Distributed Lowrate DenialofService attacks (DLDoS) exploit the vulnerability of the adaptive behaviours exhibited by network protocols and network services. Its attack efficiency and ability of concealment are far higher than the traditional floodingbased DDoS attacks, thus it is harder to detect and defense. In this paper, we first model and formalize the DLDoS attacks, and then propose an approach of detecting DLDoS based on the congestion participation rate (CPR). Experiments and analysis demonstrate that the approach can detect the DLDoS attacks accurately and reduce the false alarm rate drastically.