You are not logged in.
Hello. I am running a BEACON detection experiment on 802.11 Ref Des v1.7.0 (STA/DCF)
I am finding a -69 dBm sensitivity. What are some parameters I can play with to improve detecting BEACONS at negative SNR?
Offline
1: SDK modification?
maybe you can help me understand the arguments here in w3.phy.util.c:
wlan_phy_pktDet_autoCorr_dsss_cfg(corr_thresh, energy_thresh, timeout_ones, timeout_count)
2: would this need EDK licensing to compile? maybe here wlan_phy_rx_init.m:
PHY_CONFIG_PKT_DET_DSSS_MIN_ONES = 20;%30; PHY_CONFIG_PKT_DET_DSSS_MIN_ONES_TOT = 40;
3:maybe a different set up (other than STA/DCF...note I am in sniffer mode for captures)
I am open to other suggestion or better understanding of the parameters mentioned above. Thanks!
Last edited by OTurner2013 (2017-Jul-06 07:43:10)
Offline
I would suggest modifying the DSSS packet detector params, configured via the wlan_phy_pktDet_autoCorr_dsss_cfg() function.
This function is called in wlan_phy_init() in w3_phy_util.c. The four arguments are:
corr_thresh: 8-bit correlation threshold; higher values require higher correlation peaks for detection
energy_thresh: 12-bit energy threshold; higher values require larger I/Q values for detection
timeout_ones: minimum number of decoded "1" values for detection of SYNC, then SFD
timeout_count: number of decoded bit periods to search for timeout_ones number of decoded "1" values
To increase sensitivity of the DSSS Rx process I would suggest reducing the corr_thresh and/or energy_thresh values before reducing the timeout values.
2: would this need EDK licensing to compile? maybe here wlan_phy_rx_init.m:
You should be able to use the SDK to modify the C code and update the bitstream (download.bit) without a paid EDK license.
The wlan_phy_rx_init.m script is only used to configure the Rx PHY for simulation. All PHY params are set by the C code (mainly in w3_phy_util.c) in hardware. The M script is a good reference for decoding the various fields in PHY registers, but the register values set in the M script are overridden by the C code int he actual hardware design.
Offline
One subtlety I should mention is that the DSSS pkt det and OFDM pkt det subsystems effectively compete for detecting the same waveform. When one subsystem has asserted the other is bypassed until the detected event is complete (either because it was a valid detection and the PHY decodes the packet, or because it was an invalid detection and the pkt det subsystem times out). You can increase the sensitivity of one subsystem but this will increase the probability of false positive detection events, blocking possible detections by the other subsystem. This isn't an issue in your experiment where the whole OFDM pipeline is disabled, but is definitely a consideration when tuning params when both DSSS and OFDM are enabled.
Offline
Thank you! MANGO's board and design is impressive with so many reconfigurable features to experiment in. I am sure I will be in touch with future project questions :)
Offline