Measurement-calibrated conflict graphs for dynamic spectrum distribution

Building accurate interference maps is critical for performing reliable and efficient spectrum allocation. In this work, we use empirical data to explore the feasibility of using measurement-calibrated propagation models to build accurate interference models. Our work shows that calibrated propagation models generate location-dependent signal prediction errors. Such error pattern leads to conservative conflict graphs that actually improve the reliability of spectrum allocations by reducing the impact of unpredicted accumulative interference.