Large Language Model-Integrated Human Activity Recognition System Using Channel State Information (채널 상태 정보를 활용한 대규모 언어 모델 통합형 인간 행동 인식 시스템 연구)

Hye-Jin Park (박혜진); Inhyeok Kang (강인혁); Sol-Bee Lee (이솔비); Jung-Hyok Kwon (권정혁); Eui-Jik Kim (김의직)

doi:10.20465/KIOTS.2025.11.6.101

Large Language Model-Integrated Human Activity Recognition System Using Channel State Information

Journal of Internet of Things and Convergence
Abbr : JKIOTS
2025, 11(6), pp.109~115
DOI : 10.20465/KIOTS.2025.11.6.101
Publisher : The Korea Internet of Things Society
Research Area : Engineering > Computer Science > Internet Information Processing
Received : November 5, 2025
Accepted : December 16, 2025
Published : December 31, 2025

Hye-Jin Park ¹, Inhyeok Kang ², Sol-Bee Lee ², Jung-Hyok Kwon ², Eui-Jik Kim ²

¹한림대학교 소프트웨어학부
²한림대학교

Accredited

ABSTRACT

This paper presents a Human Activity Recognition(HAR) system integrated with a Large Language Model(LLM) using Channel State Information(CSI) in Internet of Things(IoT) environments. The main goal of the proposed system is to infer user activity through HAR analysis using CSI collected from IoT devices and generate personalized feedback using LLM. The proposed system constructs prompts based on the HAR results and the duration of the corresponding activities. During this process, the current activity state, task-specific requirements, and desired feedback length are considered to ensure the accuracy and consistency of the LLM outputs. The generated prompt is processed by the LLM-based feedback engine and converted into natural-language feedback appropriate to the user's activity. This feedback is then provided in real time through a web interface. To evaluate the feasibility of the proposed system, we conducted an experimental implementation. The results demonstrated the reliable operation of the entire process, from CSI-based activity recognition to LLM-based feedback generation.

KEYWORDS

Large Language Model, Internet of Things, Personalized Feedback, Human Activity, Recognition, Channel State Information

Journal Copyright Policy

No CCL information provided

Citation status

* References for papers published after 2025 are currently being built.

[journal] M.Kaseris / 2024 / A comprehensive survey on deep learning methods in human activity recognition / Mach. Learn. Knowl. Extr. 6(2) : 842~876

[journal] F.Serpush / 2022 / Wearable Sensor-Based Human Activity Recognition in the Smart Healthcare System / Comput. Intell. Neurosci. 2022(24) : 1~12

[journal] P.F.Moshiri / 2021 / A CSI-Based Human Activity Recognition Using Deep Learning / Sensors 21(21) : 7225~

[journal] A.Pantelopoulos / 2010 / A Survey on Wearable Sensor-Based Systems for Health Monitoring and Prognosis / IEEE Trans. Syst. Man Cybern. Part C-Appl. Rev. 40(1) : 1~12

[journal] Z.Wang / 2018 / Wi-Fi CSI-Based Behavior Recognition: From Signals and Actions to Activities / IEEE Commun. Mag. 56(5) : 109~115

[journal] A.Zhuravchak / 2022 / Human Activity Recognition based on Wi-Fi CSI Data -A Deep Neural Network Approach / Procedia Comput. Sci. 198 : 59~66

[journal] S.Yousefi / 2017 / A Survey on Behavior Recognition Using WiFi Channel State Information / IEEE Commun. Mag. 55(10) : 98~104

[journal] R.Alazrai / 2020 / An End-to-End Deep Learning Framework for Recognizing Human-to-Human Interactions Using Wi-Fi Signals / IEEE Access 8 : 197695~197710

[journal] M.N.Shoumi / 2024 / Leveraging the Large Language Model for Activity Recognition: A Comprehensive Review / Int. J. Act. Behav. Comput. 2024(2) : 1~27

[other] N.Kashyap / 2023 / Towards Enhanced Human Activity Recognition through Natural Language Generation and Pose Estimation

[other] M.Dong / 2024 / Can LLMs Serve As Time Series Anomaly Detectors?

[other] T.An / 2025 / IoT-LLM: Enhancing Real-World IoT Task Reasoning with Large Language Models

[other] Z.Li / 2025 / SensorLLM: Aligning Large Language Models with Motion Sensors for Human Activity Recognition

[web] / Nexmon / https://nexmon.org

[web] / Teach, Learn, and Make with Raspberry Pi Foundation / https://www.raspberrypi.org

This paper was written with support from the National Research Foundation of Korea.

KJCKorea
Journal Central

Journal of Internet of Things and Convergence 2025 KCI Impact Factor : 0.75

Large Language Model-Integrated Human Activity Recognition System Using Channel State Information

ABSTRACT

KEYWORDS

Journal Copyright Policy

Citation status

* References for papers published after 2025 are currently being built.

Journal of Internet of Things and Convergence 2025 KCI Impact Factor : 0.75

Large Language Model-Integrated Human Activity Recognition System Using Channel State Information

ABSTRACT

KEYWORDS

Journal Copyright Policy

Statistics

Tools

Issue List

Citation status

KCI Citation Counts (0)

REFERENCES (15) * References for papers published after 2025 are currently being built.

Search PDF

Citation

* References for papers published after 2025 are currently being built.