The DDoS attacks are the most common attacks of the network and the mechanism and nature of these attacks change day by day. DDoS attacks can simply exhaust the network communication resources, it cause the failure of the network within a less time. The fundamental technologies and network protocols have flaws and vulnerabilities that can open doors for attackers to attack the network. This problem is a hurdle for secure and reliable communication with in network. To overcome this issue, there is a need to create a more effective and accurate attack detection algorithm specifically for DDoS attack. Numerous attack detection mechanisms already developed to detect a DDoS attack in network. An effective attack detection mechanism that can provide the entire attack details is therefore required to overcome the problem. Additionally, need to develop such a diverse system to address not only DDoS attack details and also provide proactive measures to avoid the same kind of attacks in future by designing intelligent system. Therefore, in this thesis, we propose SRDD, a smart routing based DDoS detection system for entire network. In SRDD, the DDoS attacks can be detected at the initial phase. In this research, we have been developed a DDoS detection system which is based on decision tree algorithm to prevent the DDoS attack. The decision tree algorithm, include with signature based detection methodologies, perform automatic prediction, and provide effective detection of malicious traffic. The decision tree algorithm classifying the received packets and making a decision based on the classification of results. For validation of proposed system, selected some of other machine learning techniques and provide comparison with our proposed system. This proposed SRDD system evaluates the network resources and the traffic dataset in order to train the centric agent for detection system to detect legitimate and malicious traf?c. The proposed SRDD system allowed legitimate traffic to pass through network and malicious traf?c is ?agged malicious to go through detection system. Finally, the results are discussed in terms of accuracy, exposure and specificity. The system not only can identify the attacks, but also detect the attacker IPs and start a process of mitigation to provide very well protection of the network system at initial stage of an attack is identified. The proposed research focused on the security aspect of legacy protocols as the baseline for mapping our desired research objectives to align with intelligent survivable centric network agent.