You are not logged in.
Hi,
I have a question about the granularity of RSSI reporting with respect to packet processing time:
Is the RSSI / RX power reported in lower MAC framework an average value over the entire packet, or an instantaneous value taken at some point in the PHY processing pipeline?
Thanks.
Offline
The Rx power estimate in the MAC code is computed from the RSSI during the preamble.
The Rx PHY computes a continuously running sum (average without the divide) of the raw RSSI. The depth of this sum is configured from software; the reference code sets it to 4 samples by default. Each RF interface implements identical RSSI summing circuits.
The summed RSSI value is used for 3 things:
- The energy-threshold packet detector (disabled by default in the reference code; the auto-correlation detectors work well enough over the full range of SNRs).
- The energy detection input to the medium busy logic (the 802.11 spec says CCA BUSY should assert whenever the observed Rx power exceeds -62 dBm)
- The summed RSSI value is recorded in the PHY's WLAN_RX_PKT_RSSI_AB register. This register contains the value of the summed RSSI a few cycles after the AGC completes its state machine. Thus the captured RSSI reflects the Rx power of the STS in the preamble. The MAC code uses this register to compute the Rx power estimate recorded with the packet metadata.
Offline
Thank you murphpo. Is it safe to assume that Rx power stays relatively constant over the entire packet? As far as I know, the Tx signal power is kept at the same level, and Tx gains are fixed for the whole packet. But can other environmental factors, such as fast channel fade, make the Rx power fluctuates over packet reception time?
Best,
Danh
Offline
Is it safe to assume that Rx power stays relatively constant over the entire packet? As far as I know, the Tx signal power is kept at the same level, and Tx gains are fixed for the whole packet. But can other environmental factors, such as fast channel fade, make the Rx power fluctuates over packet reception time?
This is a mostly-safe assumption, and is one we make in the Rx PHY design. Whether it is totally safe is...complicated.
The propagation channel is constantly changing. The rate of change depends on how fast the transmitter, receiver, and/or scatterers are moving. In the standard frequencies, the indoor channels seen by WLAN devices vary slowly enough that you can, with high probability, consider them to be static for typical packet durations (100's of usec, even a few msec). However this is a probabilistic argument. With some small probability the channel coefficients will change very quickly. If this happens mid-reception it will probably cause a packet loss. For unicast packets this loss will be resolved by a re-transmission by the DCF.
Offline
Great answer! Thanks for the information.
Offline