A Novel Software Framework for Optimizing Edge AI Systems: Overcoming Computational and Deployment Challenges
- سال انتشار: 1403
- محل انتشار: اولین کنفرانس بین المللی برق، مکانیک، فناوری اطلاعات و کامپیوتر در علوم مهندسی
- کد COI اختصاصی: EMICWCONF01_015
- زبان مقاله: انگلیسی
- تعداد مشاهده: 74
نویسندگان
Master of Science, Department of Electrical and Computer Engineering Islamic Azad University of Yadegar-e Imam, Tehran, Iran
چکیده
The rapid adoption of Edge Artificial Intelligence (Edge AI) in autonomous systems, smart cities, and IoT applications has introduced significant software engineering challenges related to computational efficiency, real-time processing, and scalable deployment. Traditional AI deployment models struggle with the resource constraints, energy efficiency, and heterogeneity of edge devices, limiting their ability to perform high-performance inference in real-world environments. In this research, we propose a novel software framework designed to optimize Edge AI model execution, adaptive deployment, and resource allocation across diverse hardware architectures. Our approach leverages lightweight AI model transformations, dynamic resource scheduling, and federated software updates to enhance the efficiency and scalability of AI-powered edge systems. We develop and test our framework on heterogeneous edge environments, demonstrating a ۳۰-۵۰% reduction in inference latency and power consumption compared to existing deployment techniques. Additionally, we introduce an adaptive learning mechanism that enables efficient model updates without requiring cloud dependencies. This research provides a scalable and secure software foundation for next-generation Edge AI applications, addressing critical limitations in computational overhead, real-time adaptability, and security.کلیدواژه ها
Edge AI, Software Optimization, Model Deployment, Resource Allocation, Federated Learningاطلاعات بیشتر در مورد COI
COI مخفف عبارت CIVILICA Object Identifier به معنی شناسه سیویلیکا برای اسناد است. COI کدی است که مطابق محل انتشار، به مقالات کنفرانسها و ژورنالهای داخل کشور به هنگام نمایه سازی بر روی پایگاه استنادی سیویلیکا اختصاص می یابد.
کد COI به مفهوم کد ملی اسناد نمایه شده در سیویلیکا است و کدی یکتا و ثابت است و به همین دلیل همواره قابلیت استناد و پیگیری دارد.