Research On TCP Transmission Control In Air & Space Network

Air & space network is a type of wireless self-organizing network. Long delay, high BER and asymmetric bandwidth are three typical characteristics in air & space network. If TCP is applied in air & space network directly, there would be plenty of problems, such as, lots of time consumed in slow start phase, not able to correctly estimate the reason of packet loss, not able to guarante RTT fairness and so on. Therefore TCP cannot achieve good transmission performance in air & space network. The main purpose of this thesis is to provide some improvement in designing a transmission control algorithm that can be applied in air & space network.Intensive theoretical analyses of TCP, SCPS-TP, anti-long delay algorithms and anti-high BER algorithms are made in this thesis. Advantages and shortcomings of these algorithms are discussed as well. The author then designs a new transmission control algorithm based on these algorithms. The main idea of the algorithm is to increase the congestion window a little aggressively when channel is in a good state and decreasing the congestion window a little mildly when channel is in a bad state. There are two main parts in this algorithm, namely, self-adaptive window adjustment TCP (SAWA-TCP) algorithm and mixed loss notification (MLN) strategy, which can implement the above idea. The dynamic increasing algorithm is the core of the self-adaptive window adjustment algorithm. In the design of dynamic increase algorithm, this thesis adopted a method of changing the window increment dynamically based on the queue delay information, which can automatically change the increment of congestion window according to the network situation to decrease the probability of network congestion. MLN is a mix of ECN strategy and ELN strategy. It combines the merits of wire-network-based ECN and wireless-network-based ELN. In air & space network, MLN can properly distinguish the reason of packet loss to provide more accurate theoretical evidence for self-adaptive window increase algorithm. Quantity of simulations with NS2 simulation tool aiming at three different air & space network environment have been carried through in this thesis, and a contrast between SAWA-TCP and the other algorithms has been made to validate the superiority of SAWA-TCP.