A Network Access Control Algorithm Integrating User Preferences and Network States

• Liang Gen,
• Su Fengyi,
• Lu Baiyang,
• Yu Hewei,

Vol. 19, No. 10, October 31, 2025
10.3837/tiis.2025.10.019, Download Paper (Free):

• Double Deep Q-network
• Deep Reinforcement Learning
• Network Access Control
• Ultra-dense heterogeneous wireless networks (UDHWNs)
• User Preference

Abstract

In ultra-dense heterogeneous wireless networks (UDHWNs), the large-scale access of Internet of Things (IoT) devices, diverse service requirements, and system heterogeneity pose significant challenges to traditional access control algorithms in maintaining efficient and stable performance in dynamic environments. To overcome these limitations, this paper proposes a network access control algorithm that integrates user preferences with network states. The algorithm first quantifies subjective user preferences using the Analytic Hierarchy Process (AHP) and combines them with objective network state characteristics evaluated through the standard deviation method to derive comprehensive weights. Then, leveraging the Double Deep Q-Network (DDQN) within a reinforcement learning framework, the algorithm autonomously optimizes access decisions by decoupling action selection from value estimation, effectively mitigating the Q-value overestimation problem inherent in traditional Deep Q-Networks (DQN). By constructing a well-designed network state space and reward function, the access control problem is formulated as a Markov Decision Process (MDP). Simulation results demonstrate that the proposed algorithm not only satisfies differentiated Quality of Service (QoS) requirements but also significantly reduces the total number of handovers and enhances network access stability.

Statistics

Cite this article