基于LWR-ABCSVR的WiFi指纹定位算法  

Fingerprinting Positioning Algorithm for Wi Fi Based on Locally Weighted Regression and Support Vector Regression Optimized by Artificial Bee Colony

作　　者：王艳[1] 殷富成 纪志成[1] 严大虎[1] 

机构地区：[1]江南大学物联网技术应用教育部工程研究中心,无锡214122

出　　处：《系统仿真学报》2017年第6期1193-1200,共8页Journal of System Simulation

基　　金：国家自然科学基金(61572238);江苏省杰出青年基金(BK20160001);江苏省产学研联合创新资金-前瞻性联合研究项目(BY2016022-24)

摘　　要：针对在Wi Fi环境下,传统的位置指纹定位算法定位精度不够高和指纹数据库构建困难的问题提出了一种基于线性加权回归(LWR)和蜂群优化的支持向量回归机(ABCSVR)的LWR-ABCSVR定位算法。该算法通过LWR在离线阶段对采集到的位置指纹数据库进行扩充;利用ABCSVR构建物理位置和RSS之间的非线性关系,并通过构建的预测模型完成定位。实验结果表明,该算法的定位精度远高于传统的几种定位算法,并且可以在一定程度上减少构建指纹数据库的工作量,是一种综合性能良好的定位算法。Since the traditional location fingerprinting algorithms have poor positioning accuracy and cost laborious efforts constructing fingerprinting database during the offiine phase, a novel LWR-ABCSVR positioning algorithm was proposed, that the derived algorithm was based on the locally weighted regression （LWR） method and support vector regression was optimized by artificial bee colony （ABCSVR） algorithm. By using the proposed algorithm, the fingerprinting database was expanded by LWR step. The ABCSVR algorithm was employed to buiM the nonlinear relationship between the RSS values of reference points and their locations. The position of mobile terminal was predicted by the constructed model. Simulation results indicate that the proposed algorithm performs better than traditional location fingerprinting algorithms, in terms of positioning accuracy and database constructing costs.

关 键 词：WIFI LWR算法 ABC算法 SVR算法 定位技术 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]