Wi-Fi Fingerprint Positioning Using Information-Gain AP Filtering

The patent extracts and selects only the most relevant and stable WI-FI access points from the large set of available signals. By calculating information gain for each access point and selecting those with high information gain values, the system reduces the number of signals used for positioning while maintaining accuracy. This extraction process filters out redundant and unstable signals, resolving the contradiction between using many signals for accuracy and avoiding complexity.

The patent changes the parameter selection criteria by introducing information gain as a filtering metric. Instead of using all monitored access points, the system calculates information gain for each access point based on its characteristic values and selects only those above a threshold. This parameter-based filtering reduces the volume of data requiring processing while preserving the identification rate, as the selected access points are those most discriminative for location identification.

The patent performs preliminary actions by pre-calculating and storing metadata for each WI-FI access point, including information gain values and characteristic values. This pre-processing is done offline before actual positioning operations. During runtime, the system only needs to compare current fingerprints against the pre-processed database, significantly reducing time consumption while maintaining positioning accuracy, as the heavy computational work has already been completed in advance.