[LEDE-DEV] portal wifi router

[Bill Moffitt]

On 6/22/2016 10:41 AM, Dave Taht wrote:
> eero shipped something debian based (using batman-adv to mesh, I believe)
> google onhub shipped some sort of hybrid of chromeos (so far as I know)
> https://www.plumewifi.com/ is making all sorts of promises, can't tell
> what they are using.
>
> A new one just cropped up - portal - where their kickstarter claims
> they are using openwrt. Anyone know anything more about it? Not clear
> what chipset they are using but since they are based in taiwan I'd
> guess it's mediatek, but it could be qualcomm?
>
> https://www.kickstarter.com/projects/portalwifi/portal-turbocharged-wifi?ref=email
>
> http://getportal.io/ has more "details" and early backers should be
> seeing routers soon.
>
> They are all claiming massive improvements in wifi performance mostly
> based on whatever 802.11ac based technology they are using, I keep
> hoping at least some of the above were also paying attention to bql,
> fq_codel, per station queuing, etc.
>
Dave-

Thanks for pointing those out (I'll volunteer that Ayrstone AyrMesh is 
also, of course, based on OpenWRT and Open80211s, for those on U.S. and 
Canadian farms. I don't claim that we're improving on anything but 
getting moderate-bandwidth LAN connectivity over very large rural areas).

I love the idea of meshing WiFi systems adapting their channel settings 
around local interference... so the Plume in my house detects that my 
next door neighbor's Portal is interfering, so it changes its channel. 
However, the Plume in my garage is much closer to the neighbor behind me 
than the Plume in my house, and changes the channel to avoid them. This 
then interferes with the neighbor's Portal, which changes its channel to 
avoid my interference, which then interferes with the neighbor behind 
me, who changes channel to avoid that interference...

And, of course, as you point out, this is a situation ripe for 
bufferbloat (we know that mesh networks are potentially, ah, 
troublesome). On the other hand, you, Jim, and others have been banging 
the drum for 6 years or so, if memory serves. So there's no excuse for 
new products coming out to NOT use bql, etc. If the word has gotten down 
to the least capable wireless networking developers (i.e. me), everyone 
should be up to speed by now.

So, on balance, I'm not too worried about these things increasing 
congestion - in a typical apartment building I'm not sure they'll be 
actually be working long enough to congest anything. ;-)

Humorously,

Bill

P.S. for the mathematically inclined, note the discrepancy between the 3 
distinct channels in the 2.4 GHz. band and the application of the 
4-color theorem to channel selection. Sucks...