The invention discloses a network architecture of an artificial-like network and an implementation method, and relates to the field of network architecture. The network architecture herein includes a user layer, a network layer, a control layer and an application layer. The user layer is intended for providing information to the control layer. The network layer is intended for transmitting and receiving data, executing a command from the control layer, sensing network state, environment information and knowledge of other networks, and handling the network state, the environment information and the knowledge of other networks to the control layer. The control layer is intended for acquiring and processing the information from the application layer, the user layer and the network layer, and saving the processed information and the knowledge of the other networks, and the control layer is also intended for generating knowledge associated with the network state and decisions, and adding the knowledge associated with the network state and decisions to a database. A control entity is intended for based on the current network state, obtaining decisions, performing decision-making rehearsal by simulation, estimating optimal time for implementing the decisions, implementing the decisions, evaluating the decisions and feedbacking the decisions. The application layer is intended for receiving the information provided by the control layer and providing required information to the control layer. According to the invention, the network architecture herein can increase the robustness of the network, enhances defense against network attacks, and provide quality functions and performances.