Abstract: According requirements of marine observation envirement，a real-time data collection architecture in MOS（Marine Observation System），based on GSM/GPRS communication network，is proposed in this paper.Using GSM SMS and GPRS packets services of mobile network，this architecture guarantees real-time and reliability of data collection.And a novel entropy-based incremental fuzzy clustering algorithm is designed to cluster real-time datum efficiently and gives attention to real-time while improving clustering accuracy by applying a series of improved policies.Results of simulation experiments show that proposed algorithm is adaptive to the requirements of marine observation.

Key words: GSM/GPRS, real-time data collection, entropy, incremental, clustering analysis

摘要： 针对海洋监测环境的要求，提出一种基于移动GSM/GPRS通信网络的无线海洋监测实时数据采集的架构。该架构利用移动网络的GSM短消息和GPRS数据包传输服务，保证了数据采集的实时性和可靠性。在此基础上，设计了一种新的基于熵的增量式模糊聚类算法，通过实施一系列改进策略，算法能有效地对实时数据进行聚类，在提高聚类的准确性的同时兼顾实时性。仿真实验结果表明提出的算法符合海洋实时监控的要求。

关键词: GSM/GPRS, 实时数据采集, 熵, 增量式, 聚类分析