Function Computation in Wireless Sensor Networks

With the wide availability of inexpensive wireless technology and sensing hardware, wireless sensor networks are expected to become commonplace because of their broad range of potential applications. A wireless sensor network consists of sensors that have sensing, computation and wireless communication capabilities. Each sensor monitors the environment surrounding it, collects and processes data, and when appropriate transmits information so as to cooperatively achieve a global detection objective. One important feature of wireless sensor networks is that the network is often designed for a specific purpose, and the sensors are required to collaborate to achieve a global objective. This is one fundamental distinction between wireless networks used for communication and wireless networks used for sensing. In wireless communication networks, the protocols are designed so that they are not applicationspecific, and therefore the network can support a constantly evolving set of applications. Contrasting this, in sensor networks, the architecture and protocols can be designed for each specific application, exploiting its structure, to reduce the energy usage within the network. Note that the objectives of sensor networks are to retrieve useful information from sensor measurements, so many of these objectives can be regarded as a function computation of sensor measurements.