Indoor Positioning via CNN-Processed WiFi CSI Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current indoor positioning technologies, such as GNSS and GPS, face challenges in accurately positioning objects indoors due to signal interference from reinforced concrete structures and multi-path phenomena caused by obstacles, leading to non-line-of-sight errors and reduced accuracy.
Innovation Solution
An improved convolutional neural network (CNN) model is used to preprocess and analyze CSI data from Wi-Fi signals, partitioning it into subsets, and training a fingerprint database to accurately extract coordinate information, thereby enhancing indoor positioning accuracy by mitigating noise and multi-path effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If Wi-Fi signals are used for indoor positioning, then positioning can be achieved in large indoor environments without new instruments, but NLOS errors and multi-path phenomenon occur due to signal blockage by obstacles
Solution Approach 1:
The patent converts the harmful multi-path signal components into useful information by training a CNN model to recognize and process both direct and reflected signal paths. The model learns to identify positioning patterns even when signals bounce off surfaces, transforming the previously harmful multi-path effect into a可利用 signal source for positioning.
Solution Approach 2:
The patent introduces CSI (Channel State Information) data as an intermediary between the Wi-Fi signal and the positioning calculation. Instead of directly using signal strength or time of flight, the system processes CSI data through a CNN model that acts as a mediator to extract accurate position information while filtering out NLOS and multi-path interference.
2Measurement precision
If traditional positioning algorithms are used, then computation is simpler, but positioning accuracy is reduced due to inability to handle NLOS and multi-path effects
Solution Approach 1:
The patent replaces traditional mathematical positioning algorithms with a deep learning-based CNN model. Instead of using geometric calculations or signal processing formulas, the system uses a neural network that learns positioning patterns from data, substituting mechanical/mathematical computation with an adaptive learning system that handles complex signal environments more effectively.
Solution Approach 2:
The patent changes the input parameters from traditional positioning metrics (signal strength, time of flight) to CSI data, which contains detailed channel characteristics. This parameter transformation enables the system to capture subtle signal variations that indicate position information even in the presence of NLOS and multi-path effects.
3Measurement precision
If raw CSI data is processed directly, then all information is retained, but computation burden increases and noise interference affects accuracy
Solution Approach 1:
The patent extracts only the most relevant features from CSI data by using the CNN model to automatically identify and extract positioning-critical characteristics. Instead of processing all raw CSI information, the model learns to extract only the features that contribute to accurate positioning, discarding redundant and noisy components.
Solution Approach 2:
The patent performs preliminary processing of CSI data through the trained CNN model before actual positioning calculation. The model pre-processes the data by filtering noise, normalizing variations, and organizing signal characteristics in advance, so that the final positioning computation works with cleaned and structured information rather than raw noisy data.
Data Source
AI summary
An indoor target positioning method based on an improved convolutional neural network (CNN) model includes acquiring and preprocessing target camera serial interface (CSI) data of a to-be-positioned target and matching the preprocessed target CSI data with fingerprints in a positioning fingerprint database to obtain coordinate information of the to-be-positioned target. The generation method of the positioning fingerprint database includes: collecting indoor WiFi signals by a software defined radio (SDR) platform to obtain indoor CSI data corresponding to the WiFi signals, and preprocessing the indoor CSI data; partitioning the preprocessed indoor CSI data into a plurality of data subsets through a clustering algorithm; training an improved CNN model by the data subsets to obtain a trained improved CNN model; and generating the positioning fingerprint database by the trained improved CNN model and the preprocessed indoor CSI data.