[中文]

Wireless sensor networks (WSNs) are often deemed as appropriate solutions for environmental surveillance and real-time monitoring applications. In the past decade, WSNs have gained a lot of attention from the research and industry community. Countless research projects have been launched to study different aspects of WSNs and yielded fruitful results. The uses of WSNs nowadays, however, are far restricted by many theoretical or practical obstacles. For examples, most WSNs adopt un-rechargeable batteries with limited power supply. The wireless links among the sensor nodes appear to be unreliable and easily affected by various environmental factors. Nevertheless, there is hardly any fundamental result on the data collection capacity of a duty-cycled WSN with unreliable wireless links. Further considering the faulty behaviors of sensor nodes and the difficulties of outdoor deployment and maintenance, a real WSN system often fails to support long-term large-scale applications with guaranteed performance.

What are the fundamental challenges in long-term large-scale WSNs? Which research and technical issues must be addressed? Where are the potential design spaces of future WSN research? The research and industrial communities are in urge need of the answers to the three questions.

Under such circumstances, we launched GreenOrbs, a WSN system of 1000+ nodes to operate in the forest for over one year. Through the experience in GreenOrbs, we gain considerable insights on the challenges and design space in long-term large-scale WSNs, such as energy consumption, scheduling and synchronization, routing efficiency, link estimation, encapsulation, deployment, diagnosis, and fault tolerance, etc.