This technique enables sensor stream mining process more dependable and increases reliability of the overall system. To fulfill the objectives of this research, we have incorporated light weight online mining algorithms and link analysis to extract useful but hidden information from the gathered data. Context information like persons movement pattern, current location of that person, profile of the specific person and area of residence as well as importance of current location are exploited to detect anomalous behaviors. The major goal of this research is to detect those persons performing malicious activities and in turn minimizing exposure of society to risks and vulnerabilities.