Cloudy Computing: Leveraging Weather Forecasts in Energy Harvesting
Sensor Systems
Abstract
Wireless Sensor Networks (WSNs) comprise of small nodes with sensing,
computation and wireless communication capabilities. WSNs are used for
various applications such as habitat monitoring, automation, agriculture, and
security. This work is targeted for Weather Forecasting. Since numerous sensors
are usually deployed on remote and inaccessible places, the deployment and
maintenance should be made easy and scalable. Nodes in a sensor network are
severely constrained by energy and computing power. To prolong the lifetime of
the sensor nodes, efficient routing protocol that could also optimize the energy
consumption is critical.
To sustain perpetual operation, systems that harvest environmental energy must
carefully regulate their usage to satisfy their demand. Regulating energy usage is
tricky if a system’s demands are not elastic and its hardware components are not
energy-proportional. We explore the use of weather forecasts to improve a
system’s ability to satisfy demand by improving its predictions.
The simulation results are expected to indicate that the proposed prediction for
weather forecasting would increase the lifetime of sensor networks. It is also
anticipated that this could outperform than any other prediction algorithm.
Existing System
Energy harvesting systems collect and store environmental energy to sustain
continuous operation without access to external power sources. Energy
harvesting has been well studied in the context of sensor networks, because they
are often deployed in remote locations with no access to the power grid.
Regulating energy usage is current issue since the system must choose when to
satisfy its demands based on its current energy reserves and predictions of its
future energy supply.
Proposed System
We consider weather forecast, and energy data to formulate a model. We
estimate our model for energy sources with and without inelastic demands. We
show that using weather forecasts for predictions in sensor systems increases
each system’s ability to satisfy its demands compared with existing strategies.
Energy predictions derived from weather forecasts for large regions improve
nearby systems’ ability to satisfy their demand over the time-scales, when
compared against predictions derived from the immediate past.