School of Electrical, Computer and Telecommunications Engineering, University of Wollongong, Wollongong, NSW 2522, Australia.
School of Electrical and Data Engineering, University of Technology Sydney, 15 Broadway, Ultimo, NSW 2007, Australia.
Sensors (Basel). 2022 Nov 3;22(21):8472. doi: 10.3390/s22218472.
In this paper, human step length is estimated based on the wireless channel properties and the received signal strength indicator (RSSI) method. The path loss between two ankles, called the on-ankle path loss, is converted from the RSSI, which is measured by our developed wearable hardware in indoor and outdoor ambulation scenarios. The human walking step length is estimated by a reliable range of RSSI values. The upper threshold and the lower threshold of this range are determined experimentally. This paper advances our previous step length measurement technique by proposing a novel exponential weighted moving average (EWMA) algorithm to update the upper and lower thresholds, and thus the step length estimation, recursively. The EWMA algorithm allows our measurement technique to process each shorter subset of the dataset, called a time window, and estimate the step length, rather than having to process the whole dataset at a time. The step length is periodically updated on the fly when the time window is "sliding" forwards. Thus, the EWMA algorithm facilitates the step length estimation in real-time. The impact of the EWMA parameter is analysed, and the optimal parameter is discovered for different experimental scenarios. Our experiments show that the EWMA algorithm could achieve comparable accuracy as our previously proposed technique with errors as small as 3.02% and 0.30% for the indoor and outdoor scenarios, respectively, while the processing time required to output an estimation of the step length could be significantly shortened by 53.96% and 60% for the indoor walking and outdoor walking, respectively.
本文基于无线信道特性和接收信号强度指示（RSSI）方法估计人体步长。在室内和室外步行场景中，通过我们开发的可穿戴硬件测量 RSSI，将两踝之间的路径损耗（称为踝间路径损耗）转换为 RSSI。通过可靠的 RSSI 值范围估计人体步行步长。通过实验确定该范围的上限和下限。本文通过提出一种新的指数加权移动平均（EWMA）算法来更新上限和下限，从而递归地更新步长估计值，改进了我们之前的步长测量技术。EWMA 算法允许我们的测量技术处理称为时间窗口的数据集的每个较短子集，并估计步长，而不必一次处理整个数据集。当时间窗口“滑动”时，步长会实时周期性地更新。因此，EWMA 算法有助于实时进行步长估计。分析了 EWMA 参数的影响，并为不同的实验场景发现了最佳参数。我们的实验表明，EWMA 算法可以达到与我们之前提出的技术相当的精度，对于室内和室外场景，误差分别为 3.02%和 0.30%，而输出步长估计所需的处理时间可以分别显著缩短 53.96%和 60%，用于室内行走和室外行走。
