Title: Artificial Intelligence Across Disciplines: Research, Innovation, and Intelligent Solutions
Editors: Dr. Subita Bhagat, Dr. A. Balamurugan, Dr. P. Krishna Kumar, and Mrs. S. Nandhini Devi
ISBN: 978-81-69857-83-3
Chapter: 20
DOI: https://doi.org/10.59646/815/20
Author: Dr. Kavipriya
Abstract
The rapid expansion of distributed cyber-physical systems has catalyzed the transition from conventional, centrally managed machine-to-machine (M2M) communications toward autonomous digital ecosystems. These contemporary environments require dynamic, decentralized decision-making where heterogeneous devices negotiate computational, spectrum, and energy resources without continuous human intervention. This chapter presents an artificial intelligence-driven framework based on Multi-Agent Deep Reinforcement Learning (MADRL) operating under a Centralized Training with Decentralized Execution (CTDE) architecture. Focusing on an industrial IoT smart manufacturing case study containing autonomous agents, we evaluate dynamic radio channel allocation, edge computation offloading, and collaborative transmission scheduling. The proposed framework implements a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm augmented by attention-driven inter-agent state aggregation. Empirical results demonstrate an 84.7% reduction in channel contention collisions, a 43.8% decrease in end-to-end task execution latency, and an overall energy efficiency improvement of 39.4% over legacy decentralized protocols, establishing a scalable foundation for next-generation self-governing autonomous digital infrastructures.
Keywords: Machine-to-Machine (M2M) Intelligence, Autonomous Digital Ecosystems, Multi-Agent Deep Reinforcement Learning, Centralized Training Decentralized Execution, Edge Computing Task Offloading, Dynamic Spectrum Allocation.