Abstract:
TCP is the most predominant protocol offering services in the internet. The use of ACK to implement reliable data delivery is built into the protocol’s fundamental property of “self-clocking” mechanism. It uses returning ACKs as a “clock” to determine when to transmit new packets to the network without adding to the level of congestion. When congestion occurs, the traffic control scheme would momentarily stop transmitting packets to avoid further congestion. As a result, ACKs stop transiting back and no new packet can be clocked out. Therefore, the loss of “self-clocking” can severely degrade TCP performance. The Packet Retransmission Algorithm of this work is a promising solution to this problem. The simulation results show that the model reduces redundant duplicate ACKs, false retransmits and packet loss rates to a barest minimum. It achieves near-capacity goodput. In future, this proposed method can be further annexed to study the system fairness index and evaluate its effectiveness in practice.
Keywords: Duplicate Acknowledgment, Retrans-mission, Transport Control Protocol.
