In this research work, an attempt is made to enhance the performance of RTT estimation model used extensively by the researchers for network traffic analysis. The error term in the model given in Eq. (1) is called bias term. The bias term eliminates the noise that introduced into the signals during data transmission over the Internet. The bias term is calculated by the application of Kalman filtering. The error in RTT estimation using model (1) without bias term (Fig. 3) has been reduced, and the values are smoothened after adding the bias term (3) (Fig. 6). The accuracy of the performance of the modified model is validated by comparing the graphs given in Figs. 3 and 6. The estimated RTT values are nearly converging with the measured RTT values. The accuracy in the modified RTT model in turn influences the performance of queue length and the sender window dynamics. This can be observed by comparing the graphs shown in Figs. 7 and 8. The queue length gets smooth and converges to a value of 199.6 packets. Similarly, window size converges to a value of 200.4 packets. The source sending rate matches with the queue length minimizing the congestion and packet loss.
展开▼