Wireless sensor network has applied to many applications to solve critical social and military problems and attracted significant research interest from both academic and industry. However, a major hurdle of deploying this technology is the battery lifetime. Nobody wants to replace hundreds or thousands of sensor batteries on a regular basis. The promise of wireless sensor networks can only be fully realized when the wiring for both data communications and power supply is eliminated. Fortunately, energy harvesting from environment becomes more and more practical. But, typically, energy harvesting cannot provide reliable energy supplies, as it commonly depends on the environment situations. For example, solar panel will depend on the weather, and vibration harvesting will depend on the vibration intensity. Those limitations pose significant new research challenges on network uptime and reliability control, which is crucial for next-generation sensor network.
Sensor networks have scarce resources (energy, bandwidth, CPU, memory). To make the best and most efficient use of these resources, sensor nodes need to *collaborate* to allocate their resources to maximize the global data quality and network lifetime. As an example consider a network where nodes sample data and relay it via a spanning tree to a base station. Each node needs to make local decisions about how to allocate energy for sampling, storage, processing, listening for incoming packets, transmitting local data, and forwarding received data. A purely decentralized, local scheme is unlikely to work well as each node does not know the capabilities and data stored by other nodes. One could do this with a central network controller; but this has high overhead. The goal is to develop a coordinated, in-network resource allocation scheme that maximizes the network data quality while meeting lifetime and bandwidth constraints.
- Mingsen Xu
- Nghia Tang