ePMP vs the "new" Ubiquiti?

Products ePMP
#1

I'm moving in to a new area in the new year - it's a suburban micropop style environment.  I'd previously ruled out Ubiquiti because of their poor performance and reliability but it seems that, now they have fixed frame and GPS across the line, their AC product line is a good option (from what I've read).

I like ePMP a lot.  I like Cambium a lot.  I like the upcoming 3K range (albeit too far away to consider for this project).  But apparently UBNT now have a fairly stable product, that has most of the important features of ePMP, that can do 300M+ per 40mhz sector, has a good range of CPE options, and is lower cost.  All of which is perfect for my micropop requirement to deliver 100M plans.

Any advice on why I would go ePMP current gen over UBNT gen2?

Rich

Cambium
#2

Hello,

Just to sort things out .

Gen 2 means it is second generation of UBNT AC products.

But It is not 802.11ac wave 2 chipset.

80 MHz channel bandwidth and QAM-256 modulation support are the only major benуfits of 802.11ac Wave 1 in comparison with 802.11n.

However in real life it is not easy to get SINR level which will allow your link to operate with QAM-256 modulation.

80 MHz certainly is advantage in case you have 80 MHz of continuous spectrum available.

Please find document attached with comparison of throughput tests of ePMP2000 and UBNT Rocket AC.

Tests were performed on clean channel and with co-cannel interference, with different number of subscriber modules connected, with fixed QAM-64 modulation and without modulation limitations. 

PTP550, ePMP3000 and F300 are 802.11ac Wave 2 products.

Wave 2 enables the following

  • Non contiguous channel bonding
  • MU-MIMO operation
  • 80 MHz channel
  • QAM-256

ePMP1000 subscriber modules will be compatible with ePMP3000.

Thank you.

10 Likes
#3

to further add to Fedors response:

Ubnt still is a chance on firmware, many of the 8.x versions are not stable.   They've had a laundry list of major performance affecting security issues in the recent past (the single biggest reason we stopped using their products).

the EPMP product line is much more matured

the UBNT AC doesn't have a path to wave2 or at least as far as I have seen,   the epmp line already has a path to AX past AC wave2. 

the epmp does much better with path obstructions.   (we've got cpes registered at -60 running MCS15 passing through homes and trees moving 100 mbps. we did well to get 10 to 20 Mbps with ubnt AC gear in those conditions)   ubnt may have fixed that by now though. 

the epmp line does much, much better with noise.

I'm sure there are situations where the UBNT AC will outperform the epmp line in PTMP, but our experience has been the epmp line outperforms in real-world situations consistently. 

9 Likes
#4

Hi.

currently we have a similar scenario (line of sight compromise by trees) working with FSK in 900 MHz, but need to increase the link throughput... Can share the experience with ePMP working whit this kind of obstructed links?

#5

Omar, we replaced our entire pmp100 900mhz with epmp 2.4 ghz.

Most 900 cpes registered 70db or better will lock at a similar power with a kp reflector attached. Of course there are some exceptions to that, but overall i was shocked at the penetration at first. Test a few cpes to besure your results are similar.

3 Likes
#6

I'd recommend you read this post from last year about several people's experinces with ePMP 2.4ghz gear.  We also have changed many 900Mhz deployments to 2.4Ghz.

http://community.cambiumnetworks.com/t5/ePMP-2000-and-1000/Does-2-4GHz-makes-sense/td-p/51783

Yes, all things being equal, lower frequencies (900Mhz) should penetrate trees better. But the ePMP 2.4 ghz gear can output higher power, can user much higher gain antennas, and most importantly - can get MUCH higher troughput at the same signals.

As Chris has said, if you're currently getting  a -70 with FSK gear, that's likely with what... an 11 dBi yagi on the SM maybe?  And you'll get 3.3 Mbit of thoughput under perfect conditions, shared among all subscrbers.  With ePMP @ 2.4GHz, we will mate a connectorized SM with a 22 dBi Dual Pol Grid, and the AP/SM will often connect at that same -70 or so.  So what you're losing in propogation, you're gaining in real gain. And again, most importantly - usable thoughput - that AP can have 20 or 30 or 60 Mbit of usable throughput, instead of 3 Mbit.  :)

Plus, you can further help a low signal @ 2.4Ghz by narrowing your channel width sometimes to.  Again, your mileage may vary, but if you go down to a 10 Mhz wide channel, that can increase gain by a couple db in some cases.  You will lose a bit on the maximum throughput capability, but may gain some back on the 'gain' in the system.  Likewise, sometimes even going down to a 5Mhz channel can again increase the final SNR, even at the expense of halving the theoretical throughput.  But, if you have a low user count sector on FSK right now, then absolutely it's very very possible to replace that with a 2.4Ghz ePMP and even at 5Mhz widths (similar spectrum width), and to often shoot through just as many trees, and get similar signals, and to increase bandwidth by 5 times.

For us, we've kinda standardized on 10Mhz channel widths in 2.4Ghz and it's working really really really well for all but the worst conditions. At the very least, we've been able to move 75% of our 900Mhz FSK people over to ePMP 2.4, and those people are MUCH happier. And, the remaining 25% who are stuck on 900Mhz... those people now don't have to share the 3.3 Mbit agregate throughut with all those people anymore, so their internet works better, even without being abel to upgrade them. :)


5 Likes
#7

UBNT forums is following this thread... pretty interesting to see their comments. I've yet to see UBNT staff actually provide any substantive feedback or testing... the response is: "I don't believe the comparison posted on their forum is valid." I, for one, would like to see some actual testing on their side, but I don't think that they'll do this... ever... and I think this says a lot about their company, and not in a postive way.

1 Like
#8

The UBNT users bring up some interesting discussion though, especially considering antennas, price, and the tests limiting the UBNT gear to 64QAM.  However, I think that the fact that it's a competition now between Cambium using last-gen chipset and UBNT with their new AC chipset sets us up for Cambium blowing Ubiquiti out of the water once the Force 300 / ePMP 3000 are out.  UBNT's answer is going to be Airfiber LTU, but I'm skeptical about how much better that will really be than the Airfiber X line, not to mention that the firmware for PTMP won't be available for the LTU line right off the bat.  I think that the UBNT AC line and ePMP are too difficult to compare apple-to-apples right now considering all the different variables, but I'm predicting that ePMP 3000 will outperform UBNT's "Wave 2" AC products at the very least.  LTU is much more of a wildcard

#9

Hi all,

Thanks for your replies and advice.

I know vendor choice is a passionate topic so sorry if anyone got heated. My intent is purely to decide on what technology will service this new area, give me the best option for future migration, and give me the best profit. Like I said above, I had previously discarded UBNT because performance and capacity of their legacy AP's was not good, but the stories I hear of their fixed frame had me intrigued.

Over the holidays I gave this quite a lot of thought, and expanded my thinking beyond raw AP capacity to being more strategic about how each of the 4 main vendors (Cambium, Ubiquiti, Mimosa and Mikrotik) will service my business in the coming year+. I'll share what I came up with...

My business requirements are typical (sub)urban micropop:

  • Provide 100M plans now
  • Ability to provide 200M plans in future
  • Have a technology migration path from now to future
  • Competing with cable operators @ 100M+

My environment is:

  • Low noise (surprising given the density, but that's how it is) so 80MHZ is feasible
  • Max radius 1km (mostly limited by geography)
  • AP's colocated on residential properties / shops / street assets.

It seemed obvious that I would have to either deploy a technology that did 80Mhz PtMP out of the box, or go with an existing 40Mhz solution and upgrade when the 80Mhz products became available. Either is fine as I intend on "only" providing 100M plans for now, but want future options.

I weighed many factors, the most important being:

  • Solution must either be 80Mhz capable now, or provide an upgrade path.
  • Non-contiguous channels (i.e. 40+40) - not mandatory, but I think it will become important.
  • GPS
  • Stable platform
  • Ability to use cheaper CPE for lower speed clients
  • Cost

Given that I now have some criteria defined for my choice of platform, I applied it to the current and upcoming products from the 4 vendors as follows:

  • The deployment has to be same vendor now and in future. Deploying vendor "X" now and moving to vendor "Y" when they have an 80MHZ platform available is not practical because we would have to effectively deploy two networks (old and new) using twice the spectrum, and twice the colocation costs. This means our technology roadmap became a shortlist of: 
    1. Cambium e2K now; migrating to e3K in future
    2. Mimosa
    3. Mikrotik
  • Ubiquiti was effectively ruled out because I don't see any wave2 path for them, and I don't believe their LTU gear is backward compatible. So the same problem exists as if we were to go with vendor X then vendor Y - 2 incompatible networks side by side. Plus given past history I am suspicious that LTU will be usable for a further 1 year after launch. There is speculation that their 11AC product line can (and will?) support 80MHZ in PtMP mode.

  • Mimosa has 80Mhz gear now, but I ruled it out because (a) the whole solution is expensive; (b) I have no idea of their roadmap beyond the current products (if any); (c) their product hasn't yet matured beyond basic functionality; and (d) gut feeling - so far they've been a 1 trick pony and nothing new since.

  • Mikrotik ticks all the boxes except GPS. I appreciate it's criticised for lack of wifi development, but in a sub-1km low noise environment it can still work well. And they have a wave2 CPE available now. Assume an AP will be forthcoming, which combined with horns may overcome the lack of GPS. Given the extremely low cost I am prepared to make some sacrifices, such as limiting AP capacity, and choice of antenna. They're working on sync, albeit without GPS, and whilst it has issues, they're obviously developing something on this line. Based on my cost models, using Mikrotik means this new area will be 4x more profitable than with anyone else. But there's risk that the wifi will start to suffer.

  • Cambium ticks all the boxes (although yet to see on F300 price). Arguably the e3K is a new platform and you could apply the same point I made about stability of UBNT LTU, but I have more confidence that Cambium will pull off a reasonably good product out the gate with only a handful of non-showstopping teething issues. I can deploy F180's now, F300's in near future, and upgrade AP's to e3k down the line whilst maintaining a single network using the same spectrum and same towers.

So my conclusion is that Cambium is the correct choice. Despite it's 11n chipset, it will deliver my 100M plan today and give me a clear migration path for future. By time I have capacity issues with 11n, the 11ac wave 2 will be out. Mikrotik will be an experiement I'll work on. Despite having faster products today, both Ubiquiti and Mimosa doesn't
give me any viable path forward.

Rich

#10

Solid reasoning Nekomata! Sucks to see some of the name calling and quick accusations in the other forum. Not at all helpful to end users who could see benifit from the ubnt platform. Not to mention the cambium test did show sistuations when the ubnt gear did better. I’ll never understand the attacks when it comes to helping other operators pick whats best for them. One of the minor reasons we turned away from ubnt, its difficult to find clear useful information to questions like this. Each vendor has its nitch, and it would be so much more helpfully to see those adantages expressed by those forum members marked as top contributors rather than insults and accusations of rigged results. Show off what each vendor does best instead!

5 Likes
#11

Hi all,

Thanks for the discussion and input. I've had some time over the holidays to think this through and focus on our business drivers rather than which technology is the latest.  I've realised that picking a vendor with the latest whiz-bang gadget and fastest speed is wrong and I should focus on what enables us to transform our business to accomodate the growth in bandwidth in the future.

Here's what I came up with...

Business requirements

  • Provide 100M service now
  • Provide 200M (or higher) in future
  • Compete with cable operators

Remember, that I'm looking at an urban micropop setup, so short distances, high speed packages to compete with wireline services.

Based on this it was obvious I need a solution that handles 80mhz, or has a path to such a solution. Despite being urban, we have very little noise so 80mhz is feasible.  We're unlikely to use beamstearing antennas due to the antenna size and where we are hosting our AP's.

This gave us the following selection criteria:

  • Has 80mhz or roadmap to it
  • GPS
  • Low CPE cost
  • If 80mhz is in the future, must have migration path

Now that last point is golden for our business. If we deploy a non 80mhz product now and have to upgrade in the future, they must both co-exist using the same AP.  We CANNOT run two networks side by side as this will consume twice the spectrum, and we would unlikely be able to negotiate mounting space for twice the AP's.  So considering the 4 vendors that are in my target CPE price range (Cambium, Mikrotik, Mimosa, Ubiquiti) I concluded that...

  • Ubiquiti: Doesn't have 80mhz now in AC product line (although some discussion that it can be enabled by hacking the config via shell). No roadmap that I've seen for wave2, only LTU. Therefore if we deploy AC now, to upgrade to a faster platform we would need to run AC and LTU as two seprate networks. Also, LTU is a new platform, and I suspect it may be another year after released before it becomes stable for prime time. This is a bust.
  • Mimosa: Has 80mhz but the solution is quite expensive and our financial models require a much bigger capex investment to get this off the ground. I've also not seen anything from Mimosa that shows a roadmap, so it's unclear that there is anything beyond their first generation of kit (A5's, B series, and C5's). This is a bust.
  • Mikrotik: They've had 80mhz for a long while, and wave 2 is just being released now. Although they don't have GPS and their wireless tech is somewhat behind the curve, it can work with some compromises such as horns, short distances, and keeping the users per AP low. Given that their very low cost gives me more than 4 x the profit than other vendors with around half the capex, it may be worth the compromises. This is a maybe, but needs some experimentation.
  • Cambium: Only 40mhz available now, but this will deliver our 100M package. Has an upgrade path to 80mhz by swapping the AP and keeping old and new CPE's on one network. Has mature GPS. The wave2 stuff is also new, and you could argue the same point that I did for LTU above, but I have more confidence that Cambium will deliver a good product out the gate and any bugs will stabalise much quicker. 4x4 is also interesting, but unsure how I will yet implement this.

So in conclusion, it seems Cambium is the only one that ticks all boxes, even though it's 11n technology today. In our tests we get ~200M downlink in flexible mode, which isn't as good as ~260M from UBNT & MT, but it's enough for now and that extra 20-30% speed isn't as important as having an upgrade path.

Hope my reasoning makes sense, and I believe it echoes what a few of you have been saying above and on other similar threads.

Rich

6 Likes
#12
...even though it's 11n technology today. In our tests we get ~200M downlink in flexible mode...

Rich

Hi Rich.  Yes, the ePMP1000 & 2000 are 11n technology, but I've been testing the new Cambium AC gear.  The Force300 is the only member yet, and the firmware is still Beta - BUT in a 20 Mhz wide channel, with fixed ratio of 75%/25% set, I get about 160 Mbit aggregate throughput.  125+ Mbit down and 35+ Mbit up... in a 20 Mhz wide channel.  We haven't really tested 40 Mhz or 80 Mhz widths yet, but I'm expecting no less than spectactular.

http://community.cambiumnetworks.com/t5/ePMP-2000-and-1000/Force-300-25-first-impression/td-p/86606/page/2

Again, this is still Beta firmware and there are still features missing. Currently, these are made for PTP backhauls only, and there is no '3000 Access Point' ready currently, but I post this just to show you the boost from Cambium's N gear to their AC gear. The Force 300's are really impressing me.  And, unlike other brands we've experienced, where 'beta period' seems to last for years, Cambium is on Release Candidate 52 already. I now have them carrying live traffic on our network, and they are performing wonderfully. :D 

1 Like
#13

Hi Ninedd;


Great to hear and writeup on the switched to ePMP 2.4GHz portfolio from the PMP100 900MHz.  Would mind sharing which sector antenna you used for this migration?  Cambium's solution or the vendor/model?

Regards.

#14

We have two links with epmp force 300 and not impressed at all. Epmp 1000 and 2000 work great but ubiquiti is far ahead in the ac race. Especially with them coming out with the ac equipment a year earlier and how well it works. So far with the force 300 the signal is a lot higher then what it should be with a line of site at this length and wont even stay registered at a 72 to eAlign. Honestly a low end powerbeam ac works better than the force 300 more speed a stability. Also with the 4.1.1 update we lost the gui and had to reset the dish in ssh to get it back. I hope that 3 or 4 firmwares down the road epmp will catch up.

#15

That hasn’t been our Force300 experience. For us, they have been very solid and very spectrum efficient since the last FW update. We only have them running in 20 or 40mhz widths currently, but they link with zero downtime for us.

1 Like
#16
@jlkz wrote:

We have two links with epmp force 300 and not impressed at all. Epmp 1000 and 2000 work great but ubiquiti is far ahead in the ac race. Especially with them coming out with the ac equipment a year earlier and how well it works. So far with the force 300 the signal is a lot higher then what it should be with a line of site at this length and wont even stay registered at a 72 to eAlign. Honestly a low end powerbeam ac works better than the force 300 more speed a stability. Also with the 4.1.1 update we lost the gui and had to reset the dish in ssh to get it back. I hope that 3 or 4 firmwares down the road epmp will catch up.

signal strength isn't to blame on the radio,   link budget is a link budget. like gain and conducted power on equipment will land in the same general RSSI regardless of brand, if you're seeing a major difference from these than a ubnt product with like gain and transmit power, something isn't right. possibly the alignment or tx power was never turned up.     the f300 has a long way to go, but keep in mind its a gen 2 AC radio, UBNT is a gen 1.   gen 1 chips have been around much longer so I'd expect them to be much more mature at this time.   once the epmp3000 if released and made to run at its potential, won't be much of a comparison between ubnt and epmp in AC devices.   with UBNTs focus on the LTU, that's where the battle for top dog is going to come into play.   cambium went 4x4 on the AP and ubnt is trying to code to 4096.  my experience with the AF5xHD is it needs as much development as the cambium products at this point.  remember the f300 is meant to be a CPE and light PTP radio, its only half of the line in its early days.  seems to be the trend with all of the vendors right now. 

1 Like
#17

Anyone tried the 60Ghz ignitenet stuff for Micropops ?  2.5Gb at the AP  seems pretty sweet if you can feed your PoPs with fiber. And they have 5Ghz built in (and 2.4Ghz I think) that you can use as fallover or for customers that the 60Ghz can't reach correct ?  60Ghz is really short range but that's the whole point of micropops.

I think we are going to set on of these up and see what it can do. Running fiber to the neighborhood and then 60Ghz to the home is insainly cheaper than taking the fiber all the way to the home.

Wonder if there is any chance we will see a 60Ghz cabmium / ePMP solution any time soon or if they are going to continue to be a generation behind everyone else. 

#18

ubiquiti has canceled the LTU proyect?

#19
@jldiaz wrote:

ubiquiti has canceled the LTU proyect?

Supposedly the news of canecllation was an accidental/confused email sent to a European distributor. Multiple UBNT staff working on the LTU project have stated that this is not accurate and that the product is still very much alive and progressing. You can find the thread in the UBNT forums discussing it HERE.

#20
@brubble1 wrote:

Anyone tried the 60Ghz ignitenet stuff for Micropops ?  2.5Gb at the AP  seems pretty sweet if you can feed your PoPs with fiber. And they have 5Ghz built in (and 2.4Ghz I think) that you can use as fallover or for customers that the 60Ghz can't reach correct ?  60Ghz is really short range but that's the whole point of micropops.

I think we are going to set on of these up and see what it can do. Running fiber to the neighborhood and then 60Ghz to the home is insainly cheaper than taking the fiber all the way to the home.

Wonder if there is any chance we will see a 60Ghz cabmium / ePMP solution any time soon or if they are going to continue to be a generation behind everyone else. 

We are looking to do some of this. Our local city has a fiber ring. They are currently running fiber to their water towers for us install wireless gear. The downtown area runs parallel to one of the tanks 1500-2500' away from the tank. We plan to use Ignitenet 60Ghz for internet and phone service to the businesses on main street. 

I do a lot of Motorola contract work, and my sources with Nokia tell me Cambium is working with Siklu on a 60Ghz product line.