Sensor Web: Integration of Sensor Networks with Web and Cyber Infrastructure

Introduction

Sensor networks are persistent computing systems composed of large numbers of sensor nodes. These nodes communicate with one another over wireless low-bandwidth links and have limited processing capacity. They work together to collect information about their surrounding environment, which may include temperature, light or GPS information. As sensor networks grow and their ability to measure real-time information in an accurate and reliable fashion improves, a new research challenge, how to collect and analyze recorded information, presents itself.

Deployment scenarios for sensor networks are countless and diverse, sensors may be used for military applications, weather forecasting, tsunami detection, pollution detection and for power management in schools and office buildings. In many of these cases the software management tools for data aggregation and decision making are tightly coupled with each application scenario. However, as these systems grow and mature, a set of common data operations and transformations begin to emerge. All application scenarios will need to query a sensor network and retrieve some observational data. Some scenarios may require information from historic queries be stored in a repository for further analysis. They may require regular queries to be scheduled and automatically dispatched without external operator intervention. Scenarios may need to share information among themselves to aid in decision making tasks. For example, a tsunami warning system may rely on water level information from two geographically distributed sets of sensors developed by competing hardware vendors. These requirements present significant challenges in resource interoperability, fault tolerance and software reliability. A solution to these emerging challenges is to implement a set of uniform operations and a standard representation for sensor data which will fulfill the software needs of a sensor network regardless of the application or deployment scenario.

A Service Oriented Architecture (SOA) allows us to describe, discover and invoke services from heterogeneous platforms using XML and SOAP standards. Services can be defined for common operations including data aggregation, scheduling, resource allocation and resource discovery. We can exploit these properties by combing sensors and sensor networks with a SOA to present sensors as important resources which can be discovered, accessed and where applicable, controlled via the World Wide Web. We refer to this combination of technologies as the Sensor Web. Taking this concept a step further, when interlinked, geographically distributed services form what is called a Sensor Grid which is a key step in the integration of sensor networks and the distributed computing platforms of SOA and Grid Computing. The integration of Sensors Networks with the cyber infrastructure of Grid Computing brings several benefits to the community. The heavy load of information processing can be moved from sensor networks to the backend distributed systems. This separation is beneficial because it reduces the energy and power needed by the sensors, allowing them to concentrate on sensing and sending information. Cross-organizational collaboration is streamlined, because geographically distributed resources can be accessed over common Grid protocols. Data produced by heterogeneous resources can be combined with the aid of common XML formats, eliminating data incompatibility issues.

Figure 1 demonstrates an abstract vision of the Sensor Web; various sensors and sensor nodes form a web view and are treated as available services to all the users who access the Web. A researcher wishing to predict whether a tsunami is going to occur may query the entire Sensor Web and retrieve the response either from real-time sensors that have been registered on the web or from historical data in database. Data from all sources can be aggregated and used by modeling or visualization tools to aid in tsunami prediction. This can be shared among collaborative parties which may run algorithms or transformations over the raw data with the aid of grid resources. In this way, individual resources can be coupled together to perform complex tasks which where not previously achievable.

Figure 1.

Abstract vision of the Sensor Web