| （英） |
In this study, we propose a high-accuracy radio map construction method for environments where bursty errors can appear in location information. Radio maps can visualize and analyze wireless environments; however, the most existing studies assume no errors in the measured location information or simplified (e.g., independent and identically distributed (i.i.d.)) error models. In practice, however, the use of user devices with positioning systems such as GNSS on smartphones often results in positioning errors ranging from several meters to several tens of meters, and these errors frequently occur in bursty, non-i.i.d. patterns. Such errors can significantly degrade the accuracy of radio maps. To address this issue, our proposed method models the spatial correlation of both location errors and wireless propagation. It applies maximum likelihood-based estimation to enable a posterior correction of location errors, thereby achieving high-accuracy radio map construction. Numerical results based on realistic human mobility demonstrate that the proposed method can improve RMSE performance compared with baseline methods. |