Abstract: In order to improve the robustness of indoor positioning system, an access point lost prevented positioning algorithm is proposed. The detection of access point lost is based on the [K] nearest neighbor approach according to the distortion in signal space. The possibility that an access point has been lost is updated constantly and the distance metric is modified when access point lost is detected. Experimental results indicate that the proposed algorithm can provide reliable location estimation even a part of access points have been lost. A lost access points set can be determined after obtained several observations, which can provide a basis for the indoor positioning system to eliminate anomaly.

Key words: indoor positioning, robustness, access point lost, [K] nearest neighbor approach

摘要： 为了增强室内定位系统的鲁棒性提出了一种WLAN环境下防接入点丢失的室内定位算法。在[K]最近邻法的基础上根据信号空间畸变对接入点丢失情况进行实时检测并不断更新各接入点的丢失可能性，当检测到存在接入点丢失时对信号空间距离进行修正。实验结果表明，该算法可以在部分接入点丢失的条件下仍提供可靠的位置估计并在获取一定数量观测值后确定一个丢失接入点集合，为定位系统排除异常提供依据。

关键词: 室内定位, 鲁棒性, 接入点丢失, [K]最近邻法