Heterogeneous network resource allocation algorithm based on deep reinforcement learning

Although heterogeneous networks reduce network operating costs, increase network coverage and system capacity, improve user service quality and enhance user service experience, heterogeneous networks will make limited communication resources even more stretched. Moreover, it will also lead to problems such as increased interference and increased equipment energy consumption. Deep reinforcement learning methods can dynamically interact with the environment through intelligent agents, which provides an effective way to solve complex optimization problems in heterogeneous network resource management. However, traditional deep reinforcement learning methods are singleagent algorithms and cannot handle the unstable environment problems of multiple agents in heterogeneous network scenarios. Therefore, this paper optimizes the user mode selection and channel allocation of heterogeneous networks. With the goal of minimizing the user information transmission delay and energy consumption, the user mode selection and channel allocation optimization problem is first modeled. Then, in order to construct the optimization problem of maximizing the total system rate of heterogeneous networks under user QoS constraints, we proposed a resource allocation algorithm based on deep reinforcement learning, which can significantly reduce network delay and energy consumption and achieve the optimal balance between energy consumption and user delay. In addition, we introduced a priority-based allocation strategy that combines methods such as Q-learning, experience replay, and deep neural networks. It can effectively reduce user delay and achieve optimal configuration of heterogeneous network resources. Finally, the paper conducts simulation experiments on the proposed heterogeneous network resource allocation algorithm. The results show that compared with other algorithms, our proposed algorithm can not only achieve better convergence performance and better resource allocation efficiency, but also significantly reduce network latency and energy consumption, and achieve better system performance.