The internet is spreading fast and the diversity of its components affects the performance unpredictably. This leads to the continuous examination of internet hardware structure for the purpose of user experience improvement. Network congestion is one of the challenges that affects network performance, which mostly occurs when the arriving packets exceed available network resources. When this occurs, incoming packets face unpredicted losses or delay. Thus, congestion has an impact on worsening the network performance due to an increase in packet loss. Therefore, a high performance approach called CDGRED was proposed to overcome these constraints using adaptive techniques. An optimized implementation with a suitable parameter tuning for CDGRED method was proposed with results showing clearly enhanced outputs. The CDGRED approach performance is empirically tested and compared with existing methods such as GRED, DGRED, and FLRED. Experimental results prove that the proposed approach has higher performance in early congestion detection over existing approaches.