• RSSI-based Features:
    • Time related features (e.g., period of slope)
    • Statistical features (e.g, peak, mean, variance)
    • Frequency related features (e.g. low-pass filter, high-pass filter, spectral energy, entropy)
    • Edge features (e.g., rising edge, falling edge)
    • Histogram (i.e., the signal distribution)
    • Correlation (e.g., spearman, Pearson)
    • Distance (e.g., manhattan, euclidean)
  • Sensing Objects: Localization, Presence Detection (Coarse-grained), Activity Recognition (Fine-grained)
  • Methods: Filtering (with threshold), ML (e.g. KNN, SVM, Bayesian Network), Wavelet Transform
  • Another taxonomy can be “Device-based” and “Device-free”


  • Tracking Human Queues[MobiSys14] has a very detailed description of what are derived from RSSI raw data and used as feature for following sensing.

  • Although [MobiCom15-Understanding] is a CSI based work, there is a good survey on the RSSI-based human activity recognition topic. But the survey here is to citing some RSSI-based works and show their performance defects compared with (author’s) CSI-based work.

  • There is also a survey on RSSI-based sensing in [IMWUT17-Detecting] and [MobiCom15-Keystroke].

Special Topics

  • Challenges, MobiCom07
    • RSSI-based features: Statistical features (Moving average, Moving variance)
    • Sensing Object: Localization
    • Method: Filtering (with threshold)
  • IODetector, SenSys12
    • RSSI-based features: Statistical features (RSSI variation)
    • Sensing Object: Localization
  • Leveraging RF-channel fluctuation for activity recognition Active, MoMM13
    • RSSI-based features: Statistical features , Frequency related features
    • Sensing Object: Activity Recognition
    • Method: KNN

RSSI Features

  • Predicting Length of Stay at WiFi Hotspots, INFOCOM13
    • RSSI-based features: Statistical features (e.g, peak, mean, variance)
    • Sensing Object: Presence Detection
    • Method: ML (SVM)
  • Tracking Human Queues, MobiSys14
    • RSSI-based features: Time related features, Statistical features
    • Sensing Objects: Presence Detection
    • Method: Filtering, ML (Bayesian Network)
    • Three features are extracted from raw RSSI data: (1) Longest period of negative slopes in the trace; (2) Stable RSS & largest RSS value during service; (3) Largest change of RSS in the trace.

RSSI Features

(Figure from the Paper)

  • The Telepathic Phone, PerCom14
    • RSSI-based features: Statistical features (peak value, mean, variance)
    • Methods: ML (KNN)
  • RF-Sensing of Activities, TMC14
    • RSSI-based features: Statistical features (peak value, mean, variance), Frequency related features (energy, entropy)
    • Sensing Objects: Activity Recognition
    • Methods: ML (KNN)
  • UbiBreathe, MobiHoc15
    • RSSI-based features: Frequency related features (e.g. low-pass filter, high-pass filter)
    • Sensing Objects: Activity Recognition
    • Method: Wavelet Transform
  • WiGest: INFOCOM15
    • RSSI-based features: Edge features (e.g., rising edge, falling edge)
    • Sensing Objects: Activity Recognition
    • Methods: Wavelet Transform
  • Inferring Person-to-person Proximity Using WiFi Signals, IMWUT17
    • RSSI-based features: Correlation (spearman and pearson correlation of RSSI from two APs), Distance (manhattan and euclidean RSSI from two APs).
    • Sensing Objects: Presence Detection


[MobiCom07] Youssef, M., Mah, M., & Agrawala, A. (2007, September). Challenges: device-free passive localization for wireless environments. In MobiCom(pp. 222-229). ACM.

[SenSys12] Zhou, P., Zheng, Y., Li, Z., Li, M., & Shen, G. (2012, November). IODetector: A generic service for indoor outdoor detection. In SenSys (pp. 113-126). ACM.

[MoMM13] Sigg, S., Shi, S., Buesching, F., Ji, Y., & Wolf, L. (2013, December). Leveraging RF-channel fluctuation for activity recognition: Active and passive systems, continuous and RSSI-based signal features. In MoMM(p. 43). ACM.

[INFOCOM13] Manweiler, J., Santhapuri, N., Choudhury, R. R., & Nelakuditi, S. (2013, April). Predicting length of stay at wifi hotspots. In IEEE INFOCOM (pp. 3102-3110). IEEE.

[MobiSys14] Wang, Y., Yang, J., Chen, Y., Liu, H., Gruteser, M., & Martin, R. P. (2014, June). Tracking human queues using single-point signal monitoring. In MobiSys (pp. 42-54). ACM.

[PerCom14] Sigg, S., Blanke, U., & Tröster, G. (2014, March). The telepathic phone: Frictionless activity recognition from wifi-rssi. In PerCom (pp. 148-155). IEEE.

[TMC14] Sigg, S., Scholz, M., Shi, S., Ji, Y., & Beigl, M. (2014). RF-sensing of activities from non-cooperative subjects in device-free recognition systems using ambient and local signals. IEEE TMC, 13(4), 907-920.

[MobiCom15-Keystroke] Ali, K., Liu, A. X., Wang, W., & Shahzad, M. (2015, September). Keystroke recognition using wifi signals. In MobiCom (pp. 90-102). ACM.

[MobiCom15-Understanding] Wang, W., Liu, A. X., Shahzad, M., Ling, K., & Lu, S. (2015, September). Understanding and modeling of wifi signal based human activity recognition. In MobiCom(pp. 65-76). ACM.

[INFOCOM15] Abdelnasser, H., Youssef, M., & Harras, K. A. (2015, April). Wigest: A ubiquitous wifi-based gesture recognition system. In INFOCOM (pp. 1472-1480). IEEE.

[MobiHoc15] Abdelnasser, H., Harras, K. A., & Youssef, M. (2015, June). UbiBreathe: A ubiquitous non-invasive WiFi-based breathing estimator. In MobiHoc(pp. 277-286). ACM.

[IMWUT17-Detecting] Ohara, K., Maekawa, T., & Matsushita, Y. (2017). Detecting state changes of indoor everyday objects using Wi-Fi channel state information. IMWUT, 1(3), 88.

[IMWUT17-Inferring] Sapiezynski, P., Stopczynski, A., Wind, D. K., Leskovec, J., & Lehmann, S. (2017). Inferring person-to-person proximity using WiFi signals. IMWUT, 1(2), 24.