In recent years, pavement instrumentation became a common practice for structural monitoring. Instrumented roadway includes different type of sensors such as strain gauges, accelerometers or geophones. Data processing consists often in extracting different values such as extremum, number of positives or negatives peaks, calculate threshold, on so on. These values are generally representative of variations in the whole structure, but make it more difficult to follow a specific characteristic of the structure. To overcome this limitation, an original method has been defined. It consists in finding a weighting function of the signal that is particularly sensitive to a specific characteristic of the structure without being sensitive to the other characteristics. This weighting function is calculated using a model of the structure. The sensitivity of the response of the simulated sensor to different variations in the characteristics of the structure is studied. From these sensitivities, it is possible to define specific weighting functions. The application of these weighting functions to the measured signals makes it possible to deduce indicators specific to different characteristics of the structure. This method has been applied during an experiment on the Nantes fatigue carousel. Stain gauges equipped the roadways and lot of signals has been recorded during the loading phase. A viscoelastic pavement model has been defined and the response under a running load was calculated. Optimal weighting functions have been estimated and specific indicators constructed. Stain gauge measurements have been processed to calculate the indicator; The paper presents a comparison between constructed indicators and conventionally used indicators.