Estimating Channel States by Density-Based Rival Penalized Competitive Learning

摘要：

We propose a channel states estimation method named as the density-based RPCL (DERPCL) to prune the structure of a rival penalized competitive learning (RPCL) method by evaluating the data density of each unit. Although the RPCL has been shown to perform clustering methods without knowing the extract number of clusters, they may not solve the problems of local optima and slow learning speed for complicated circumstances. This newly proposed method is applied to estimating channel states, and its results are compared with the other RPCL methods. Our results show that the proposed DERPCL method outperforms the traditional ones in terms of convergence accuracy and speed.

关键词： Density-based RPCL channel states estimation clustering method RPCL DERPCL

作者: Yao-Jen Chang Chia-Lu Ho

作者单位: Department of Communication Engineering Central University, Chung-Li, 320

会议类型: 国际会议

会议名称: 2010 Cross-Strait Conference on Information Science and Technology(2010 海峡两岸信息科学与技术学术交流会)

会议地点: 秦皇岛

会议语种:英文

页码: 221-224

在线出版日期: 2010-07-09（万方平台首次上网日期，不代表论文的发表时间）