Browse Prior Art Database

Smart house systems: sensors that can be linked to form neural network

IP.com Disclosure Number: IPCOM000247946D
Publication Date: 2016-Oct-13
Document File: 3 page(s) / 124K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method to link existing sensors in a house to form a neural network capable to learn and thus improve smart house capabilities

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 01 of 3

Smart house systems :

:

sensors that can be linked to form neural network

sensors that can be linked to form neural network

The article generally relates to smart homes/workspaces/etc. The aim is to deal with common challenges in existing solutions ( https://www.cleverism.com/smart-home-intelligent-home-automation/) and provide means to easily extend and orchestrate various sensors in order to provide more and more sophisticated capabilities. The idea is build a system using layers of sensors, each layer provides more sophisticated cognitive capabilities, starting from layer 2 sensors may have embedded learning capabilities/track history and adopt to individual

For example:

Layer 1:
- Light sensors
- Temperature sensor
- Noise sensor
- Humidity sensor
- Motion Senors
- Door sensors
- IoT sensors
- Cameras

Layer 2: (Perceived values): - Is it relatively dark - take into account input from: - Light sensors - Clock - Calendar - Is it relatively loud - take into account input from: - Noise sensors - Clock - Calendar - Is cooking in progress - take into account
- Light sensors (kitchen)

- IoT sensors (cooker/fridge/etc)

- Motion sensors

Layer N:
- Is it comfortable - take into account
Perceived darkness, loudness, activity, etc
The core ideas:
1) Extending the capabilities of a sensor to let it act a node in neural network
2) Organization of sensors:
While each sensor may act more or less independently or be a part of some integrated system (as in currently existing solution) it will also have a means (see 1) to take input(s) from other sensors, apply some function on the input and pass the results to one or more outputs, effectively acting as a node in neural network

The network can then be trained and over the time recommendation from the network may take priority over the built-in logic in individual sensors, what gives the

1


Page 02 of 3

user the learning capabilities of neural network without suffering the pain of training process

Related documents:

Intelligent home monit...