Congestion avoidance algorithm using ARIMA(2,1,1) model-based RTT estimation and RSS in heterogeneous wired-wireless networks

Abstract

In the past two decades, there has been a fast growth in the heterogeneous wired-wireless network which connects variety of networks such as wired network, wireless network, high speed network and sensor network with the Internet. Since the wired network and wireless network use the different technologies, there is a necessity of developing new protocols or redesigning of existing protocols. Transmission Control Protocol is redesigned to achieve the reliable and effective communication in the heterogeneous wired-wireless network. The basic assumption made in Transmission Control Protocol is that the packet losses are caused by the network congestion but in case of wireless network, the packet losses may be due to temporary link failure, high Bit-Error-Rate etc. Hence the congestion avoidance algorithm of Transmission Control Protocol needs to be redesigned to cope with heterogeneous wired-wireless network. Therefore in this paper, we propose a congestion avoidance algorithm which differentiates the cause of packet losses. The essential components of proposed algorithm are round-trip-time estimation, backlogged packet estimation, channel status notification, loss differentiation algorithm and congestion window control. A round-trip-time estimation using ARIMA(2,1,1) model is proposed in this paper which estimates the sudden changes in the round trip time. It helps us to estimate the backlogged packet and thereby differentiating the congestive packet loss from non-congestive packet loss. In addition, we use the received signal strength of wireless receiver to differentiate the packet losses. Based on the packet loss differentiation, the congestion window size is controlled to avoid the congestion in the network. The proposed algorithm is simulated using network simulator under different network environments. From the simulation results, it is realized that the proposed congestion avoidance algorithm performs well in the heterogeneous wired-wireless network.