Networking technologies have given rise to worldwide social, business, and military networks, and commercial networks in US has been growing explosively. Cyber-attacks are increasing in frequency, impact, and complexity, which demonstrate extensive network vulnerabilities with the potential for serious damage. To defend against cyber-attacks, we propose a Network Sensor-Based Defense Framework for Active Network Security Situation Awareness and Impact Mitigation, with the aim of handling network security awareness, mitigation, and prediction. In particular, we will conduct the following research tasks: (i) develop a generic framework for integrating data from various sources, (ii) Develop information theory, image, and signal processing techniques to efficiently process alerts information and provide accurate detection decision; (iii) Develop novel techniques to carry out attack scene investigation via network forensic analysis and visualization; (iv) Develop the game theory to investigate the interaction between the attacker and defender with different strategies; and (v) implement a prototype system and validate our proposed solutions. Our work will have immediate and significant impacts on the security of U.S. Military and commercial networks. The proposed security SA framework, algorithms and toolkits are well sought by military and commercial cyber defense.