Fingerprint Error Detection for Radiomap Quality

The patent performs preliminary error detection and device qualification before devices begin participating in data collection. A testing phase is conducted where devices execute test routines and their fingerprint data is validated against known reference data. Only devices that pass these preliminary tests are authorized to contribute data to the radiomap, preventing erroneous data from entering the system in the first place.

The patent implements continuous feedback mechanisms where collected fingerprint data is validated against existing radiomaps and consistency checks are performed. Devices that exhibit patterns of erroneous data contribution are identified through feedback analysis and their participation is restricted or terminated. This feedback loop maintains data quality while allowing broad device participation.

The patent divides the filtering task into multiple segments performed at different levels: (1) Device-level segmentation where each device performs self-validation and basic error checks on its own data before transmission; (2) Network-level segmentation where aggregate statistics and consistency checks are performed on received data batches; (3) Server-level segmentation where final validation against the radiomap is performed. This segmented approach distributes computational load and reduces the complexity of any single filtering stage.

The patent extracts and removes erroneous data elements at the source (device level) before transmission, rather than attempting to filter complete datasets centrally. By taking out only the problematic portions identified through local validation routines and transmitting only validated data, the system reduces transmission bandwidth requirements and lowers the computational burden on centralized systems.