Neural Network Vehicle Device Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining the 3D location of a mobile device with respect to a vehicle using received signal strength indicators (RSSIs) or time of flight (TOF) values are inefficient, requiring extensive computational resources and time, and often yield inaccurate results due to the complexity of solving simultaneous linear equations and the instability of numerical methods.
Innovation Solution
A neural network is trained to determine the 3D location of a mobile device by processing RSSIs or TOFs from multiple antennas, eliminating the need for solving simultaneous linear equations and optimizing the number of antennas required, thereby reducing computational steps and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods using received signal strength indicators or time of flight values are used to determine 3D location, then location determination can be achieved, but computational resources and time are excessively consumed and accuracy is reduced due to the complexity of solving simultaneous linear equations
Solution Approach 1:
The patent replaces the traditional mechanical/mathematical computation system (solving simultaneous linear equations) with a neural network system. The neural network is trained offline to learn the mapping between signal characteristics and 3D locations, then performs rapid inference during operation. This substitution transforms a computationally intensive algebraic problem into a pattern recognition problem that can be solved efficiently by the trained neural network, thereby reducing real-time computational complexity while improving location accuracy.
2Measurement precision
If traditional methods using received signal strength indicators or time of flight values are used to determine 3D location, then location determination can be achieved, but excessive computational time is required
Solution Approach 1:
The patent applies preliminary action by training the neural network offline before actual deployment. During the training phase, the neural network learns the complex mapping relationships between signal characteristics and 3D locations using labeled training data. Once trained, the network contains pre-computed knowledge that enables rapid location determination during operation without requiring real-time solving of complex equations. This shifts the computational burden from runtime to training time, significantly reducing operational computational time while maintaining high accuracy.
3Measurement precision
If multiple antennas are used to improve location accuracy, then measurement precision improves, but device complexity and computational requirements increase
Solution Approach 1:
The patent replaces the traditional computational approach that becomes increasingly complex with more antennas with a neural network approach. The neural network efficiently processes inputs from multiple antennas simultaneously, learning optimal feature combinations and relationships during training. This allows the system to leverage multiple antennas for improved accuracy without the computational complexity scaling linearly or exponentially with the number of antennas, as the neural network has already learned efficient processing patterns during offline training.
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 neural network approach significantly reduces computational time and increases accuracy in determining the 3D location of a mobile device, allowing for efficient vehicle operations such as unlocking, locking, and lighting control based on precise device positioning.
Implementation Method 1
determining a plurality of received signal strength indicators or time of flight values for a mobile device from each of a plurality of sensors included in a vehicle
Data Source
AI summary
A computer, including a processor and a memory, the memory including instructions to be executed by the processor to determine a plurality of received signal strength indicators or time of flight values for a mobile device from each of a plurality of sensors included in a vehicle, determine a location of the mobile device with respect to the vehicle by processing the received signal strength indicators or time of flight values with a neural network wherein each received signal strength indicator is input to an input neuron included in an input layer of the neural network wherein each input neuron inputs at least one received signal strength indicator or time of flight value, and operate the vehicle using the located mobile device.