Skip to Main Content
In wireless sensor networks, a good cost metric encapsulating wireless link quality is essential to an energy-efficient routing topology. For many wireless network scenarios, rapid variation in the channel precludes an efficient mechanism for knowing instantaneous link quality at the time of transmission, thus making it difficult to estimate the instantaneous value of the cost metric. This paper explores what a good cost metric may be and how it can be measured in an energy-efficient way. We present an experimental study of wireless link quality variation over a period of several days in a sensor network placed in two different indoor office environments. The nodes are equipped with low power transceivers operating in the 900 MHz band. Results are documented for several different link configurations, i.e., relative placement of the transmitter and receiver. Based on detailed observations of link quality variation, we form quantitative measures of link quality, and propose a cost metric. We find that reasonably few channel measurements are sufficient to obtain a good estimate of the cost metric, hence even during the initialization phase one can obtain sufficient information about links in order to design an efficient topology. The estimate can be improved further as more measurements are taken during the normal operation of the network.