Abstract: In view of the limitation of the traditional Transmission Control Protocol (TCP) used in the heterogeneous network, a network cognitive model based on fuzzy comprehensive evaluation was proposed. This model, by building the membership functions and different dynamic weight distribution under the different network environment, distinguished the wireless error loss from congestion loss according to fuzzy comprehensive evaluation. The simulation results show: compared with the traditional TCP, under different network conditions, the model can more accurately distinguish the wireless error loss from congestion loss, and improve the TCP throughput and network performance.

Key words: heterogeneous network, membership function, weight, fuzzy comprehensive evaluation, network cognitive

摘要： 针对传统传输控制协议(TCP)应用于异构网络的局限性，提出了一种基于模糊综合评判的网络认知模型。该模型通过建立隶属度函数和不同网络环境下的动态权重分布，运用模糊综合评判的方法来区分无线误码丢包和网络拥塞丢包。仿真实验证明：与传统TCP协议相比，该模型在不同的网络条件下，能够较准确地区分无线误码丢包和网络拥塞丢包，提高了TCP的吞吐量，改善了网络性能。

关键词: 异构网络, 隶属度函数, 权重, 模糊综合评判, 网络认知