A Distortion-Aware Scheduling Approach for WirelessSensor NetworksPeriklis Liaskovitis1, Curt Schurgers11 University of California San Diego, Electrical and Computer Engineering Department{pliaskov, curts}@ucsd.eduAbstract. An important class of applications for wireless sensor networks is touse the sensors to provide samples of a physical phenomenon at discretelocations. Through interpolation-based reconstruction, a continuous map of themonitored environment can be built. In this paper, we leverage the spatialcorrelation characteristics of the physical phenomenon and find the minimumset of nodes that needs to be active at each point in time for a sufficientlyaccurate reconstruction. Furthermore, multiple such sets of nodes are found sothat a different set can report at each point in time in a rotating fashion. This iscrucial in improving network lifetime. To perform all related scheduling taskswe employ a novel approach which does not assume a-priori knowledge of theunderlying phenomenon. Instead it jointly estimates process characteristics andperforms node selection online. We illustrate that significant gains in networklifetime can be achieved with minimal impact on the overall reconstructionquality, measured in terms of distortion.Keywords: spatial random process, irregular sampling, distortion, lifetime,energy efficiency.1 IntroductionLarge scale networks of wireless micro-sensors are envisioned to enable the monitoring of physical phenomena without supervision for long periods of time. As sensor nodes often contain a limited energy supply, the individual nodes, and more importantly the network as a whole, have to operate in a highly energy efficient fashion. At the same time, it is crucial that any energy saving mechanism preserves a certain monitoring quality of the network, as this is its primary functionality.It has been realized that sensors in close physical proximity can generate correlated readings, which can be exploited to increase the overall network lifetime with minimal effect on the monitoring quality. The basic principle is that at each moment in time, only a subset of all available sensor nodes is kept active, while the others are in an energy efficient sleep mode. If the active set of sensors is chosen appropriately, the negative impact on overall sensing quality can be kept to a minimum. It is crucial to note here that such a strategy does not lead to an increase in network lifetime by itself. In fact, only if multiple such sets of sensors, i.e. sets providing adequate coverage, are found and activated sequentially, does the overall network lifetime
本文来源：https://www.2haoxitong.net/k/doc/4d3cbe03a6c30c2259019e26.html
