Bare necessities - Knowledge-driven WSN design

The viability of wireless sensor applications often hinges on minimising power consumption whilst maximising the informational output. Although many low-level platform-oriented energy saving mechanisms have been developed, considerable savings are possible at application level. This work presents an approach to pushing the calculation of application-level state closer to the information source. The context in which this approach is evaluated is a residential building monitoring application. Combined with the Spanish Inquisition Protocol (SIP), this is shown, based on deployment data, to reduce the average transmission period for temperature data from once every 5 minutes to an average of once every 38 days for an allowed error threshold of 10% on any component of the application-level state. For combined sensing of temperature, relative humidity and CO ₂, the average transmission period drops to 13 days. This transmission reduction should considerably extend network life while having minimal effect on the usefulness of the information gathered. Most importantly, the underlying approach generalises to a wide variety of applications.