Title of Presentation: Architecture of A Satellite-Based Sensor Network for Environment Observation

Primary (Corresponding) Author: Wei Ye

Organization of Primary Author: University of Southern California

Co-Authors: Fabio Silva, Annette DeSchon and Spundun Bhatt

Abstract: Sensor networks are a promising technology to enable real-time, in-situ environment observation. Despite the rapid advancement in research, there are challenges for the scientists to widely adopt this technology. For example, such a system must be easily deployable at remote locations. The system should be flexible to meet the requirements of different scientific applications. It also needs to be easily reconfigurable by scientists, who may not be software or networking experts. To address these challenges, we have designed a robust and flexible sensor network with satellite communications, called Sensor Processing and Acquisition Network (SPAN). This paper describes SPAN architecture and its first prototype that supports ecological research.

The SPAN architecture consists of a front-end system and a back-end system, connected by satellite communications. The front end includes all components deployed in the field, such as sensors, data acquisition, meta-data tagging, local wireless network and reliable data transfer. The back end provides data storage and system reconfiguration tools. Our architecture currently allows remote, centralized decision feedback to change the system's sampling behavior. We are exploring further support of local and distributed triggers for rapid sampling response and network adaptation.

The design of SPAN system emphasizes robustness, flexibility, ease of use, and extensibility. To achieve robustness we perform extensive system monitoring and data caching in stable storage. We developed a unified sensor integration framework that allows streamlined integration of various industry-grade sensors for different science applications. In addition, we provide intuitive user interfaces and tools for scientists to easily control and reconfigure the system. While initially targeted for individual scientists, SPAN can easily scale up to support a group of scientists or even community infrastructure.

We have deployed our first prototype at Stunt Ranch in Santa Monica Mountains in collaboration with biologists at UCLA to support their study of the impact of drought conditions in southern California to plants. The system has successfully operated in the field for several months, and validated our design approaches.