Skip to Main Content
In this paper we discuss utility-based network optimization for Wireless Sensor Networks (WSNs) and Machine-to-Machine communications. In the Internet setting the utility function to be maximized is usually assumed to be a sum of individual utilities of end-to-end flows. For WSNs the situation is highly different. Usually a sensor network is deployed to monitor the evolution of a particular phenomenon over time in some region of interest. The key performance metric of the network becomes then how accurately the network measures this evolution compared to the requirements of the application. This makes utility-based reasoning on network performance much more complex for WSNs compared to the traditional Internet model, since the utility will depend in a complex manner on phenomena under study, the topology of the network, as well as on the algorithms used for state estimation from the individual measurements.We focus here on spatial and temporal monitoring problems, give an overview of the various estimation approaches that are applicable for data processing in the corresponding WSN deployments, and give examples on how utility-based optimization and be used to enhance the lifetime and performance of WSNs. We also discuss the related architectural issues, in particular how utility functions for WSNs could be expressed in a form suitable for automated processing.