search

espressif / esp-csi

Applications based on Wi-Fi CSI (Channel state information), such as indoor positioning, human detection
Apache License 2.0
690 stars 110 forks source link

how to interpret "data" collected from record function of "console_test" (AEGHB-540) #162

Closed nixonmok closed 8 months ago

nixonmok commented 8 months ago

I am collecting CSI training data for motion detection for my final year project, and I used "console_test"'s function to collect (which is, 5.5 Collect CSI data for a specific action). The data was collected properly but I have some questions:

  1. the collected data has only 100 data, is that normal? Since I read other issue from this project (#145 ), 64 subcarriers. from my understanding after reading several issues and espressif website, in my situation, only 50 subcarriers are record right? if yes, can I collect more subcarrier? my ESP32 model is ESP32-WROOM-32D. SAMPLE DATA type,seq,timestamp,taget_seq,taget,mac,rssi,rate,sig_mode,mcs,cwb,smoothing,not_sounding,aggregation,stbc,fec_coding,sgi,noise_floor,ampdu_cnt,channel_primary,channel_secondary,local_timestamp,ant,sig_len,rx_state,len,first_word_invalid,data CSI_DATA,360521,2024-01-25 14:59:40.255,1,static,1a:00:00:00:00:00,-51,11,1,0,1,1,1,0,0,0,1,-96,0,11,2,3847298315,0,44,0,104,0,"[-14, -10, -15, -9, -15, -8, -16, -8, -16, -8, -15, -8, -15, -7, -15, -7, -15, -6, -15, -6, -16, -6, -17, -6, -16, -5, -16, -5, -17, -5, -17, -4, -17, -4, -16, -4, -17, -3, -17, -3, -17, -3, -17, -3, -17, -3, -18, -2, -19, -2, -20, -2, -18, 0, -20, 0, -20, 0, -20, 0, -20, -1, -20, -1, -20, -1, -20, -1, -20, -1, -21, -1, -21, -1, -20, -2, -20, -2, -21, -2, -21, -2, -21, -2, -21, -2, -21, -3, -20, -4, -20, -4, -19, -5, -18, -6, -18, -6, -18, -7, -17, -7, -16, -7]"

  2. Do I need to do "5.2 Calibration threshold" before collecting CSI data? Not so sure what's the use of doing calibration.

Thank you

MacChu0315-Espressif commented 8 months ago

For question one, the provided CSI data consists of information corresponding to 52 subcarriers, and this data has been processed to exclude pilot subcarriers and all-zero subcarriers.

Regarding question two, this example offers two detection algorithms: motion detection and human presence detection. Calibration is required for the human presence detection.