Research On SCTP Differentiate Packet Loss Mechanisms In Wireless And Wired Network

With the rapid development of the Internet and related technology, the increasingly widespread application of the network, network traffic also will surge. In this case, The in-herent bad performance of TCP is becoming increasingly apparent, SCTP with the next gen-eration of the Internet and other related multistreaming multihoming properties, and therefore it is a deal that SCTP will replace TCP transport protocol for the next generation of the Inter-net industry consensus. However, SCTP was originally used to transport in the wired network, making it apply to hybrid wireless and wired network environment will produce a lot of prob-lems, such as wireless mobile environment, BER (bit error rate) significantly increased, the latency spikes occur more frequently, a sudden noise of the device, and the quality of the ra-dio channel is unstable, so the wireless network is worse than the wired network transmission environment. These features will lead to frequent packet loss. In a wired network, SCTP packet loss all comes down to simple network congestion, packet loss once occurs on the path, the the congestion window of the path is halved. However, in wireless networks, packet loss is not entirely due to network congestion, reducing the congestion window is unWISE once phenomenon of packet loss occurs, and this behavior will greatly reduce the SCTP perfor-mance in wireless networks. In order to improve the utilization of both wired and wireless network environment, conducting research on the next generation loss differentiation mecha-nism has important theoretical significance and application value.So far, there have been many researchers study the mechanism of loss differentiation for the next SCTP in wired and wireless network environment, but these existing mechanisms have some more or less imperfect place, their main defects are:(1) Most of the mechanism is to rely on a single parameter and trends of loss differentiation. Such as WISE agreement based on the bandwidth estimation techniques just use the remaining bandwidth of the net-work to distinguish the cause of loss, which is comparatively more one-sided, in the ev-er-changing network environment is difficult to obtain satisfactory results. (2) The current network parameters statistics, evaluation method is to obtain criteria for a network model through simulation, these criteria are mostly identified certain threshold or interval, thereby parameter values currently measured or estimated with this threshold or interval determine the relationship between the loss reasons. Obviously this simple two-valued logic characteristic is not suitable to judge the probability of the evaluation process in the comprehensive integra-tion network.In response to these problems, this paper established a loss differentiation mechanism based on fuzzy comprehensive evaluation model (Fuzzy Comprehensive Evaluation Model). Fuzzy comprehensive evaluation model is a very effective multifactor decision model; it is a valuable method is proposed for accurate ures difficult to quantify the direct evaluation of the problem. This method has had only vague factors and non-quantitative characteristics, through mathematical treatment; it has a quantifiable form of expression, which can be com-pared to provide a basis for decision-making and judgment, to improve scientific and correct decisions. It can objectively evaluate the relationship between the multiple factors and deci-sion-making goals, has certain advantages in handling multi uncertainty factors and evalua-tion outcomes. Therefore, the study of this paper, we introduce fuzzy comprehensive evalua-tion model to describe the evaluation process of network packet loss, the use of fuzzy logic intelligent processing capabilities at the sending end through comprehensive evaluation model to analyze the handling of packet loss reasons, thereby increasing congestion judgment accu-racy, to avoid false starts SCTP congestion control algorithms. Finally, the simulation results verify the accuracy and reasonableness of the model design.Finally, we have a summary and outlook of the issues and propose future research direc-tions.