Abstract: Given that the wireless communication between fishing boats far away from the coast and the shore will be seriously affected in case of emergency situations such as bad weather or natural disasters at sea in the complex and volatile marine environment, a classification algorithm based on decision tree integration algorithm was proposed to locate the fishing boats with the best communication conditions according to the classification results. First, data preprocessing was performed on the data set containing 200 fishing boats. Then based on the decision tree C4.5 algorithm, the four attributes including the signal strength of the fishing boat itself and its environmental conditions (wind speed, sea surface temperature, and sea surface pressure)were extracted as classification attributes to classify the data set. Finally, the Bagging algorithm and Boosting algorithm in ensemble learning were used to build strong classifiers to improve classification performance. The performance comparison results showed that the ensemble learning algorithm can improve the classification performance of the decision tree and the decision tree with Boosting algorithm has the highest classification accuracy of 95.76%. It has been proved that this method can help to quickly and accurately locate one or more fishing boats with the optical communication singals, through which, information can be conveyed to the target fishing vessels in the distance and the quality of maritime wireless communications can be guaranteed
.

Key words: wireless communication, decision tree, optimal communication fishing boats, ensemble learning

摘要： 针对海洋环境复杂多变，如遇到海上恶劣天气或自然灾害等紧急情况时，离海岸较远的渔船与岸边的无线通信会受到严重影响，提出了一种基于决策树集成算法的分类算法，根据决策树的分类结果寻找具备最佳通信条件的渔船。首先对包含200艘渔船的数据集进行数据预处理，然后根据决策树C4.5算法提取渔船自身的信号强度和所处的环境条件（风速、海面温度和海面气压）等4个属性作为分类属性对数据集进行分类，最后使用集成学习中的Bagging算法和Boosting算法构建强分类器以提高分类精度。性能比较结果显示，集成学习算法能提升决策树的分类性能，基于Boosting算法的决策树分类精度最高，达到95.76%。研究表明，该方法有助于快速准确地寻找一艘或多艘最佳通信渔船，通过该渔船传播信息给远处的目标渔船，保障海上无线通信质量。

关键词: 无线通信, 决策树, 最佳通信渔船, 集成学习