Indoor Space Recognition Training Data Using RSSI Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing indoor positioning technologies face challenges in accurately measuring location within indoor spaces due to poor GPS signal reception, and manual data collection for training indoor space recognition models is inefficient and inconvenient for users.
Innovation Solution
A method for generating training data using a mobile terminal that automatically collects RSSI values from multiple signal sources, performs clustering, and selects a feature set based on RSSI overlap and distribution scores to efficiently generate training data for an indoor space recognition model, reducing the need for manual input and prolonged user presence in each space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual data collection is used for training indoor space recognition models, then data quality can be controlled, but user convenience deteriorates and data collection efficiency is low
Solution Approach 1:
The mobile terminal automatically performs data collection, feature extraction, and training data generation without requiring manual user intervention. The system self-services by autonomously measuring RSSI values, selecting features based on overlap and distribution scores, and generating training data, thereby eliminating the need for manual data collection while maintaining data quality.
Solution Approach 2:
The system performs preliminary actions by automatically collecting RSSI data from multiple signal sources and pre-processing it to identify meaningful features before training. The automatic feature selection based on overlap and distribution scores occurs in advance, preparing high-quality training data without requiring manual curation or user presence in each space.
2Measurement precision
If GPS signals are used for location measurement, then outdoor positioning accuracy is high, but indoor positioning accuracy deteriorates due to poor signal reception
Solution Approach 1:
The patent uses RSSI values from multiple signal sources (Wi-Fi, Bluetooth, cellular) as intermediary indicators to indirectly determine indoor location. Instead of directly using GPS signals that fail indoors, the system measures signal strengths from available sources and uses these intermediary measurements to infer spatial information, enabling accurate indoor positioning without relying on GPS.
Solution Approach 2:
The system replaces the mechanical/electromagnetic GPS signal-based positioning mechanism with a different approach using wireless network signal (Wi-Fi, Bluetooth) RSSI measurements. This substitution allows the positioning system to function in indoor environments where GPS signals are unavailable, maintaining location measurement capability through alternative signal sources.
3Quantity of substance
If all signal sources are used for training, then training data comprehensiveness is high, but data processing complexity increases
Solution Approach 1:
The system extracts and selects only the most informative features from among all available signal sources. By calculating overlap scores (indicating spatial distinguishability) and distribution scores (indicating signal variability) for each signal source, the system extracts and retains only those signals that provide meaningful discrimination between spaces, filtering out redundant or uninformative signals to reduce processing complexity.
Solution Approach 2:
The patent applies local quality by evaluating and selecting signal sources based on their specific characteristics in different indoor environments. Instead of uniformly treating all signal sources equally, the system assesses each signal's overlap and distribution properties and selectively uses only those signals that exhibit locally optimal characteristics for space recognition, optimizing the training data for specific spatial contexts.
Data Source
AI summary
A method is directed to generating training data for training an indoor space recognition model. The method includes collecting, by using a mobile terminal in a plurality of spaces in an indoor space, data obtained by measuring received signal strength indicator (RSSI) values of signals transmitted from a plurality of signal sources; obtaining reference data in which the RSSI values measured for the signals transmitted from the plurality of signal sources are matched and stored with the plurality of spaces; selecting one or more signal sources of the plurality of signal sources as a feature set based on the collected data and the reference data; performing clustering on the collected data and the reference data based on the feature set; and generating the training data by matching and storing the collected data with at least one of the plurality of spaces based on a result of the clustering.