报告详情
Deep Reinforcement Learning-Driven Dynamic Spectrum Access for AI-Enabled Cognitive Radio Networks
编号：117
访问权限：仅限参会人
更新：2026-11-10 16:48:59 浏览：9次
Online
报告开始：2026年11月21日 12:15（Asia/Kolkata）
报告时间：15min
所在会场：[S2] Internet of Things and Intelligent Sensing Systems [S2-1] Internet of Things and Intelligent Sensing Systems
摘要
Widespread access to connectivity takes its toll on wireless spectrum and the ever-growing demand as well potential spectral inefficiencies, which are exasperated under static allocation paradigms. In this paper, we investigate a new framework of Deep Reinforcement Learning Based Cognitive Radio (DRL-CR) which combines A3C agent with CNN based spectrum feature extraction and enables dynamic and autonomous access to the cordial space. In contrast to traditional energy-detection or single-agent DQN scenarios, the proposed system allows a secondary-user (SU) agent to develop an optimal policy for channel selection and power adaptation directly from raw spectrum inputs, without requiring any explicit knowledge of the radio environment or PU activity statistics [1]. In A3C, these states are explored in parallel actor-learner threads with the goal of launching a myriad of state exploration at any one time to find converging policy faster and reduce gradient variance. Through MATLAB-based simulations in a ten-channel wideband cognitive radio scenario, we show that the proposed DRL-CR obtains 42.6 Mbps throughput with corresponding gains of 35.2% over DQN-based baselines, and attains spectrum detection probability of 0.96, along with false alarm rates as low as 0.03. Meanwhile, the framework features 89% average spectrum utilization and reaches 4.7 bits/Joule energy efficiency, confirming it is ready for deployment in both 5G heterogeneous networks, IoT ecosystems and spectrum-scarce emergency communication scenarios alike.
关键词
Cognitive Radio, Deep Reinforcement Learning, Spectrum Sensing, Dynamic Spectrum Access, A3C, Convolutional Neural Networks, 5G, Interference Management.
报告人
Karthik M
ASSISTANT PROFESSOR K.Ramakrishnan College of Engineering； Trichy稿件作者
全部评论
重要日期
-
会议日期
11月20日
2026
至11月22日
2026
-
10月31日 2026
初稿截稿日期
-
11月07日 2026
提前注册日期
-
11月22日 2026
注册截止日期
主办单位
United Societies of Science
承办单位
Pondicherry University
协办单位
IEEE Section
IEEE Madras Section
IEEE Madras Section
联系方式
- ICon INDIA 2026
- ic******@usssociety.org
发表评论