Reliable information delivery and dynamic link utilization in MANET cloud using deep reinforcement learning

• Published in: Transactions on emerging telecommunications technologies Vol. 35; no. 9
• Main Authors: Kuang, Shuhong, Zhang, Jiyong, Mohajer, Amin
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.09.2024
• ISSN: 2161-3915; 2161-3915
• DOI: 10.1002/ett.5028

Modern networking demands efficient and reliable information delivery within Mobile Ad‐hoc Network (MANET) and cloud environments. This paper introduces a novel approach that employs Multi‐Agent Deep Learning (MADL) for adaptive resource allocation, addressing the challenges of optimizing traffic and ensuring dependable information delivery while adhering to Service Level Agreement (SLA) constraints. Our method dynamically allocates resources across nodes, leveraging the synergy between Advanced Cloud Computing and Edge Computing to balance centralized processing and localized adaptability. The integration of Graph Neural Networks (GNNs) further enhances this process by adapting resource allocation decisions based on network topology. Through iterative learning, our algorithm fine‐tunes continuous‐time resource optimization policies, resulting in substantial improvements in throughput and latency minimization. Simulations validate the effectiveness of our approach, demonstrating its potential to contribute to the advancement of MANET cloud networks by offering adaptability, efficiency, and real‐time optimization for reliable information delivery and dynamic link utilization. This paper introduces a novel approach that employs Multi‐Agent Deep Learning (MADL) for adaptive resource allocation, addressing the challenges of optimizing traffic and ensuring dependable information delivery while adhering to Service Level Agreement (SLA) constraints.