Changeset 2801

Timestamp:

10/30/09 18:17:42 (4 weeks ago)

Author:

jaein

Message:

 

Files:

• HydroWatch/Tim/doc/ipsn10/ipsn10-energy.tex (modified) (1 diff)

HydroWatch/Tim/doc/ipsn10/ipsn10-energy.tex

@@ -45 +45 @@
 %Long-term, environmental sampling by networks of embedded sensors has long been a vision of the computer-science community. This vision has subsequently driven much of the research over the past decade around low-power communications and network protocols, with the primary focus to date on reducing radio power consumption whilst maintaining good communications reliability. Despite these advances however, almost all existing applications of sensor networks still hold to the traditional ``data logger'' paradigm, where a user preselects a fixed operating point (such as sensor sampling rate and reporting interval) and then deploys. In this paper we propose a new architecture for environmental sensor networks, where nodes can adapt their behavior based on the amount  of energy available to harvest from the environment and within the constraints of a user-policy. Rather than nodes attempting to use as little energy has possible, nodes attempt to consume as \emph{much} energy as possible to ensure the greatest amount information can be returned as frequently as possible, whilst still keeping within necessary constraints to ensure long network lifetimes. 
 
-Long-term, environmental sampling by networks of embedded sensors has long been a vision of the computer-science community. This vision has subsequently driven much of the research over the past decade around low-power communications and network protocols, with the primary focus to date on reducing radio power consumption. Despite these advances however, almost all existing applications of sensor networks still hold to the traditional ``data logger'' paradigm, where a user preselects a fixed operating point (such as sampling and reporting interval) and then deploys. In this paper we argue that "sample-and-send" protocols are the wrong paradigm for environmental sensing. We demonstrate that in moving to a model where the radio is completely off for large periods, a whole new suite of functionality can be utilized in order to optimize the performance of nodes, subject to environmental conditions. Within this framework, we explore how the information returned by a network can be maximized while at the same time meeting fundamental requirements around responsiveness of the network and energy budgeting.
+Long-term, environmental sampling by networks of embedded sensors has long been a vision of the computer-science community. This vision has subsequently driven much of the research over the past decade around low-power communications and network protocols, with the primary focus to date on reducing radio power consumption. Despite these advances however, almost all existing applications of sensor networks still hold to the traditional ``data logger'' paradigm, where a user preselects a fixed operating point (such as sampling and reporting interval) and then deploys. In this paper we argue that "sample-and-send" protocols are the wrong paradigm for environmental sensing. We demonstrate that in moving to a model where the radio is completely off for long periods, a whole new suite of functionality can be utilized in order to optimize the performance of nodes, subject to environmental conditions. Within this framework, we explore how the information returned by a network can be maximized while at the same time meeting fundamental requirements around responsiveness of the network and energy budgeting.
 
 \end{abstract}