AI-Driven Dynamic Bandwidth Allocation for Smart IoT Environments Using Reinforcement Learning

Abstract

As the Internet of Things (IoT) expands rapidly, the main issue is now limited bandwidth on computer networks. Since many connected devices are coming online, the ways shared bandwidth is managed still fail to satisfy the variety and constant needs of IoT systems. Such approaches raise the threat of resource overuse, network overcrowding and poor performance when used in mobile or swiftly changing situations. Devices in the IoT communicate in many ways, some needing fast speed for sensors and others requiring big bandwidth for video. Therefore, static allocation strategies do not adjust to new requirements in demand. To solve this, this study develops a strategy for bandwidth allocation that makes use of Deep Q-Networks within the field of Reinforcement Learning. Thanks to the RL model, the system can quickly change the amount of bandwidth needed for optimal operation and top service quality in IoT applications. Thanks to this operation, the IoT’s capacity grows without increasing delay. Tests using RL technology demonstrate that it improves the system’s performance more than older approaches. The proposed method will help IoT networks to handle increased usage and use bandwidth effectively. It makes future reachable for the IoT network optimization, specifically by handling the increased need for connection and data sharing in different IoT solutions.