(Reactive + Proactive Behavior) Situation Awareness in Sensor Networks

In order to fully exploit all the potentials of various heterogeneous resources in sensor network applications in a context-aware manner, coordinated research efforts need to be pursued in multiple technical areas including networking, information management, control, and spatio-temporal data/events management. At the heart of the motivation for our research is the observation that there is a lack of a unified paradigm that will enable dynamic balancing between the Quality of Data (QoD) assurances and the coordination of the resource-utilization, not only as reaction to an occurrence of a particular event, but also with a proactive impact on how the future tracking and data gathering should be managed. These are crucial capabilities that are needed for implementing any system that relies on automated situation awareness. Our work is focusing on adaptive and environment-aware integrated management of spatiotemporal queries and events in heterogeneous sensor networks. In lieu of the Network is the Database perspective [2], some recent works have addressed the goal of declarative networking which essentially aims at providing high-level declarative languages for managing the behavior of the network (e.g., [9]). In particular, the DSN (Declarative Sensor Networks) paradigm combines a language, compiler and run-time system for high-level specification and deployment of large variety of applications [4]. The language – Snlog, is a variant of Datalog [8] with two notable extensions: “@”-symbol for denoting that the type of the variable is a node, and “#”denoting a link. In the mid-1990s, deductive databases and active databases (ADb) [5] co-developed, both having a similar goal: providing expressive languages to enable users to easily specify what the system needs to do, without getting into the details of how. However, ADb leaned more towards the specification of the reactive behavior of the systems and arguments were made that the deductive and active databases can/should be viewed as two end-points of the spectrum of the family of rule-based languages in databases [10], along with attempts for their integration [3]. Although declarative languages are emerging in sensor networks settings, the triggers, which are the basic mechanism for specifying the reactive behavior in ADb, have not yet found their counterpart. In this position paper, we focus on efficient management of two types of information pertaining to mobile entities: (1) location-in-time and (2) context, and we combine the behavioral and semantic-knowledge with the optimal resource utilization. We will consider the collaborative behavior of two types of sensors used for detecting the information of interest: (1) low-cost and low-energy acoustic and seismic (vibration-based) sensors, and (2) multiple video sensors. Towards the goal of situation awareness, we present a methodology that, in reaction to a particular event in a given state of the system, will proactively adjust the resources’ exploitation, so that the highest Quality of Data (QoD) level is assured in balance with some parameters of interest, e.g., networks connectivity and lifetime.