From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sergey Ryazanov Date: Sun, 28 Jul 2013 17:11:05 +0400 Subject: [ath9k-devel] sensitivity control for ath9k with mac80211 In-Reply-To: <51EC6B5B.5030400@candelatech.com> References: <2DF5040C00EDCD438915E297DC199E9232512664@CVA-MB001.centreville.ads.sparta.com> <51E92EA9.6000202@gmail.com> <51E940D7.1030607@gmail.com> <51E9436E.5060208@candelatech.com> <51E959E8.9010706@openwrt.org> <51EC6B5B.5030400@candelatech.com> Message-ID: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ath9k-devel@lists.ath9k.org 2013/7/22 Ben Greear : > On 07/19/2013 05:53 PM, Sergey Ryazanov wrote: >> >> 2013/7/19 Felix Fietkau : >>> >>> On 2013-07-19 4:58 PM, Sergey Ryazanov wrote: >>>> >>>> 2013/7/19 Ben Greear : >>>>> >>>>> On 07/19/2013 06:36 AM, Flavio Leonel wrote: >>>>>> >>>>>> Ok i know that but the command iw not permited set sensitivity limit >>>>>> >>>>>> how i can seting this limit on atth9k , this question.. >>>>> >>>>> >>>>> I tried messing with this some months ago and got nowhere. I could >>>>> not figure any way to make the NIC ignore fainter signals, and I am >>>>> not sure it is possible to make the hardware do this... >>>>> >>>> >>>> At least 5k chips have a special register for CCA threshold >>>> configuration. Modern 9k chips have not inherited this register? >>> >>> The CCA threshold is not the same as the threshold for signal detection. >>> In many ways, 5k and 9k cards have a somewhat similar set of registers >>> to configure various aspects of detection sensitivity. Some of those are >>> set in the initvals, some are controlled by ANI. >>> The main issue is that the driver does not expose any convenient knobs >>> to control this. >>> >> Yeah. It would be useful for developers to get delicate and operative >> control over Tx/Rx process (at least via debugfs). But I never faced >> with situations where this would be useful for end user. So, IMHO, >> even if somebody submit appropriate patches, they will not be >> accepted. > > > I do a lot of testing in cases where there are a few APs with signal level > of around -70 to -85, > primarily due to not-that-great shielding, as well as APs in surrounding > buildings. > Do you tried to tune CCA or ANI related stuff, as Felix suggested? > I'd like to be able to tune a NIC so that it just plain ignored those weak > signals > to increase throughput with systems with much better signal quality. > Even if you force NIC to completely ignore weak signals during clear channel detection procedure you can't avoid interference, which could lead to rise of packet error rate, what lead to drop of the throughput. On other hand, if your NIC transmit in the same time as weak APs, you are jamming them. Shielding is the single way for correct testing, until you do not try to create a jammer. -- BR, Sergey