Energy-Efficient Resource Allocation in Wireless Networks Using Federated-Double-Deep Q-Network

Now a days, the resource allocation manages the resource networks such as bandwidth, power usage for ensuring Quality of Service (QoS) without overloading. However, the optimization fails to adapt traffic loads due to its dependance on static resource allocation that cannot adjust to user demands and power control. To overcome this limitation in this research resource allocation method Federated-Double-Deep Q-Network (F-DDQN) is proposed for maximizing throughput and achieve minimum QoS requirements for all active Device-to-Device communication(D2D) pairs and Cellular Users (CUE). whereas Federated Learning enhances adaptability to dynamically communication environments. Initially, the data is collected from D2D-enabled heterogeneous cellular networks (HetNet) dataset and 3-tier HetNet system model are designed. Further, for resource management the F-DDQN enhances the energy-efficient by decentralizing action states and minimizes communication overhead and power consumption by sharing model updates instead of raw data, and DDQN effectively improves the energy efficiency in resource allocation by estimating Q-values and optimizes resource allocation through selecting the most energy-efficient transmitting power and channels using target estimation. From the results the proposed model achieved better results to evaluate its effectiveness in increasing system throughput with 58.90% compared with both CUE and D2D Users Equipment (DUE) with existing method a Deep Q-network-based Reinforcement Learning algorithm (DRL).