Detection of Denial of Service by access pattern assessment

Denial of Service (DoS) attacks are increasing in popularity due to the fact that they are much more difficult to detect as well as mitigate compared to other types of attacks. In this paper, we propose, a step by step verification mechanism to identify legitimate users as well as attack traffic. The mechanisms employed here include random generation of interactive content as well as corresponding encoded solution sent to the client when a service is requested. This avoids further connections to database to verify if the solutions to the interactive content are right or not. Further measures such as dual timers and termination of connections are done if pre-set period of inactivity is detected to minimize the DoS attacks over a large period of time. Hence, this security model ensures that spoofed IP addresses also fail to successfully attack web servers.