Granted Patent B2

US 12,452,265 · App. 18/510,105 · Granted Oct 21, 2025

Inventors: Ki Ryong Kwon (Busan, KR); Mamunur Rashid Md (Busan, KR)

Assignee: PUKYONG NATIONAL UNIVERSITY INDUSTRY-UNIVERSITY COOPERATION FOUNDATION

H04L63/1416G06N3/045G06N3/08

Abstract

Claims (23)

1. An intrusion detection system for an Internet of Things (IoT) network, comprising:

a plurality of local clients, each configured to be interconnected with a corresponding one of a plurality of IoT device groups to establish an individual IoT network; and

a central server configured to be interconnected with the plurality of local clients and implemented on a blockchain network,

wherein at a beginning of operation, the central server is configured to:

generate an initial global neural network model by using data held by the central server, and

transmit the generated global neural network model to each of the plurality of local clients,

wherein each of the plurality of local clients is configured to:

input data collected from a IoT device group interconnected with each of the plurality of local clients into the initial global neural network model to train the initial global neural network model, and

generate an initial local neural network model, and

wherein each of the plurality of local clients includes a local neural network model for intrusion detection, and is further configured to update a current local neural network model based on an updated global neural network model transmitted from the central server.

2. The intrusion detection system of claim 1 , wherein each of the plurality of local clients trains the current local neural network model by using data collected from the IoT device group interconnected with each of the plurality of local clients as input, and transmits learning content to the central server.

3. The intrusion detection system of claim 1 , wherein the central server generates a global neural network model updated by combining learning content transmitted from each of the plurality of local clients, and transmits the updated global neural network model to each of the plurality of local clients.

4. The intrusion detection system of claim 1 , wherein the central server includes a plurality of nodes implemented on the blockchain network, and the plurality of nodes generate an updated global neural network model based on learning content transmitted from a local client among the plurality of local clients and learning content transmitted from other nodes.

5. An operating method of an intrusion detection system for an Internet of Things (IoT) network, the system detecting an intrusion into an IoT network established by interconnecting each of a plurality of IoT device groups with a corresponding one of a plurality of local clients, each of the plurality of local clients including a local neural network model for intrusion detection, the method comprising:

generating, by a central server implemented on a blockchain network, an initial global neural network model by using data stored by the central server at a beginning of operation;

transmitting, by the central server, the generated initial global neural network model to each of the plurality of local clients;

inputting, by each of the plurality of local clients, data collected from a IoT device group interconnected with each of the plurality of local clients into the initial global neural network model to train the initial global neural network model;

generating, by each of the plurality of local clients, an initial local neural network model;

training, by each of the plurality of local clients, a current local neural network model by using data collected from the IoT device group interconnected with each of the plurality of local clients as input and transmitting learning content to the central server;

generating, by the central server, a global neural network model updated by combining the learning content transmitted from each of the plurality of local clients, and transmitting the generated global neural network model to each of the plurality of local clients; and

updating, by each of the plurality of local clients, the current local neural network model based on the updated global neural network model.

6. The operating method of claim 5 , wherein the central server includes a plurality of nodes implemented on the blockchain network, and

wherein the generating, by the central server, the updated global neural network model includes generating, by the plurality of nodes, the global neural network model updated by combining the learning content transmitted from a local client among the plurality of local clients and learning content transmitted from other nodes.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (23)