Mobile Device Locationing Using Inference Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing location determination methods for mobile devices in environments without GPS, such as warehouses and retail stores, are costly due to infrastructural requirements and suffer from reduced accuracy due to multipath RF propagation and noisy measurements.
Innovation Solution
A method and system that uses a mobile device to collect proximity indicators from fixed transmitters, such as wireless access points, and employs inference model data to determine its location within a predefined frame of reference, utilizing machine learning processes like neural networks to correlate transmission data with specific locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If navigational aids and host computing devices are deployed to determine mobile device location, then locationing capability is achieved, but system cost increases
Solution Approach 1:
The mobile device performs location determination autonomously using its own processor and stored inference model data, without requiring a host computing device or server to calculate location. The device collects proximity indicators, applies the inference model locally, and generates location information independently, eliminating the need for centralized computational infrastructure.
Solution Approach 2:
The location determination function is extracted from the host computing device and transferred to the mobile device itself. By deploying the inference model to the mobile device, the system removes the dependency on centralized servers for location calculation, allowing the mobile device to standalone and determine its own location using only proximity indicators from fixed transmitters.
2Reliability
If traditional locationing methods are used in environments with multipath RF propagation, then locationing is achieved, but measurement accuracy deteriorates
Solution Approach 1:
The inference model acts as an intermediary between raw proximity indicator measurements and final location determination. Instead of directly converting signal strength to location, the inference model processes the proximity indicators through learned patterns and relationships, filtering out noise and multipath effects to produce more accurate location estimates.
Solution Approach 2:
The system transforms the approach by changing from direct geometric calculation methods to machine learning-based inference. The inference model learns optimal parameter relationships between proximity indicators and location from training data, adapting to environmental characteristics including multipath propagation patterns to improve measurement accuracy.
Data Source
AI summary
A method of determining a location of a mobile device in a system having a plurality of fixed transmitters includes: obtaining, at the mobile device, inference model data defining a plurality of node operations; collecting, at the mobile device, respective proximity indicators corresponding to a subset of the fixed transmitters, each proximity indicator representing a proximity of the mobile device to the respective fixed transmitter; at the mobile device, generating a location according to the proximity indicators and the node operations; and presenting the location.