Further note: You can see I did those calcs at 1024QAM, so reduce that
down the 256QAM for closer to real numbers :)

On Tue, Feb 13, 2018 at 12:42 AM, Josh Reynolds <j...@kyneticwifi.com> wrote:
> Let's break this down a bit.
>
> Firstly, what outdoor PTMP platform is really using WiFi anymore? *shakes 
> head*
>
> Mu-MIMO only works if the clients are sufficiently spread apart
> (physically), and their tx/rx windows can fit into almost the same
> timeframe. Any degradation in signal of one client that ends up in the
> same window as other clients reduces the overall capacity of the AP
> (like in many other situations). It can, in some situations, lead to
> cumulative transfer windows where overall throughput ends up getting
> reduced as the rx/tx hold time for the other clients end up taking a
> hit in efficiency. This is one of the few failings of MU-MIMO, not
> even taking into account "massive" systems like 14x14 that end up
> costing quite a bit in overall power budget due to the number of
> elements, further meaning that your range is severely limited in a
> system like this... so only decent in very dense situations. That's a
> unique niche.
>
> So, 80 clients. That's a pretty average number for a modern system
> (450, Mimosa, AC Prism Gen2).
>
> 30Mbps per client... okay, but most customers are actually streaming.
> Let's throw another margin on top of that and say a few Mbps for
> gaming. 10Mbps is a nice round number. Now, that data gets sent in
> most services in bursts and buffered, so it's not continuous. Let's
> take that average number down to about 8 Mbps. Now let's assume that
> maybe 70% of those 80 customers is doing something like that, and
> that's probably a generous number. 56 customers. So 56 customers x
> 8Mbps = 448Mbps. On a 20Mhz channel? Wait, this doesn't seem to work
> out!
>
> Soo.... 1024 QAM on a 20MHz channel gives you 250Mbps, very roughly.
> If you're optimistic about modern patterns, you're between an 80/20
> and a 60/40 Download/Upload ratio on a split GPS synced system.
>
> 80/20 = 200Mbps Down, 50Mbps Up
> 60/40 = 150 Down, 100Mbps Up
>
> Let's say for the sake of argument that you're in the 80/20 camp,
> giving you 200Mbps to work with in above perfect conditions, gives you
> 3.57 Mbps per subscriber. Roughly 4M/sub, good for 480p streaming.
>
> That's a very expensive platform for that kind of throughput and
> subscriber count with such limitations in range and needed a "perfect
> storm" of client distribution and data patterns to really take
> advantage of. With working GPS in all modern platforms, I would be
> hard pressed to not use an additional 20mhz channel if available, or
> just cut the channel width in half to 10MHz each, and put up 4 Mimosas
> or 4 Gen2 Prism radios and have far more than 4x the possible
> subscriber account, improved tx/rx efficiency, improved range
> (increasing distance and SNR in many situations), and greatly reduced
> cost.
>
> Again, I'm far more excited about the 4x increase in spectral
> efficiency via OFDMA that doesn't cause you to cut down on tx/rx
> chains for multi-client transmission (costing your range, per client
> snr, and per-client throughput in the process). MU-MIMO is and will
> always be a niche hack that never lived up to what was promised.
>
> On Tue, Feb 13, 2018 at 12:12 AM, Sean Heskett <af...@zirkel.us> wrote:
>> Being able to load a 450m AP with 80 subs and deliver 30mbps service to all
>> of them at peak Netflix time in a 20mhz channel without breaking a sweat is
>> worth every penny.
>>
>> But it’s one tool in the tool box and isn’t the best solution for every
>> deployment.
>>
>> 2 cents
>>
>> -sean
>>
>>
>>
>> On Mon, Feb 12, 2018 at 10:32 PM Josh Reynolds <j...@kyneticwifi.com> wrote:
>>>
>>> The more I dig into MU-MIMO, the more I realize it's not all that great.
>>>
>>> I am far more excited by the 9 client simultaneous transmissions in
>>> 802.11ax via OFDMA.
>>>
>>> On Mon, Feb 12, 2018 at 8:00 PM, Adam Moffett <dmmoff...@gmail.com> wrote:
>>> > 450 still does a few things that ePMP doesn't.
>>> > Plus there's that 14 chain MU-MIMO thing......ePMP will probably never
>>> > have
>>> > something like that.
>>> > UI is still sluggish on ePMP.
>>> >
>>> > On the other hand ePMP has gotten so many feature improvements over
>>> > these
>>> > past few years that it's gotten really hard to argue with the value it
>>> > provides.
>>> >
>>> >
>>> > ------ Original Message ------
>>> > From: "Chuck McCown" <ch...@wbmfg.com>
>>> > To: af@afmug.com
>>> > Sent: 2/12/2018 8:27:56 PM
>>> > Subject: Re: [AFMUG] mini wisp
>>> >
>>> > The UI server was probably the worst I have ever seen.
>>> >
>>> > So, less than 25 subs per site, what speed packages do you sell to those
>>> > 25?
>>> >
>>> > Packetflux GPS sync.
>>> >
>>> > From: Joe Novak
>>> > Sent: Monday, February 12, 2018 5:20 PM
>>> > To: af@afmug.com
>>> > Subject: Re: [AFMUG] mini wisp
>>> >
>>> > What didn't you like about it? The interface came a long way since the
>>> > early
>>> > days of EPMP. We've got quite a bit deployed. A lot of people are having
>>> > weird GPS situations come up with the on-board GPS, we have this problem
>>> > once in a while too. Our packetflux sites are rock solid though. That is
>>> > assuming density isn't more then 25 per AP, because then I don't exactly
>>> > have enough experience with it. Most of our APs are sitting right around
>>> > 25
>>> > customers, and according to airtime we still have quite a bit of room.
>>> >
>>> > On Mon, Feb 12, 2018 at 6:09 PM, Jaime Solorza
>>> > <losguyswirel...@gmail.com>
>>> > wrote:
>>> >>
>>> >> I separated frequencies to three I found cleanest on AFx5s...On Rockets
>>> >> and Powerbeams I choose one frequency and shut off the rest on APs and
>>> >> on
>>> >> PowerBeams I only use two...this method has worked well since August of
>>> >> 2017
>>> >> when I replaced all the radios on this network and have had to change
>>> >> them
>>> >> since.  Two of the WISPs live in Fabens and work with us on issues.
>>> >> The
>>> >> other one from El Paso uses my services once in a while and works with
>>> >> us as
>>> >> well.  Texas Gas put up allot of 5GHz units around Fabens but still no
>>> >> issues. I used larger dishes at Wells and lift stations as well.
>>> >>
>>> >> Jaime Solorza
>>> >>
>>> >> On Feb 12, 2018 4:50 PM, "Jaime Solorza" <losguyswirel...@gmail.com>
>>> >> wrote:
>>> >>>
>>> >>> Two AF5x on same tower, One AP on second tower 20 ft away...all other
>>> >>> radios within 4 mile radius...
>>> >>>
>>> >>> Jaime Solorza
>>> >>>
>>> >>> On Feb 12, 2018 4:43 PM, "Chuck McCown" <ch...@wbmfg.com> wrote:
>>> >>>>
>>> >>>> All on the same tower, right?
>>> >>>>
>>> >>>> From: Jaime Solorza
>>> >>>> Sent: Monday, February 12, 2018 4:41 PM
>>> >>>> To: Animal Farm
>>> >>>> Subject: Re: [AFMUG] mini wisp
>>> >>>>
>>> >>>> Yes..I have two AF5X links as PTP and 25 radios all in 5 GHz off 4
>>> >>>> APs
>>> >>>> in Fabens, Texas sharing spectrum with 3 WISPs...no issues...
>>> >>>>
>>> >>>> Jaime Solorza
>>> >>>>
>>> >>>> On Feb 12, 2018 4:32 PM, "Chuck McCown" <ch...@wbmfg.com> wrote:
>>> >>>>>
>>> >>>>> Talking to a friend that wants to build a small wisp.  He is about
>>> >>>>> 5.5
>>> >>>>> miles from a backbone connection.  I would suggest AF5X to him but
>>> >>>>> he is
>>> >>>>> gonna want to use 5 GHz for his wisp I presume.
>>> >>>>>
>>> >>>>> Can an AF5X and some 5 GHz cambium (or others) access points
>>> >>>>> peacefully
>>> >>>>> coexist on a tower?
>>> >>>>> Very rural area.  Not expecting much interference other than home
>>> >>>>> routers.
>>> >
>>> >

Reply via email to