居家長照AI物聯網之DDS通訊網路開發和佈署平台;Development and Deployment Platform of DDS Communication Network for AIoT in Home-based Long-term Care

NCU Institutional Repository > 資訊電機學院 > 資訊工程研究所 > 博碩士論文 > Item 987654321/98293

請使用永久網址來引用或連結此文件: https://ir.lib.ncu.edu.tw/handle/987654321/98293

題名:	居家長照AI物聯網之DDS通訊網路開發和佈署平台;Development and Deployment Platform of DDS Communication Network for AIoT in Home-based Long-term Care
作者:	彭至偉;Peng, Chih-Wei
貢獻者:	資訊工程學系
關鍵詞:	物聯網;閘道器;居家長照;DDS;ROS 2;IoT;Gateway;Home-based Long-term Care;DDS;ROS 2
日期:	2025-07-21
上傳時間:	2025-10-17 12:35:50 (UTC+8)
出版者:	國立中央大學
摘要:	隨著物聯網（IoT）設備數量的迅速增加，異質設備之間的資料傳輸與管理逐漸面臨通訊延遲、系統整合困難以及資料格式不一致等挑戰。現今主流的輕量級通訊協定如 MQTT 與 CoAP 雖具備低功耗與簡易開發等優點，但在處理多模態資料、高頻寬需求與即時性應用方面仍存在明顯限制。有鑑於此，本研究提出一套以 DDS（Data Distribution Service）為基礎的通訊閘道器系統，透過其具備的即時性、去中心化與高可靠性特性，作為異質設備整合與資料單一化處理的核心解決方案。本研究應用 MIAT 系統設計方法論，透過 IDEF0 模組化建模與 GRAFCET 離散事件建模設計整體系統架構。系統以 Docker 為容器環境，將影像與音訊資料流分別導入邊緣推論模組，分別採用 YOLOv8 與 YAMNet 進行即時辨識與事件判定，並由系統閘道器統一處理與回應。為提升系統擴充性與部署效率，本論文亦提出單一化的設備配置流程，能根據不同設備自動產生並套用 DDS 通訊參數與 QoS 設定。透過實際模擬居家照護情境進行多模態設備測試，實驗結果顯示本研究所開發之 DDS 閘道器具備良好的即時資料處理能力與系統穩定性，能有效降低系統建置時間與整合成本，提供一套具備彈性與單一化的開發架構，對於未來智慧感測應用場域具高度實用性與可擴展性 ;The rapid proliferation of Internet of Things (IoT) devices has introduced challenges in managing heterogeneous data streams, ensuring low-latency communication, and achieving flexible system integration. While lightweight protocols like MQTT and CoAP dominate current IoT ecosystems, they often fall short in reliability, scalability, and multimodal data handling. To address these limitations, this study proposes a gateway system based on the Data Distribution Service (DDS) communication protocol. DDS offers real-time, decentralized data transmission and is the backbone of modern ROS 2 and Micro-ROS systems. In this research, we design and implement a DDS-based multimodal gateway, capable of standardizing data from various sensors and automating network configuration across heterogeneous microcontroller units (MCUs). By applying the MIAT system design methodology, the gateway is modularized using IDEF0 and GRAFCET frameworks to support scalability, resilience, and efficient data processing. Each subsystem, including image and audio data streams, is independently handled via Dockerized services and edge inference using models such as YOLOv8 and YAMNet. Experimental validation demonstrates the gateway’s ability to process and distribute sensor data in real-time within a smart home simulation, enabling timely alerts through a cloud-integrated notification system. Results confirm the gateway’s effectiveness in simplifying deployment, reducing system configuration time, and enhancing multimodal data processing performance in resource-constrained environments. This architecture establishes a foundation for scalable DDS-based IoT systems and offers a template for future development in smart sensing applications.
顯示於類別:	[資訊工程研究所] 博碩士論文

文件中的檔案:

檔案	描述	大小	格式	瀏覽次數
index.html		0Kb	HTML	6	檢視/開啟

在NCUIR中所有的資料項目都受到原著作權保護.

社群 sharing

資料載入中.....