Research Of Network State Cognitive Model Based On Fuzzy Comprehensive Evaluation

Future telecommunication network covers cable network, satellite network,wireless network and other hybrid networks which is a huge, complex and openheterogeneous networks collection. which due to the complexity of the networkstructure, network transmission performance cognitive becomes more difficult whichis mainly because of the great differences between the various networks-differences inthe underlying transmission media and underlying protocols. By shielding underlyingdifferences, TCP/IP OSI model realizes the data transparent transmission, also causesthat the network cannot identify the differences which cannot perform the targettransmission regulation.To solve the problem of the traditional TCP protocol performance degradation inthe heterogeneous network environment, Firstly, the paper makes use of GreyRelational Analysis to calculate the relational degree of different network parametersand the network status, gets the quantitative description of the network status on thebasis of the network parameters, and according to the quantitative analysis results,selects the parameters which cognizes network status and reflects networkperformance better to make up the factor set; Secondly, by using fuzzy statistical andfuzzy distribution law, the paper establishes the factor set‘s membership functions,which are made use of to independently make judgment of the network status, getsevery factor’s judgment values of current network status, and preliminarily cognizesthe network status according to the values. Thirdly, using close degree of fuzzy sets,this paper makes research of dynamic weights algorithm of network parameters, getsdynamic weights of the network parameters in complicated network environments;Finally, it combines the single factor evaluation set with weights to make fuzzycomprehensive evaluation, and gets network status cognitive model and classifies thenetwork performance into multiple levels on the basis of the model, in order to cognizethe status of network. The simulation results show that the TCP protocol, which adopts the fuzzycomprehensive evaluation-based cognitive model of network status, is capable ofmaking a more accurate judgment of network performance level in complicatednetwork environments comparing to the original TCP protocol; this model improvesthe utilization and the whole throughput value of the TCP protocol and its transmissionperformance in heterogeneous network environments. Based on the work mentionedabove, the paper proposes a detailed division algorithm of network performance levelwhich adopts the parameter of wireless bit-error loss. Under the premise of notreducing the whole network throughput, the algorithm is able to improve the reliabilityof data transmission outstandingly and reduce the system extra overhead effectively.