WiFi Signal Deep Learning Object Detection System

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing security systems in indoor and vehicular environments lack efficient, device-free methods for detecting and identifying objects using WiFi signals, particularly in real-time, with limitations in accurately determining location, type, and number of objects.

Innovation Solution

A deep learning neural network-based security system utilizing WiFi nodes and a deep learning module that processes Channel State Information (CSI) and Received Signal Strength Indicator (RSSI) data to detect and identify objects, including humans and pets, by filtering noise, generating Short-Time Fourier Transform (STFT) and Continuous Wavelet Transform (CWT) of CSI data, and classifying presence, type, and number of objects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If existing security systems use traditional detection methods, then device complexity is reduced, but measurement precision and detection accuracy deteriorate

Engineering Contradiction:
Improvedetection accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces traditional mechanical sensors and detection devices with a deep learning neural network that processes WiFi signal data. The system uses software-based object detection and classification algorithms to identify objects, people, and pets without physical contact, substituting mechanical detection systems with intelligent signal processing that achieves higher precision while maintaining reasonable system complexity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces WiFi signals as an intermediary medium for detection. Instead of using direct sensor contact with objects, the system uses WiFi signals to indirectly detect object presence, location, and characteristics. This intermediary approach enables non-contact detection with high accuracy while avoiding the complexity of deploying multiple physical sensors throughout the environment.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Ease of operation

If device-free passive detection is implemented, then ease of operation is improved, but measurement precision deteriorates

Engineering Contradiction:
Improveease of deploymentVSAvoiddetection accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent transforms raw WiFi signal parameters (CSI, RSSI) into meaningful detection data through deep learning processing. By changing the parameters from simple signal strength measurements to complex neural network feature extractions, the system achieves high measurement precision while maintaining device-free operation. The neural network learns to extract relevant features from signal variations that indicate object presence and characteristics.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent performs preliminary training of the deep learning model offline before deployment. During this preliminary action phase, the system learns from labeled data to recognize patterns associated with different objects, people, and pets. This pre-trained knowledge enables the device-free system to achieve high detection accuracy without requiring complex real-time processing or additional sensors during operation.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If real-time detection is implemented, then productivity is improved, but use of energy deteriorates

Engineering Contradiction:
Improvereal-time detection capabilityVSAvoidenergy consumption
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent implements periodic detection cycles rather than continuous monitoring. The deep learning model processes WiFi signal data at intervals, analyzing signal variations over time periods to detect objects and track their movements. This periodic approach maintains real-time detection capability for safety-critical applications while reducing energy consumption compared to continuous high-frequency processing.

Inventive Principle:
Principle #19Periodic action

Data Source

PatentEP3531350B1Deep learning neural network based security system and control method therefor
Publication Date: 2021.04.21 LG ELECTRONICS INC
  • EP3531350B1 patent drawingFigure 1
  • EP3531350B1 patent drawingFigure 2
  • EP3531350B1 patent drawingFigure 3

AI summary

The present invention relates to a deep learning neural network based security system and a control method therefor and, more particularly, to a deep learning neural network based security system comprising: at least one WiFi node; and a deep learning module for detecting an object from a WiFi signal received from the WiFi node, wherein the deep learning module identifies information on the object when the object is detected.