Ghosh helps develop an intelligent resource management technique for 5G and beyond communication networks

5G and beyond technology has great potential to positively impact smart health care networks, the Internet of Things for disaster management, and other areas of health care. However, the dynamic resource management of these complex networks must be improved to achieve those benefits.

Dr. Uttam Ghosh, associate professor of cybersecurity, and colleagues have proposed a solution using a multi-layered SFC (Service Function Chain) formation for adaptive VNF (Virtual Network Function) allocation on dynamic slices. The team has formulated an ILP to address the popular VNF-EAP (VNF-Embedding and Allocation Problem) over real network topology (AT&T Topology).
Their work builds on intelligent research using machine learning techniques for VNF selections. Further, they have also studied a VNF typecasting technique for service backup on outage slices in the field of disaster management.
“ML is a strong tool to improve the quality of service and the selection accuracy of the VNF instances, irrespective of the service types,” says Dr. Ghosh.
He adds that researchers have proposed methods like PDRA (Priority-Based Dynamic Resource Allocation) Scheme, CMDP (Constrained Markov Decision Process), and Agent-based Resource Allocation to solve problems on Linear Programming (LP), ILP, MILP and a few heuristics of VNF-EAP. However, they have not developed a suitable method to overcome issues that include service accuracy and resource outage together.
Dr. Ghosh and colleagues Mr. Deborsi Basu, Mr. Soumyadeep Kal and Prof. Raja Datta from the Indian Institute of Technology in Kharagpur, India, have proposed Dynamic Resource Introspection and VNF Embedding for 5G using machine learning or DRIVE.
“We have proposed two novel algorithms, namely VNF-READ Algorithm (VNF Re-allocation, VNF Evoke, VNF Allocation, VNF Delete or Detention) and SFC-DRIVE using multi-graph layering,” says Dr. Ghosh. “Then, we applied them on the real network topology of AT&T North America for various networking conditions.”
Dr. Ghosh and his team followed support vector regression, kernel ridge regression, and other machine learning algorithms to help classify multiple VNF classes at once. Their approach projects various commands for the type-specific applications that makes the system more accurate.
“We performed the experiments on several distinct networking conditions to enhance the network performance,” says Dr. Ghosh.
The authors intend to continue to improve their proposed model.
“One issue we hope to address is that automated SFC for starving processes are resource hungry. Especially during an outage, the existing resources can work as a backup,” says Dr. Ghosh. “Further research might overcome these difficulties, and an optimal FL-based algorithm with advanced security learning features that improve future network reliability. We are working on it.“
“A rapid paradigm shift has been observed in the domain of Edge-Cloud coupling in modern health care system, where intelligent resource provisioning becomes a must. Health care is an extremely sensitive area where margin of error is almost nil. AI-driven techniques may further help in VNF selections or even in existing VNF allocations. Our team is also developing solutions in this direction,” Dr. Ghosh stated.
This work partially funded from his recent NSF grant under award number 2219741.
The full paper, DRIVE: Dynamic Resource Introspection and VNF Embedding for 5G using Machine Learning, is available online.