【摘要】 针对室内无线电发射源的位置指纹定位问题,提出一种改进加权KNN定位算法。在分析现有定位算法的基础上,建立测试点和参考点的余弦相似度关系,并把该余弦相似度用作KNN在线定位计算的权重,计算出第一次加权质心定位结果,根据此结果判断是否进行二次加权来确定测试点最终的估计位置,最后进行算法的仿真测试。结果表明,较之传统位置指纹算法该算法定位准确度提高17%左右,不仅克服传统算法在发射源定位中由于在线阶段针对测试点接收到的信号强度不同造成的定位稳定性差的问题,还避免当存在离测试点较远的参考点时造成的定位误差大的问题。更多还原

【Abstract】 An improved weighted KNN location algorithm was proposed for the position fingerprint location problem of indoor radio source. Firstly, based on the analysis of the existing location algorithm, the cosine similarity relationship between the test point and reference point was established, and the cosine similarity was used as the weight of KNN online location calculation. Furthermore, the first weighted centroid location result was obtained. According to the result, it was judged whether a secondary weighting should be performed to determine the final estimated position of the test point. Finally, the simulation test of the algorithm was carried out,the results show that compared with the traditional position fingerprint algorithm the positioning accuracy of the algorithm is improved by about 17%. It not only overcomes the problem that the traditional algorithm has poor location stability due to different received signal strengths for the test points in the online stage,but also avoids large location error when there is a reference point far from the test point.更多还原