Making wireless sensor node platforms energy efficient is one of the major research thrusts in the sensor network community. Energy metering lies at the foundation of this research, either by providing direct measurements for profiling, or by serving as the base for the formulation and fitting of energy usage models. Most of the literature and tools, however, make their measurements on a very small subset of the node population, and usually at a single point in time, before deployment. In this paper we set out to evaluate the cost, in loss of precision, of not having constant and ubiquitous measurement. Through experiments on a 240-node sensor-network testbed, we find that the variations in energy consumption due to temperature change are small, and we establish a model between environmental temperature changes and power consumption of Quanto testbed motes. We also find that different nodes of the same kind can have up to 15% variation in power draw, suggesting a need to deploy instrumentation on a subset of nodes. We quantify the energy estimation error of different metering techniques and characterize the conditions in which the errors disappear. Overall, we find that a small number of measurements in time and across nodes is adequate for accurate estimation of network-wide energy use.

We present our experiences evaluating the power-performance tradeoffs of a sensornet network protocol on a power-aware testbed. We characterize the power draw of the entire network while running the Collection Tree Protocol (CTP), as a function of low-power-listening interval. We find that message transmission counts are poor predictors for energy consumption on the CC2420 radio, that CTP routinely creates energy hotspots in the routing tree, and that conclusions based on protocol evaluation performed without low-power listening enabled provide little insight about the same protocol performance using low-power listening.