From: James Prestwood <prestwoj@gmail.com>
To: Denis Kenzior <denkenz@gmail.com>, iwd@lists.linux.dev
Subject: Re: [PATCH v2 2/4] knownnetworks: sort known frequencies by BSS rank
Date: Wed, 24 Jan 2024 10:33:55 -0800 [thread overview]
Message-ID: <ac057182-4b8e-44ef-ac32-9b039d13e45f@gmail.com> (raw)
In-Reply-To: <703a48ba-41f8-4104-bb1d-6b017ae76e15@gmail.com>
Hi Denis,
On 1/24/24 10:10 AM, Denis Kenzior wrote:
> Hi James,
>
> On 1/24/24 07:39, James Prestwood wrote:
>> Currently a quick scan uses the entire known frequency list so
>> ordering really doesn't matter but improvements will be made here
>> to make quick scans "quicker" for large network deployments (with
>> many known frequencies).
>>
>> To prepare for this the known frequency list has been changed to
>> be sorted by BSS rank rather than most recently seen. This makes
>> a lot more sense because IWD should prefer to scan frequencies
>> that had higher ranked BSS's, not just frequencies that were scanned
>> last on the most recent scan.
>
> Interesting. I definitely agree with the argument for treating
> frequencies with higher-ranked candidates as preferred, but I'm not
> sure that directly basing the frequency preference on the bss rank is
> the right thing to do. A client might start right under an AP,
> ranking it quite high, but then rapidly move away where no APs exist
> on this frequency. So you end up always scanning the original AP's
> frequency. Some sort of temporal rank decay would be needed?
Yeah, I see your point...
>
> Perhaps an easier way to accomplish this would be to add known
> frequencies in reverse bss->rank sorted order. That way last seen
> frequency with best ranked BSS would be first?
I'm not sure I understand, how would this be any different than just
reversing the list I already have? Or are you saying sort by both rank
and last seen? Essentially have the list segmented into chunks of
frequencies found with prior scans:
[Last Scan] [N - 1] [N - 2] [ etc ]
And if a duplicate is found, move it to the front, sorted by rank?
>
> Do note however that the quick scan approach is really meant for
> smallish networks, to cut down scanning time when moving between Home
> / Office location for example. It works surprisingly well, but might
> not be appropriate for rapid roaming in a large network.
Exactly, e.g. with 50 APs (assuming you've seen them all) a "quick" scan
takes 5 seconds.
>
>>
>> As far as the disk sync goes the ranking is not included, but
>> ordering is. This really isn't a limitation because when IWD starts
>> up there isn't any guarantee its in the same physical location so
>> old scan ranks are likely not valid anymore. The first set of scans
>> will begin replacing the frequencies loaded from disk.
>> ---
>> src/knownnetworks.c | 16 ++++++++++++++--
>> src/knownnetworks.h | 1 +
>> 2 files changed, 15 insertions(+), 2 deletions(-)
>>
>
> Regards,
> -Denis
>
next prev parent reply other threads:[~2024-01-24 18:33 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-01-24 13:39 [PATCH v2 1/4] knownnetworks: pass scan_bss to known_network_add_frequency James Prestwood
2024-01-24 13:39 ` [PATCH v2 2/4] knownnetworks: sort known frequencies by BSS rank James Prestwood
2024-01-24 18:10 ` Denis Kenzior
2024-01-24 18:33 ` James Prestwood [this message]
2024-01-24 18:44 ` Denis Kenzior
2024-01-24 18:55 ` James Prestwood
2024-01-24 19:06 ` Denis Kenzior
2024-01-25 13:21 ` James Prestwood
2024-01-25 15:39 ` Denis Kenzior
2024-01-24 13:40 ` [PATCH v2 3/4] station: knownnetworks: limit quick scans to 5 freqs per network James Prestwood
2024-01-24 18:21 ` Denis Kenzior
2024-01-24 13:40 ` [PATCH v2 4/4] auto-t: add test for known frequency sorting/maximum James Prestwood
2024-01-24 18:16 ` [PATCH v2 1/4] knownnetworks: pass scan_bss to known_network_add_frequency Denis Kenzior
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=ac057182-4b8e-44ef-ac32-9b039d13e45f@gmail.com \
--to=prestwoj@gmail.com \
--cc=denkenz@gmail.com \
--cc=iwd@lists.linux.dev \
/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