Two-Neural Network Indoor Localization System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Indoor wireless localization systems face challenges in accurately determining the location of targets due to complexities in wireless signal propagation, such as multipath interference, especially when using radio technologies like Wi-Fi, Bluetooth, or ultra-wideband, which limits their precision in indoor environments.
Innovation Solution
A method involving a neural network system comprising two neural networks is trained to infer the location of a target from a plurality of localization parameters, including velocity-related parameters derived from Doppler shifts, and other parameters like channel impulse response and angle of arrival, to improve location determination accuracy by compensating for signal interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional single neural network approaches are used for indoor localization, then the system complexity is low, but the location accuracy deteriorates due to multipath interference and signal propagation complexities
Solution Approach 1:
The patent applies segmentation by dividing the localization task into two distinct neural networks: a first neural network that processes signal strength parameters (RSSI, TOA, TDOA, AOA) to generate candidate locations, and a second neural network that processes velocity parameters (Doppler shifts) to refine the location estimate. This segmentation allows each network to specialize in specific signal characteristics, improving overall accuracy while managing complexity through functional decomposition.
Solution Approach 2:
The patent introduces a temporal dimension by incorporating velocity parameters derived from Doppler shifts of consecutive signal measurements. This adds a time-based dimension to the localization problem, transforming it from a static spatial estimation to a dynamic estimation that leverages motion information. The second neural network operates in this extended dimension to refine locations by considering the target's movement trajectory.
2Measurement precision
If velocity parameters from Doppler shifts are incorporated into the neural network system, then location accuracy improves through refinement of candidate locations, but the computational complexity and data processing requirements increase
Solution Approach 1:
The first neural network performs preliminary processing by generating candidate locations from signal strength parameters before the second neural network refines them using velocity information. This preliminary action reduces the search space for the second network, as it only needs to evaluate refinements among pre-identified candidate locations rather than considering all possible positions, thereby reducing computational power requirements.
Solution Approach 2:
The candidate locations generated by the first neural network serve as an intermediary between the signal strength measurements and the final refined location. This intermediary representation allows the second neural network to focus computational resources on refining specific candidate positions rather than processing raw signal data directly, optimizing the balance between accuracy and computational power.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The two-neural-network system enhances location accuracy by refining candidate locations based on velocity and other signal parameters, reducing errors and improving precision in indoor localization compared to conventional single-neural-network approaches.
Implementation Method 1
the plurality of localization parameters comprise one or more parameters relating to a velocity of a target and one or more other parameters
Data Source
AI summary
A neural network system for inferring a location of a target from a plurality of localization parameters derived from a wireless signal and a method of training thereof. The neural network system comprises first and second neural networks. The plurality of localization parameters comprise one or more parameters relating to a velocity of a target and one or more other parameters. The first neural network is trained to infer a set of candidate locations of a target from values of the one or more other parameters. The second neural network is trained to infer a location of the target from values of the one or more parameters relating to a velocity of the target and a set of candidate locations of the target.