摘要
针对传统的基于贝叶斯的位置指纹算法存在着后验概率可能为0以及计算量较大的问题,对算法进行了改进。离线阶段,选取固定的AP进行信息采集,防止突然加入的AP造成无法定位的情况发生,并采用高斯概率分布描述信号强度分布;在线阶段,通过判断信号最强的AP,去除一部分不可能的参考节点,以此减小计算量,同时对后验概率最大的若干个参考点进行加权处理,进一步减小误差;测试阶段,利用Android客户端采集数据信息,并在Matlab平台上进行仿真测试。结果表明,改进的算法可以有效地提高定位精度并减少计算量。
Aiming at the problems that the probability of the traditional location fingerprint algorithm based on Bayesian may be null and the large amount of calculation,this paper improves the algorithm. In the off-line phase,we select the fixed AP for information collection to prevent non-located situation caused by the sudden addition of AP,and we use the Gaussian probability distribution to describe the signal strength distribution. In the on-line phase,in order to reduce the computation and error,we remove some impossible reference nodes by judging the strongest signal AP,and several reference points which have maximum posteriori probability are weighted. On the testing stage,we use the Andriod client to collect data information,and carry on simulations on Matlab platform. The results show that the improved algorithm can effectively improve the positioning accuracy and reduce the amount of calculation.
出处
《江南大学学报（自然科学版）》
CAS
2015年第5期527-531,共5页
Joural of Jiangnan University (Natural Science Edition)
基金
国家自然科学基金项目(60973095)
关键词
位置指纹
贝叶斯
高斯分布
location fingerprint
bayesian
gaussian distribution
