Low Storage Radio Map Using Statistical Inference

A crowdsourcing-assisted radio map has attracted attention for providing statistical information about primary users to secondary users. The radio map is constructed by averaging the received signal power samples observed by mobile terminals in each location. However, most conventional works do not consider the required sample size to estimate the average received signal power. If the crowdsourcing is continuously conducted, the observation time of the radio environment and registered data size may be enormous according to the area size. Thus, we propose two methods for determining the required sample size to estimate the average received signal power. In the first method, the interval estimation and central limit theorem are utilized to estimate the required sample size in each mesh. As the second method, the t-test in hypothesis testing is used. The simulation results reveal that the t-test-based method can accurately estimate the average received signal power while notably reducing the registered data size and observation time.