Future wireless and Internet paradigm are envisioned to be more intelligent, adaptive, recovering from transient faults and problems without service interruption, and contineously optimizing the use of resources. In this work, we present agent based network control and management framework that enables domain based dynamic policy formulation. The article discusses the proposed architecture that adapts the hierarchical interactions in the typical telecommunication chain in defining different types of cognitive agents running on specific network elements for realizing the end-to-end goals. We also discuss the communication model, different behaviors of the proposed framework, the developed software tools, and the experimental demonstrator that we developed for realizing this framework. We believe that the proposed framework can meet the requirements of adaptability, self-x functionalities, sustainability, etc. The paper also provide details on how the proposed framework may be applied for a representative case-study. The performance of the proposed approach is evaluated for different criteria and results are discussed.