AI Bibliography
AI Bibliography |
|
 |
 |
 |
|
Munir, M. S., Abedin, S. F., & Hong, C. S. 2019, Artificial intelligence-based service aggregation for mobile-agent in edge computing. Paper presented at 2019 20th Asia-Pacific Network Operations and Management Symposium (APNOMS). |
| Resource type: Proceedings Article BibTeX citation key: Munir2019 View all bibliographic details  |
Categories: Artificial Intelligence, Computer Science, Data Sciences, Decision Theory, Engineering, General, Innovation, Military Science Subcategories: Big data, Cloud computing, Command and control, Cross-domain deterrence, Cyber, Edge AI, Informatics, Internet of things, JADC2, Mosaic warfare, Networked forces, Neural nets Creators: Abedin, Hong, Munir Publisher: Collection: 2019 20th Asia-Pacific Network Operations and Management Symposium (APNOMS) |
| Attachments |
| Abstract |
|
The ongoing development of edge computing in fifth-generation (5G) networks promises to provide an artificial intelligence-as-a-service (AIaaS) for meeting the stringent requirements of everything as a service (XaaS) in the edge of the networks. Therefore, the concept of edge-artificial intelligence (edge-AI) is not only evolving but also emergent enabler toward AI service fulfillment. In this paper, we investigate an AI-based service aggregation problem for a mobile agent in AIaaS-enabled edge computing. First, we propose an optimization problem for the mobile agent and the objective is to maximize the AI service fulfillment achieved rate while satisfying the computational, memory, and delay requirements. Thus, we show that this optimization problem is NP-hard. Second, we compel the formulated problem in a community discovery problem and derive a solution by executing a data-driven approach. To do this, we incorporate density-based spatial clustering of applications with noise (DBSCAN) and flow control algorithm, and propose a low computational complexity algorithm for AI service aggregation of the mobile agent. Finally, numerical analysis shows the proposed model can perform better over other baseline methods in terms of deprived AI services, server utilization, and complexity analysis.
|