Indoor multiple target tracking is a promising research field that attracts many efforts. Traditional approaches for tackling this problem are usually model-based methods. WiFi-based tracking approaches suffer from high cost in retrieving the CSI information. Most RF signal-based methods provide a mathematical framework correlating movement in space to a link’s RSS value. Real RSS values are used to model the signal attenuation, and the distance correlation with signal attenuation is used to estimate locations. In this paper, we propose DCT, a noise-tolerant, unobtrusive and device-free tracking framework. DCT adopts density-based clustering to find the centers. We further use a linear function of mean RSS variances and target amount and FCM algorithm to adjust the number of targets and positions. The multiple particle filter (MPF) is adopted to refine the target tracking accuracy. DCT is tolerant for noise and multi-path effects, and can fast simultaneously tracking with a O(N) time complexity. The extensive experiments in trace-driven simulations and real implementations show that DCT is efficient and effective in tracking multiple target, and can achieve a high precision.

中文翻译：

---

移动环境中的无设备室内多目标跟踪

室内多目标跟踪是一个有前途的研究领域，吸引了许多努力。解决此问题的传统方法通常是基于模型的方法。基于WiFi的跟踪方法在检索CSI信息方面成本很高。大多数基于RF信号的方法都提供了一个数学框架，将空间运动与链接的RSS值相关联。实际RSS值用于对信号衰减进行建模，而距离与信号衰减的相关性用于估计位置。在本文中，我们提出了DCT，一个噪音-宽容，不显眼和设备-免费跟踪框架。DCT采用基于密度的聚类来查找中心。我们进一步使用均值RSS方差和目标数量的线性函数以及FCM算法来调整目标和位置的数量。采用多粒子滤波器（MPF）来提高目标跟踪精度。DCT容忍噪声和多径效应，并且可以以O（N）时间复杂度快速同步跟踪。跟踪驱动模拟和实际实现中的大量实验表明，DCT在跟踪多个目标方面是高效且有效的，并且可以实现高精度。