This paper presents a propagation model generating time series of large-scale power variations for small-cell radio links intersected by vehicular traffic. The model combines stochastic processing and geometric computation. For each road crossing a link, a two-state process parameterized by mobility statistics represents the obstruction status. When the status is set to obstructed, a fluctuation pattern is generated. Based on previously published measurements, both mobility statistics and time series results are validated through the comparison of respectively inter-obstruction duration distributions and outage probabilities. The proposed model avoids resource-consuming iterative propagation prediction while providing realistic and frequency adaptive results.