[Cerowrt-devel] dlte

[Cerowrt-devel] dlte

[Dave Taht]

I expect dave reed to comment, so I'll withhold mine for now

https://kurti.sh/pubs/dLTE-Johnson-HotNets-2018.pdf


-- 

Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-205-9740

Re: [Cerowrt-devel] dlte

[Mikael Abrahamsson]

On Tue, 4 Dec 2018, Dave Taht wrote:

> I expect dave reed to comment, so I'll withhold mine for now
>
> https://kurti.sh/pubs/dLTE-Johnson-HotNets-2018.pdf

When I read the first page I was hopeful, then unfortunately I got 
disappointed and just quickly scanned the rest. It's still tunneled and 
the same architecture, just more distributed.

I'd prefer if the device had two radios and could handle its own handover 
for most traffic. Most traffic today is streaming video so when you want 
to do handover just attach to the new radio, get new address, deprecate 
the old address, open new connections using the new address/radio close 
connections based on the old address, and move on.

Traffic that needs to be always the same can be tunneled, but most traffic 
has no need for this. My guesstimate is that 90-95% of traffic doesn't 
need mobility and doesn't need to go via some tunnel to some core-node.

It's also frustrating to talk "wifi" with mobile people, because they 
conflate "wifi" into multiple things at once. For me it's just that people 
want well working wireless access, and they couldn't care less how they 
get it. The more seamless the better. In that aspect this proposal is kind 
of neat in that I could make my residential radios join a larger network 
(similar to eduroam).

So yes, I think this proposal has some merit.

-- 
Mikael Abrahamsson    email: swmike@swm.pp.se

Re: [Cerowrt-devel] dlte

[Toke Høiland-Jørgensen]

Mikael Abrahamsson <swmike@swm.pp.se> writes:

> On Tue, 4 Dec 2018, Dave Taht wrote:
>
>> I expect dave reed to comment, so I'll withhold mine for now
>>
>> https://kurti.sh/pubs/dLTE-Johnson-HotNets-2018.pdf
>
> When I read the first page I was hopeful, then unfortunately I got 
> disappointed and just quickly scanned the rest. It's still tunneled and 
> the same architecture, just more distributed.
>
> I'd prefer if the device had two radios and could handle its own handover 
> for most traffic. Most traffic today is streaming video so when you want 
> to do handover just attach to the new radio, get new address, deprecate 
> the old address, open new connections using the new address/radio close 
> connections based on the old address, and move on.
>
> Traffic that needs to be always the same can be tunneled,

Totally agree. I'll add that even persistent "connections" can do
end-to-end mobility with changing addresses. Mosh is an example of an
application that does this right. I believe Apple uses MPTCP to achieve
the same thing with TCP...

-Toke

Re: [Cerowrt-devel] dlte

[Toke Høiland-Jørgensen]

Mikael Abrahamsson <swmike@swm.pp.se> writes:

> On Tue, 4 Dec 2018, Dave Taht wrote:
>
>> I expect dave reed to comment, so I'll withhold mine for now
>>
>> https://kurti.sh/pubs/dLTE-Johnson-HotNets-2018.pdf
>
> When I read the first page I was hopeful, then unfortunately I got 
> disappointed and just quickly scanned the rest. It's still tunneled and 
> the same architecture, just more distributed.

OK, now I read the paper, and I think you may have missed the part where
they say that they terminate the tunnelling at the AP and assign new IPs
whenever a client roams. So it's basically WiFi APs over the LTE
layer-2... Which is pretty cool, I think :)

-Toke

Re: [Cerowrt-devel] dlte

[Mikael Abrahamsson]

[-- Attachment #1: Type: text/plain, Size: 939 bytes --]

On Wed, 5 Dec 2018, Toke Høiland-Jørgensen wrote:

> Mikael Abrahamsson <swmike@swm.pp.se> writes:
>
>> On Tue, 4 Dec 2018, Dave Taht wrote:
>>
>>> I expect dave reed to comment, so I'll withhold mine for now
>>>
>>> https://kurti.sh/pubs/dLTE-Johnson-HotNets-2018.pdf
>>
>> When I read the first page I was hopeful, then unfortunately I got
>> disappointed and just quickly scanned the rest. It's still tunneled and
>> the same architecture, just more distributed.
>
> OK, now I read the paper, and I think you may have missed the part where
> they say that they terminate the tunnelling at the AP and assign new IPs
> whenever a client roams. So it's basically WiFi APs over the LTE
> layer-2... Which is pretty cool, I think :)

Thanks, I now looked again and 4.1 said "local cores" and that's when my 
eyes rolled and I skipped on. But 4.2 is indeed what I have been talking 
about...

-- 
Mikael Abrahamsson    email: swmike@swm.pp.se

Re: [Cerowrt-devel] dlte

[Dave Taht]

Toke Høiland-Jørgensen <toke@toke.dk> writes:

> Mikael Abrahamsson <swmike@swm.pp.se> writes:
>
>> On Tue, 4 Dec 2018, Dave Taht wrote:
>>
>>> I expect dave reed to comment, so I'll withhold mine for now
>>>
>>> https://kurti.sh/pubs/dLTE-Johnson-HotNets-2018.pdf
>>
>> When I read the first page I was hopeful, then unfortunately I got 
>> disappointed and just quickly scanned the rest. It's still tunneled and 
>> the same architecture, just more distributed.
>
> OK, now I read the paper, and I think you may have missed the part where
> they say that they terminate the tunnelling at the AP and assign new IPs
> whenever a client roams. So it's basically WiFi APs over the LTE
> layer-2... Which is pretty cool, I think :)

It's still based on the false optimism that users will ever get to own
and control their own LTE AP.

>
> -Toke
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel

Re: [Cerowrt-devel] dlte

[Toke Høiland-Jørgensen]

Dave Taht <dave@taht.net> writes:

> Toke Høiland-Jørgensen <toke@toke.dk> writes:
>
>> Mikael Abrahamsson <swmike@swm.pp.se> writes:
>>
>>> On Tue, 4 Dec 2018, Dave Taht wrote:
>>>
>>>> I expect dave reed to comment, so I'll withhold mine for now
>>>>
>>>> https://kurti.sh/pubs/dLTE-Johnson-HotNets-2018.pdf
>>>
>>> When I read the first page I was hopeful, then unfortunately I got 
>>> disappointed and just quickly scanned the rest. It's still tunneled and 
>>> the same architecture, just more distributed.
>>
>> OK, now I read the paper, and I think you may have missed the part where
>> they say that they terminate the tunnelling at the AP and assign new IPs
>> whenever a client roams. So it's basically WiFi APs over the LTE
>> layer-2... Which is pretty cool, I think :)
>
> It's still based on the false optimism that users will ever get to own
> and control their own LTE AP.

Well, they did say they had done proof of concept tests; and that they
could build a base station for $8000... So might not be completely
impossible...

-Toke

Re: [Cerowrt-devel] dlte

[David P. Reed]

Conquer the spectrum licensing and device certification nexus. Or else your cell is will pwn yr physical world.

LTE over UNII band is not even as good as CSMA at sharing and cooperation, and without coordination at installation planning time, it doesn't work well.

802.11ax has the same fragility in Multi Unit Dwellings due to requiring a radio propagation plan and coordination so neighbors don't completely jam neighbors.

Don't obsess about throughput at the link layer, when the design assumes exclusive rights to transmit.

There are techniques for cooperative space-time-rate multiplexing that scale. LTE licensed or unlicensed or 802.11ax are not such techniques.

Small cells are a fantasy of the carriers that they can put their licensed gear on your property at points they choose. Technically it appears to work in an abstract fantasy prototype. In the real world, it can't scale unless you let the phone company invade your premises and control all your placement of furniture, doors, mirrors, etc.

----Original Message-----
From: "Toke Høiland-Jørgensen" <toke@toke.dk>
Sent: Fri, Dec 7, 2018 at 4:08 am
To: "Dave Taht" <dave@taht.net>
Cc: "Dave Taht" <dave@taht.net>, "cerowrt-devel" <cerowrt-devel@lists.bufferbloat.net>
Subject: Re: [Cerowrt-devel] dlte

Dave Taht  writes:

> Toke Høiland-Jørgensen  writes:
>
>> Mikael Abrahamsson  writes:
>>
>>> On Tue, 4 Dec 2018, Dave Taht wrote:
>>>
>>>> I expect dave reed to comment, so I'll withhold mine for now
>>>>
>>>> https://kurti.sh/pubs/dLTE-Johnson-HotNets-2018.pdf
>>>
>>> When I read the first page I was hopeful, then unfortunately I got 
>>> disappointed and just quickly scanned the rest. It's still tunneled and 
>>> the same architecture, just more distributed.
>>
>> OK, now I read the paper, and I think you may have missed the part where
>> they say that they terminate the tunnelling at the AP and assign new IPs
>> whenever a client roams. So it's basically WiFi APs over the LTE
>> layer-2... Which is pretty cool, I think :)
>
> It's still based on the false optimism that users will ever get to own
> and control their own LTE AP.

Well, they did say they had done proof of concept tests; and that they
could build a base station for $8000... So might not be completely
impossible...

-Toke
_______________________________________________
Cerowrt-devel mailing list
Cerowrt-devel@lists.bufferbloat.net
https://lists.bufferbloat.net/listinfo/cerowrt-devel

Re: [Cerowrt-devel] dlte

[Dave Taht]

"David P. Reed" <dpreed@deepplum.com> writes:

> Conquer the spectrum licensing and device certification nexus. Or else
> your cell is will pwn yr physical world.
>
> LTE over UNII band is not even as good as CSMA at sharing and
> cooperation, and without coordination at installation planning time,
> it doesn't work well.
>
> 802.11ax has the same fragility in Multi Unit Dwellings due to
> requiring a radio propagation plan and coordination so neighbors don't
> completely jam neighbors.

I have not been tracking ax well of late. I had heard they were doing
smarter things with smaller channels, which gave me hope. 

>
> Don't obsess about throughput at the link layer, when the design
> assumes exclusive rights to transmit.

Given how hard it has been to get a mere lte dongle to work right, I am
inclined to disregard a lot of the 5G hype. 

> There are techniques for cooperative space-time-rate multiplexing that
> scale. LTE licensed or unlicensed or 802.11ax are not such techniques.
>
> Small cells are a fantasy of the carriers that they can put their
> licensed gear on your property at points they choose. Technically it
> appears to work in an abstract fantasy prototype. In the real world,
> it can't scale unless you let the phone company invade your premises
> and control all your placement of furniture, doors, mirrors, etc.

Well, I kind of envision that happening, unless we make wifi work really
really well again. I loved learning about low-e glass recently, I think
that is going to interfere with celluar, and make wifi deployments better.

> ----Original Message-----
> From: "Toke Høiland-Jørgensen" <toke@toke.dk>
> Sent: Fri, Dec 7, 2018 at 4:08 am
> To: "Dave Taht" <dave@taht.net>
> Cc: "Dave Taht" <dave@taht.net>, "cerowrt-devel"
> <cerowrt-devel@lists.bufferbloat.net>
> Subject: Re: [Cerowrt-devel] dlte
>
> Dave Taht  writes:
>
>> Toke Høiland-Jørgensen  writes:
>>
>>> Mikael Abrahamsson  writes:
>>>
>>>> On Tue, 4 Dec 2018, Dave Taht wrote:
>>>>
>>>>> I expect dave reed to comment, so I'll withhold mine for now
>>>>>
>>>>> https://kurti.sh/pubs/dLTE-Johnson-HotNets-2018.pdf
>>>>
>>>> When I read the first page I was hopeful, then unfortunately I got
>>>> disappointed and just quickly scanned the rest. It's still
>>>> tunneled and
>>>> the same architecture, just more distributed.
>>>
>>> OK, now I read the paper, and I think you may have missed the part
>>> where
>>> they say that they terminate the tunnelling at the AP and assign
>>> new IPs
>>> whenever a client roams. So it's basically WiFi APs over the LTE
>>> layer-2... Which is pretty cool, I think :)
>>
>> It's still based on the false optimism that users will ever get to
>> own
>> and control their own LTE AP.
>
> Well, they did say they had done proof of concept tests; and that they
> could build a base station for $8000... So might not be completely
> impossible...
>
> -Toke
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel

Re: [Cerowrt-devel] dlte

[Aaron Wood]

[-- Attachment #1: Type: text/plain, Size: 598 bytes --]

On Tue, Dec 11, 2018 at 11:02 Dave Taht

>
> Well, I kind of envision that happening, unless we make wifi work really
> really well again. I loved learning about low-e glass recently, I think
> that is going to interfere with celluar, and make wifi deployments better.


It definitely messes with cellular.  I’ve been in multiple rural houses
where cell signal can’t get through the windows, but open a door and
suddenly it works.

Great for bottling up WiFi in the house, too, but may need a set of APs for
coverage if the house is L or H shaped.

> --
- Sent from my iPhone.

[-- Attachment #2: Type: text/html, Size: 1074 bytes --]

Re: [Cerowrt-devel] dlte

[Dave Taht]

Aaron Wood <woody77@gmail.com> writes:

> On Tue, Dec 11, 2018 at 11:02 Dave Taht 
>
>     Well, I kind of envision that happening, unless we make wifi work
>     really
>     really well again. I loved learning about low-e glass recently, I
>     think
>     that is going to interfere with celluar, and make wifi deployments
>     better.
>
> It definitely messes with cellular. I’ve been in multiple rural houses
> where cell signal can’t get through the windows, but open a door and
> suddenly it works.
>
> Great for bottling up WiFi in the house, too, but may need a set of
> APs for coverage if the house is L or H shaped.

This is weirdly kind of awesome.