linux-wireless.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Mattias Nissler <mattias.nissler@gmx.de>
To: Stefano Brivio <stefano.brivio@polimi.it>
Cc: linux-wireless <linux-wireless@vger.kernel.org>,
	"John W. Linville" <linville@tuxdriver.com>,
	Johannes Berg <johannes@sipsolutions.net>
Subject: Re: [RFC/T][PATCH v2 2/3] rc80211-pid: introduce PID sharpening factor
Date: Mon, 10 Dec 2007 07:28:01 +0100	[thread overview]
Message-ID: <1197268081.7490.10.camel@localhost> (raw)
In-Reply-To: <20071210032811.5cea4909@morte>


On Mon, 2007-12-10 at 03:28 +0100, Stefano Brivio wrote:
> This patch introduces a PID sharpening factor for faster response after
> association and low activity events.
> 
> 
> Signed-off-by: Stefano Brivio <stefano.brivio@polimi.it>
> Cc: Mattias Nissler <mattias.nissler@gmx.de>
> 
> ---
> 
> Ok, this isn't a no-op anymore. :)
> 
> ---
> 
> Index: wireless-2.6/net/mac80211/rc80211_pid.c
> ===================================================================
> --- wireless-2.6.orig/net/mac80211/rc80211_pid.c
> +++ wireless-2.6/net/mac80211/rc80211_pid.c
> @@ -23,13 +23,16 @@
>   *
>   * The controller basically computes the following:
>   *
> - * adj = CP * err + CI * err_avg + CD * (err - last_err)
> + * adj = CP * err + CI * err_avg + CD * (err - last_err) * (1 + sharpening)
>   *
>   * where
>   * 	adj	adjustment value that is used to switch TX rate (see below)
>   * 	err	current error: target vs. current failed frames percentage
>   * 	last_err	last error
>   * 	err_avg	average (i.e. poor man's integral) of recent errors
> + *	sharpening	non-zero when fast response is needed (i.e. right after
> + *			association or no frames sent for a long time), heading
> + * 			to zero over time
>   * 	CP	Proportional coefficient
>   * 	CI	Integral coefficient
>   * 	CD	Derivative coefficient
> @@ -65,6 +68,11 @@
>  #define RC_PID_SMOOTHING_SHIFT 3
>  #define RC_PID_SMOOTHING (1 << RC_PID_SMOOTHING_SHIFT)
>  
> +/* Sharpening factor (used for D part of PID controller) */
> +#define RATE_CONTROL_SHARPENING_SHIFT 2
> +#define RATE_CONTROL_SHARPENING (1 << RATE_CONTROL_SHARPENING_SHIFT)
> +#define RATE_CONTROL_SHARPENING_DURATION 1
> +
>  /* Fixed point arithmetic shifting amount. */
>  #define RC_PID_ARITH_SHIFT 8
>  
> @@ -131,8 +139,11 @@ struct rc_pid_sta_info {
>  	 */
>  	s32 err_avg_sc;
>  
> -	/* Last framed failes percentage sample */
> +	/* Last framed failes percentage sample. */
>  	u32 last_pf;
> +
> +	/* Sharpening needed. */
> +	u8 sharp_cnt;
>  };
>  
>  /* Algorithm parameters. We keep them on a per-algorithm approach, so they can
> @@ -282,19 +293,19 @@ static void rate_control_pid_sample(stru
>  
>  	mode = local->oper_hw_mode;
>  	spinfo = sta->rate_ctrl_priv;
> +
> +	/* In case nothing happened during the previous control interval, turn
> +	 * on the sharpening factor. */
> +	if (jiffies - spinfo->last_sample > RC_PID_INTERVAL)
> +		spinfo->sharp_cnt = RATE_CONTROL_SHARPENING_DURATION;
> +
>  	spinfo->last_sample = jiffies;
>  
> -	/* If no frames were transmitted, we assume the old sample is
> -	 * still a good measurement and copy it. */
> -	if (spinfo->tx_num_xmit == 0)
> -		pf = spinfo->last_pf;

Please don't remove this check. We can never know when anybody starts
calling rate_control_pid_sample() without packets transmitted. Then it's
good to have it, else we divide by zero in the next line.

> -	else {
> -		pf = spinfo->tx_num_failed * 100 / spinfo->tx_num_xmit;
> -		pf <<= RC_PID_ARITH_SHIFT;
> +	pf = spinfo->tx_num_failed * 100 / spinfo->tx_num_xmit;
> +	pf <<= RC_PID_ARITH_SHIFT;
>  
> -		spinfo->tx_num_xmit = 0;
> -		spinfo->tx_num_failed = 0;
> -	}
> +	spinfo->tx_num_xmit = 0;
> +	spinfo->tx_num_failed = 0;
>  
>  	/* If we just switched rate, update the rate behaviour info. */
>  	if (pinfo->oldrate != sta->txrate) {
> @@ -321,8 +332,11 @@ static void rate_control_pid_sample(stru
>  	spinfo->err_avg_sc = spinfo->err_avg_sc - err_avg + err_prop;
>  	err_int = spinfo->err_avg_sc >> RC_PID_SMOOTHING_SHIFT;
>  
> -	err_der = pf - spinfo->last_pf;
> +	err_der = pf - spinfo->last_pf
> +		  * (1 + RATE_CONTROL_SHARPENING * spinfo->sharp_cnt);
>  	spinfo->last_pf = pf;
> +	if (spinfo->sharp_cnt)
> +			spinfo->sharp_cnt--;
>  
>  	/* Compute the controller output. */
>  	adj = (err_prop * pinfo->coeff_p + err_int * pinfo->coeff_i
> 
> 


  reply	other threads:[~2007-12-10  6:28 UTC|newest]

Thread overview: 45+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-12-09 20:15 [RFC/T][PATCH 0/3] rc80211-pid: PID controller enhancements Stefano Brivio
2007-12-09 20:19 ` [RFC/T][PATCH 1/3] rc80211-pid: introduce rate behaviour learning algorithm Stefano Brivio
2007-12-09 22:25   ` Mattias Nissler
2007-12-09 23:21     ` Stefano Brivio
2007-12-10  0:17       ` Stefano Brivio
2007-12-10  2:24       ` [RFC/T][PATCH v2 " Stefano Brivio
2007-12-10  6:51         ` Mattias Nissler
2007-12-10  7:23           ` Stefano Brivio
2007-12-11 23:29           ` [RFC/T][PATCH v3 " Stefano Brivio
2007-12-12  0:25             ` [RFC/T][PATCH v4 " Stefano Brivio
2007-12-10  6:48       ` [RFC/T][PATCH " Mattias Nissler
2007-12-10  8:03         ` Stefano Brivio
2007-12-10 20:48           ` Mattias Nissler
2007-12-10 20:56           ` Mattias Nissler
2007-12-10 21:30             ` Stefano Brivio
2007-12-10 22:05               ` Mattias Nissler
2007-12-10  8:08     ` Stefano Brivio
2007-12-10 20:51       ` Mattias Nissler
2007-12-10 21:22         ` Stefano Brivio
2007-12-10 21:31           ` st3
2007-12-10 22:09           ` Mattias Nissler
2007-12-11 14:52         ` Johannes Berg
2007-12-11 17:23           ` Mattias Nissler
2007-12-12 17:13             ` Johannes Berg
2007-12-12 20:06               ` Mattias Nissler
2007-12-12 21:34                 ` Stefano Brivio
2007-12-13 11:42                 ` Johannes Berg
2007-12-14  5:27                   ` Jouni Malinen
2007-12-14 12:09                     ` Johannes Berg
2007-12-13  8:00             ` Holger Schurig
2007-12-11 14:51       ` Johannes Berg
2007-12-09 20:21 ` [RFC/T][PATCH 2/3] rc80211-pid: introduce PID sharpening factor Stefano Brivio
2007-12-09 22:29   ` Mattias Nissler
2007-12-09 23:31     ` Stefano Brivio
2007-12-09 23:53       ` Mattias Nissler
2007-12-10  2:28         ` [RFC/T][PATCH v2 " Stefano Brivio
2007-12-10  6:28           ` Mattias Nissler [this message]
2007-12-10  7:21             ` Stefano Brivio
2007-12-10  7:44               ` Mattias Nissler
2007-12-10  8:17                 ` Stefano Brivio
2007-12-11 23:31                 ` [RFC/T][PATCH v3 " Stefano Brivio
2007-12-09 20:28 ` [RFC/T][PATCH 3/3] rc80211-pid: allow for parameters to be set through sysfs Stefano Brivio
2007-12-09 22:30   ` Mattias Nissler
2007-12-10  2:31     ` [RFC/T][PATCH v2 " Stefano Brivio
2007-12-16  9:40   ` [RFC/T][PATCH " Stefano Brivio

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1197268081.7490.10.camel@localhost \
    --to=mattias.nissler@gmx.de \
    --cc=johannes@sipsolutions.net \
    --cc=linux-wireless@vger.kernel.org \
    --cc=linville@tuxdriver.com \
    --cc=stefano.brivio@polimi.it \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).