The increase in complexity of network management systems and a consequent lack of association to business requirements has driven the need for autonomic communications. By integrating context information, autonomic computing can provide more efficient means to counter technical problems found in complex network systems and at the same time address associated business requirements. In this paper, we propose an autonomic communications architecture that manages complexity through policy-based management where we incorporate a shared information model integrated with knowledge-based reasoning mechanisms to provide selfgovernaning behavior.