Analyzing the Heterogeneous Edge AI systems to facilitate the profiling of AI models in achieving efficient computation offloading
Analyzing the Heterogeneous Edge AI systems to facilitate the profiling of AI models in achieving efficient computation offloading
Edge AI leverages computation offloading to overcome the resource limitations of user devices. However, traditional methods often assume homogeneous infrastructure and neglect the runtime characteristics of AI models, which becomes a critical challenge in heterogeneous edge environments. This paper presents a comprehensive literature review on computation offloading strategies in edge systems and introduces the concept of profiling AI models to enable efficient offloading decisions. Through comparative analysis, we highlight how profiling — capturing parameters such as model type, hyperparameters, hardware specifications, and dataset characteristics — can dramatically improve resource utilization, energy efficiency, and latency. We propose profiling-based approaches informed by prior modelling and optimization techniques to enhance adaptivity in dynamic heterogeneous edge scenarios. A new framework is outlined to guide future research, emphasizing accurate prediction of resource consumption and task completion times. The key insights reaffirm that profiling augments offloading strategies beyond rule-based or optimization-only methods, paving the way toward more responsive and efficient edge AI systems.
"Analyzing the Heterogeneous Edge AI systems to facilitate the profiling of AI models in achieving efficient computation offloading", IJSDR - International Journal of Scientific Development and Research (www.IJSDR.org), ISSN:2455-2631, Vol.8, Issue 9, page no.1351-1356, September-2023, Available :https://ijsdr.org/papers/IJSDR2309191.pdf
Volume 8
Issue 9,
September-2023
Pages : 1351-1356
Paper Reg. ID: IJSDR_304942
Published Paper Id: IJSDR2309191
Downloads: 00049
Research Area: Science and Technology
Country: -, -, India
ISSN: 2455-2631 | IMPACT FACTOR: 9.15 Calculated By Google Scholar | ESTD YEAR: 2016
An International Scholarly Open Access Journal, Peer-Reviewed, Refereed Journal Impact Factor 9.15 Calculate by Google Scholar and Semantic Scholar | AI-Powered Research Tool, Multidisciplinary, Monthly, Multilanguage Journal Indexing in All Major Database & Metadata, Citation Generator
Publisher: IJSDR(IJ Publication) Janvi Wave
