ABSTRACT
Clearly, the untethered and unattended nature of sensors constrain their actions as individual
devices, since they must independently and efficiently utilize their limited energy resources. However,
designing sensor network solutions that only optimize energy consumption will not always lead to
efficient architectures, since the above constraints do not account for collaborative trade-offs between
groups of sensors. Note that collaborative interaction among sensors provides some network-wide
benefits (as opposed to ‘‘energy’’ benefits to individual sensors), where network-wide is a semantic term
referring to overall goals of the entire network or to a sufficiently large group of sensors. Consider, for
example, collaborative data mining/information fusion among sensors to respond meaningfully to
queries [5,6]. Too many sensors simultaneously participating in the collaborative decision making
required for aggregation of mined data will lead to excessive routing paths in the network, thereby
increasing energy consumption and competition for communication resources. On the other hand, too
little collaborative data aggregation will make distributed mining inaccurate and ineffective. Thus,
sensors are implicitly constrained by a third factor: to increase information utilization of the network,
sensors must cooperate to maximize network-wide objectives while maximizing their individual
lifetimes. We label this paradigm for broad sensor network operation as sensor-centric.