Fang M, Malone D. (2010). Experimental verification of a radiofrequency power model for wi-fi technology. Health Phys. 98(4):574-83.
Wi-Fi networks use radiofrequency energy for data transfers. Currently, there is a tendency towards denser Wi-Fi deployments, especially in urban areas. Offices, classrooms, conference halls and some houses or apartments have large numbers of devices, such as laptops, phones, personal digital assistants, game consoles and media players, with high levels of activity. Although Wi-Fi has not attracted the same level of interest as mobile phone networks, it is timely to carry out further research on this technology. In particular, it is useful to model how the power output depends on the number of stations and level of activity in order to determine whether or not radiation levels are within the accepted guidelines. Some existing models, such as the one of Malone and Malone (2009) are based on the assumption that each device always has data to transmit. Other models account for the intermittent nature of Wi-Fi transmissions.
The objective of this study was to extend the model of Malone and Malone (2009) for estimating the power output of a network that is not always busy and to compare the theoretical predictions of transmitted power with experimental results.
The model of Malone and Malone (2009) was extended using the non-saturated models described by Duffy and Ganesh (2007) and Malone et al. (2007). The total power output was calculated based on the saturated and unsaturated models. Some important factors, such as the distances between devices, reception errors, interference from other devices sharing the same frequency, were not taken into account. Experiments were carried out on a wireless testbed with configuration similar to that which might be found in a home or in a public place. The testbed included 9 stations; each station was representative of a laptop or other wireless device. The model predictions for the saturated and non-saturated models were compared to the results of experimental measurements with varied numbers of stations and the offered load.
Results and Interpretation
There was a good agreement (discrepancies within 5-10%) between model predictions and experimental results. As expected, transmitted power increased as a function of the number of stations and the traffic level. When there was a small amount of traffic in the network, the power increased linearly with increasing the offered load. The greatest power was emitted when the network was saturated with traffic. This suggests that the power of a saturated network can be a reasonable approximation of the upper bound on the output power of an unsaturated network. In the appendix, the authors provide examples of how to predict power output from a Wi-Fi network based on traffic levels.
This study provides a simple and accurate technique for estimation of power output based on traffic levels