Assessing the impact of field-measurement on the design of Spectrum Sensing WSN

Measuring the pathloss between points on a network (i.e., a site survey) is a time-consuming and costly process. Understandably, WSN designers rely on (empirical) models of signal propagation for indoor environments combined with a calibration phase to estimate a priori the sensor coverage area. However, the accuracy of these indoor models decreases as the signal crosses floors in the same building or between buildings. Thus, when the sensing range of a sensor is relatively large compared with the communication range for which the models are developed - as it is the case in a Spectrum Sensing WSN - a model-based WSN design will incur a significant error. Previously, authors have concentrated on intermediate figures of error: either the pathloss estimation or the WSN coverage area. Our work focuses on the "bottom line", that is, the error in the total number of sensors required by the design, which directly associates with the implementation cost. The final results show that in the case of spectrum sensing for a WiFi network in typical school buildings, the use of actual pathloss measurements results in significant savings of up to over the 50% in the number of sensors required.