[论文解读] Spot: An accurate and efficient multi-entity device-free WLAN localization system

Ibrahim Sabek, Moustafa Youssef|arXiv (Cornell University)|Jul 18, 2012

Indoor and Outdoor Localization Technologies参考文献 1被引用 9

一句话总结

Spot 提出了一种结合条件随机场与马尔可夫模型的概率能量最小化框架，实现了准确、高效且可扩展的多实体免设备WLAN定位。通过引入一种新型交叉校准技术，并将能量函数映射为二值图割问题，Spot 实现了小于1.1米的中位跟踪误差，且在估计实体数量方面达到100%准确率，相比最先进单实体系统，中位误差至少降低36%。

ABSTRACT

Device-free (DF) localization in WLANs has been introduced as a value-added service that allows tracking indoor entities that do not carry any devices. Previous work in DF WLAN localization focused on the tracking of a single entity due to the intractability of the multi-entity tracking problem whose complexity grows exponentially with the number of humans being tracked. In this paper, we introduce Spot as an accurate and efficient system for multi-entity DF detection and tracking. Spot is based on a probabilistic energy minimization framework that combines a conditional random field with a Markov model to capture the temporal and spatial relations between the entities' poses. A novel cross-calibration technique is introduced to reduce the calibration overhead of multiple entities to linear, regardless of the number of humans being tracked. This also helps in increasing the system accuracy. We design the energy minimization function with the goal of being efficiently solved in mind. We show that the designed function can be mapped to a binary graph-cut problem whose solution has a linear complexity on average and a third order polynomial in the worst case. We further employ clustering on the estimated location candidates to reduce outliers and obtain more accurate tracking. Experimental evaluation in two typical testbeds, with a side-by-side comparison with the state-of-the-art, shows that Spot can achieve a multi-entity tracking accuracy of less than 1.1m. This corresponds to at least 36% enhancement in median distance error over the state-of-the-art DF localization systems, which can only track a single entity. In addition, Spot can estimate the number of entities correctly to within one difference error. This highlights that Spot achieves its goals of having an accurate and efficient software-only DF tracking solution of multiple entities in indoor environments.

研究动机与目标

• 为解决WLAN中多实体免设备定位的不可行复杂性问题，该问题迄今已将系统限制在单实体追踪。
• 将多个实体的校准开销从与位置数量呈指数增长的关系降低为线性增长。
• 设计一种能量最小化框架，以保持实体追踪中的空间与时间一致性。
• 仅利用现有WiFi基础设施，实现对多个免设备实体数量与位置的准确估计。

提出的方法

• 将多实体免设备定位建模为使用条件随机场与马尔可夫模型的概率能量最小化问题，以捕捉空间与时间依赖性。
• 提出一种新型交叉校准技术，使校准复杂度在位置数量上呈线性增长，与追踪实体数量无关。
• 将能量最小化函数映射为二值图割问题，实现高效求解，平均时间复杂度为线性，最坏情况为立方级。
• 对估计的位置候选点进行聚类，以抑制异常值并提升连续空间中的追踪精度。
• 利用标准WiFi接入点和监测点的指纹化RSS测量值，检测人体存在并推断位置。
• 采用基于历史的模型，通过强制时间平滑性并检测聚类合并/分裂事件，提升实体数量估计的准确性。

实验结果

研究问题

• RQ1 是否可以使WLAN中的多实体免设备定位在保持高精度的同时具备计算可行性？
• RQ2 如何将多实体的校准开销从指数复杂度降低为线性复杂度？
• RQ3 联合空间-时间建模框架能否提升免设备定位中的追踪一致性和准确性？
• RQ4 纯软件实现的、基于WiFi的系统在多人追踪中，相较于依赖硬件或单实体系统，能实现多大程度的性能超越？
• RQ5 系统在无先验知识的情况下，能否实时准确估计实体数量？

主要发现

• Spot 在两个真实世界测试平台中实现了小于1.1米的中位跟踪误差，相比最先进单实体系统提升了至少36%。
• 系统将多实体校准开销降低至线性复杂度，使多实体追踪成为可行且可扩展的方案。
• Spot 在实体数量估计上实现了100%准确率，误差不超过一个实体，展现出强大的实体计数鲁棒性。
• 能量最小化框架可通过二值图割高效求解，平均时间复杂度为线性，最坏情况为立方级。
• 对位置候选点的聚类能有效减少异常值，并提升连续空间中的追踪精度。
• 系统在无需特殊硬件的情况下保持高精度，完全依赖标准WiFi基础设施与软件处理。