Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

[STDS-802-11-TGBN] P-EDCA open issues



Hi all, 

Let us continue the discussion of today’s presentation on open issues of P-EDCA in this thread
https://mentor.ieee.org/802.11/dcn/25/11-25-0092-02-00bn-high-priority-edca-management.pptx

[Dmitry]: "for the second part (resuming backoff counter). I think the question should be set in a bit different way (i.e. broader) - do we think it is beneficial to reuse AC_VO BK counter and simply maintain it thoughout EDCA and P-EDCA periods or isolate one from another(ie. have separate non-overlapping counters) “. 
[ML] Keeping one retry counter seems fine. Right now, I don’t see a case when this could lead to cheating or fairness issues. Moreover, this would greatly simplify the rules for switching between EDCA and P-EDCA operation. 

[Reza]: " For the 1st part, I prefer that the duration is fixed for DS-CST (given its impact of the choice of scrambler sequence). The current spec has it fixed at 97us. It'd be good to first converge if there is a problem before proposing a solution. Btw, does this problem occur during DL  or UL case?  Regarding the 2nd part, we've settled on counting each DS-CTS transmission as one P-EDCA opportunity and have set the consecutive such opportunities to be 1 (the default value). We need to have more sim and analysis before deciding on your suggestion. Did you by any chance have such analysis e.g. the improvement for P-EDCA STAs and the impact on legacy devices?  
[ML]: 
  1. That’s a valid point. However, all STAs in BSS would use same Duration value, so a chance of a collision of different DS-CTS frames isn't very high. Also,  we haven’t seen any evidence that setting same scrambler sequence may improve reception at all. 
  2. The problem is real: 24Mb/s RTSs are often used, NAV rules do not allow responding to an RTS until NAV expires. 
  3. The problem occurs in both UL and DL
  4. We have another degree of freedom: implementation complexity, but let me get back to the tradeoff between P-EDCA and legacy performance, and bring some illustrative results shortly. 

Thanks

Mikhail Liubogoshchev

To unsubscribe from the STDS-802-11-TGBN list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBN&A=1