ABSTRACT
A sensor network consists of a collection of (possibly mobile) sensing devices that can coordinate their actions through wireless communication and aim at performing tasks such as exploration, surveillance, or monitoring and tracking “target points” over a specific region, often referred to as the “mission space.” Collected data are then further processed and often support
of Complex
higher-level decision-making processes. Nodes in such networks are generally inhomogeneous, they have limited on-board resources (e.g., power and computational capacity), and they may be subject to communication constraints. It should be pointed out that sensor networks differ from conventional communication networks in a number of critical ways. First, they allow us to interact with the physical world, not just computers, databases, or human-generated data. By inserting decision-making, and control functionality into such networks one can envision closing the loop on remote processes that would otherwise be inaccessible. Thus, sensor networks are expected to realize a long-anticipated convergence of communication, computing, and control [1], [2]. Second, at least some nodes in such a network are “active,” for example, they execute sensing processes or they are mobile; therefore, they are characterized by dynamics, making a sensor network as a whole a challenging dynamic system. In addition, nodes are typically small and inexpensive, operating with limited resources, often in adverse stochastic environments. This implies that optimization in designing and operating sensor networks is a real need and not a mere luxury. Moreover, the limited computational capabilities of nodes often make distributed control or optimization methods indispensable. Finally, when it comes to measuring the performance of sensor networks, the metrics can be quite different from those used in standard communication networks, giving rise to new types of problems. For example, because of limited energy, we recognize that nodes have finite lives and we often seek control mechanisms that maximize an appropriately defined “network lifetime.” Part of such mechanisms may involve switching nodes on and off so as to conserve their energy or finding means to periodically replenish their energy supply. When the nodes are mobile, mechanisms are also needed to determine desired trajectories for the nodes over the mission space and cooperative control comes into play so as to meet specific mission objectives.
