HuAc: Human Activity Recognition Using Crowdsourced WiFi Signals and Skeleton Data

Abstract

The joint of WiFi-based and vision-based human activity recognition has attracted increasing attention in the human-computer interaction, smart home, and security monitoring fields. We propose HuAc, the combination of WiFi-based and Kinect-based activity recognition system, to sense human activity in an indoor environment with occlusion, weak light, and different perspectives. We first construct a WiFi-based activity recognition dataset named WiAR to provide a benchmark for WiFi-based activity recognition. Then, we design a mechanism of subcarrier selection according to the sensitivity of subcarriers to human activities. Moreover, we optimize the spatial relationship of adjacent skeleton joints and draw out a corresponding relationship between CSI and skeleton-based activity recognition. Finally, we explore the fusion information of CSI and crowdsourced skeleton joints to achieve the robustness of human activity recognition. We implemented HuAc using commercial WiFi devices and evaluated it in three kinds of scenarios. Our results show that HuAc achieves an average accuracy of greater than 93% using WiAR dataset.

Human Activity Sensing with Wireless Signals: A Survey

“…The weighted moving average filter utilizes the scheme that the values closer to the processed point should occupy a higher weight. HuAc [75] smooths and reduces the serrates of CSI waveform by using the weighted moving average filter. Single-Sideband Gaussian (SSG) applies a convolution to smooth the raw signal, which is used to preprocess the CSI waveform in [76].…”

“…Time-domain Pros: relatively simple calculation; Cons: vulnerable to environmental changes and noise maximum, minimum, mean, standard deviation, kurtosis, skewness, variance, median and median absolute deviation, percentiles, root sum square, interquartile range [13,18,37,40,42,43,46,[48][49][50]62,72,73,75,76,[78][79][80][83][84][85][88][89][90]97,108,127,130], time lag [37], power decline ratio [37], amplitude sequence [20,54,74,91], phase sequence [39,51,107], CV [38] Frequency-domain Pros: capture the periodical characteristics of human motion; Cons: large amount of calculation FFT coefficient [12,83], dominant frequency [43], power spectrum density [88], spectral entropy [43,57,70,72,…”

“…Template matching recognition is often training-free [10,19,20,39,55,74,91,93,94]. Training-once classification requires the valid features robust to the variations in the surrounding environment [12,13,32,37,40,42,44,45,[47][48][49]56,59,60,70,72,73,75,76,78,79,83,84,89,90,92,127,130]. Deep learning automatically extracts features, which often requires only one time of training [16,17,21,48,59,67,77,88,95,96,108].…”

“…Pros: no training needed; Cons: accuracy depending on the specific template Euclidean distance [19,94], DTW [10,20,39,42,55,74,91,93,97], EMD [54,70,81,89], Jaccard coefficients [69] Machine Learning Pros: high efficiency and robustness; Cons: a lot of training data required KNN [18,40,46,48,51,52,59,82,99,124], SVM [12,13,32,37,40,42,44,45,[47][48][49]56,59,60,70,72,73,75,76,78,79,83,84,89,...…”

Real-time GIS for smart cities

“…Many smart business applications can benefit from such low-cost, high-availability techniques, such as evaluation of business rates on properties or estimating unit rental values (Golderzahi andPao 2018, Sahni et al 2018). Wifi-based real-time indoor positioning technology can also be used in other smart-city applications, such as mapping urban population (Kontokosta and Johnson 2017), human movement analytics (Zhang et al 2016), and automatic control of smart-home devices (Guo et al 2018).…”