How does the ePMP compare to the new Ubiquiti Airmax AC system

Thought I might jump in with a couple comments.  While I am not an expert, I have been running some side by side tests with a Ubiquiti Rocket 5AC PtMP radio on a Ubiquiti 60 degree AC sector next to an ePMP 1000 GPS into a RF Elements 120 degree sector.

Radios are on a tower 1700' above town.  Users are 1.5 miles to 3 miles out.  RF environment is somewhat quiet. Gets a little noisy on some 5.8 channels at tower.

Test clients for these two radios are wanting 50 - 75 Mbps. This is being tested to be an alternative to cable Internet. Low CPEs per AP (10 - 15 subs). Ubiquiti CPEs are mostly NBE-5AC-19's with a couple PBE-5AC-300's.  Cambium CPEs are currently ePMP 1000 integrated's.  Ordered a couple Force 110's to test.

I have much more experience with Ubiquit gear than Cambium, so my natural tendency is to push for UBNT for this deployment.  

What I have discovered:

- Ubiquiti software is much better for RF surveys and general RF picture.  Especially the AirMax 7 and 8 versions.  Very well done. Big benefit to look at the RF picture at AP and CPE without taking any links down.

- Ubiquiti PtP with AirMax 7 and 8 is very fast and reliable.

- Ubiquiti PtMP is extremely inconsistent. There are too many firmware versions. Few are compatible. Some flat out don't work (AirMax 8 will not connect PtMP at all in our environment).

- As soon as the SNR creeps up, Ubiquiti dies off.  Trust me, I want them to work - simply because of my comfort level and the software.

- Every time I plug in a ePMP radio, it trains up and works.  In speed tests, it beats the Ubiquiti radio 75% of the time and is 100% more reliable.

I've put both UBNT and Cambium CPE's on a tripod and tested them side by side all over town for hours and hours. UBNT is really fast when it works, but it simply has not been reliable enough for us.  Despite my dislike of the Cambium firmware, we are starting to buy more Cambium gear and rolling out more test customers on Cambium CPEs.  And they are happier than the Ubiquiti customers.

My next test customer wants to ditch all television providers and use our wireless for 4K Netflix streams.  He is getting the first Force 110 to test. Luckily, as a co-op, we have customers that are willing to play along with us and be part of a test.  Some of our customers have both a UBNT and Cambium CPE on their roof and bounce back and forth for us.  I'm trying to do as complete of a real world test as possible before choosing one solution.  Cambium is moving to the front.  

Oh, and being able to do 5.2 is a huge plus...

3 Likes

I second this. UBNT Hardware is more capable due it is based on 802.11ac but firmware is beta. There are bugs which need a nightly reboot of the AP. With good signal and clean spectrum UBNT outperforms ePMP but with difficult conditions ePMP is much more reliable.

For PTP you better take the ptp product. This is 650 for cambium and airfiberX for UBNT. 


@Au Wireless wrote:

What I have discovered:

- Ubiquiti software is much better for RF surveys and general RF picture.  Especially the AirMax 7 and 8 versions.  Very well done. Big benefit to look at the RF picture at AP and CPE without taking any links down.

- Ubiquiti PtP with AirMax 7 and 8 is very fast and reliable.

- Ubiquiti PtMP is extremely inconsistent. There are too many firmware versions. Few are compatible. Some flat out don't work (AirMax 8 will not connect PtMP at all in our environment).

- As soon as the SNR creeps up, Ubiquiti dies off.  Trust me, I want them to work - simply because of my comfort level and the software.

- Every time I plug in a ePMP radio, it trains up and works.  In speed tests, it beats the Ubiquiti radio 75% of the time and is 100% more reliable.

I've put both UBNT and Cambium CPE's on a tripod and tested them side by side all over town for hours and hours. UBNT is really fast when it works, but it simply has not been reliable enough for us.  Despite my dislike of the Cambium firmware, we are starting to buy more Cambium gear and rolling out more test customers on Cambium CPEs.  And they are happier than the Ubiquiti customers.

Oh, and being able to do 5.2 is a huge plus...


I'll take an efficient, well tuned PtMP engine over a pretty facade UI any day of the week. Cambium's dedicated to getting the engine working right, and then making the UI look pretty... seems a lot of vendors pushing WiFi platforms are doing it backwards.

There are many vendors that are building custom MAC's on top of a WiFi chip-set that can do PtP with pretty good results... this is relatively easy... I don't know of any that have truly succeeded in building a stable, efficient PtMP MAC on top of a WiFi chip-set... let alone getting GPS sync and frequency reuse working properly. This is where Cambium stands alone. I've said this before and I'll say it again... Cambium has been able to 'squeeze' more efficiency out of an 802.11n chipset than any other vendor IMHO. Think about this, and than think about Cambium working on an 802.11ac product that is designed around this philosophy.

2 Likes

@Eric Ozrelic wrote:

@Au Wireless wrote:

What I have discovered:

- Ubiquiti software is much better for RF surveys and general RF picture.  Especially the AirMax 7 and 8 versions.  Very well done. Big benefit to look at the RF picture at AP and CPE without taking any links down.

- Ubiquiti PtP with AirMax 7 and 8 is very fast and reliable.

- Ubiquiti PtMP is extremely inconsistent. There are too many firmware versions. Few are compatible. Some flat out don't work (AirMax 8 will not connect PtMP at all in our environment).

- As soon as the SNR creeps up, Ubiquiti dies off.  Trust me, I want them to work - simply because of my comfort level and the software.

- Every time I plug in a ePMP radio, it trains up and works.  In speed tests, it beats the Ubiquiti radio 75% of the time and is 100% more reliable.

I've put both UBNT and Cambium CPE's on a tripod and tested them side by side all over town for hours and hours. UBNT is really fast when it works, but it simply has not been reliable enough for us.  Despite my dislike of the Cambium firmware, we are starting to buy more Cambium gear and rolling out more test customers on Cambium CPEs.  And they are happier than the Ubiquiti customers.

Oh, and being able to do 5.2 is a huge plus...


I'll take an efficient, well tuned PtMP engine over a pretty facade UI any day of the week. Cambium's dedicated to getting the engine working right, and then making the UI look pretty... seems a lot of vendors pushing WiFi platforms are doing it backwards.

There are many vendors that are building custom MAC's on top of a WiFi chip-set that can do PtP with pretty good results... this is relatively easy... I don't know of any that have truly succeeded in building a stable, efficient PtMP MAC on top of a WiFi chip-set... let alone getting GPS sync and frequency reuse working properly. This is where Cambium stands alone. I've said this before and I'll say it again... Cambium has been able to 'squeeze' more efficiency out of an 802.11n chipset than any other vendor IMHO. Think about this, and than think about Cambium working on an 802.11ac product that is designed around this philosophy.


Yes. Only problem there is no .ac product line announced from cambium. The 11n chipsets are limited in what can be done. So UBNT PTMP is a preview of what will be possible with next gen chipsets and at some time in the future they will get their Firmware working. Not sure they will make GPS working on their PTMP line as they suffered doing this with 11n based gear. But as the newer HW is more capable they might get it done this time.

This airview/airmagic stuff looks like fancy UI-toy but it is a very valuable tool cambium should consider for their nextgen gear.

Adding a second radio for monitoring will be a must have  in the future of unlicensed PTMP. It will help decreasing downtime with dfs channels as you know the condition of a channel to hop to before using it. Gathering this data from all connected CPEs the AP has a much better knowlegde for selecting a channel.


@ste wrote:

Yes. Only problem there is no .ac product line announced from cambium. The 11n chipsets are limited in what can be done.


Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse.

At the same time  current ePMP 1000 802.11n  point to multipoint  solution  outperforms  twice Ubnt 802.11ac with:

- max quantity  CPEs  per sectors

- max sector (  8+ active CPEs) throughput in 10/20/40 MHz channel bandwidth 

I assume your speaking about ePMP 2000. Sadly I didn't get to attend that talk at Wispa. Did cambium comment on ePMP 2000's backwards compatability details? Will you still see the extra performance on the sector with the new SM types (assuming there will be new SMs) and run the current N based SM?


@matkix wrote:

I assume your speaking about ePMP 2000. Sadly I didn't get to attend that talk at Wispa. Did cambium comment on ePMP 2000's backwards compatability details? Will you still see the extra performance on the sector with the new SM types (assuming there will be new SMs) and run the current N based SM?


ePMP 2000 radio is  802.11n, it is  based on   new  powerfull hardware (latest Atheros SOC),   has some new features, one of them   gives extra  7-10 dB Uplink  RSSI from ordinary    ePMP CPE , another one  provides   extra  interference rejection from adjacent channels . This gear  will be available  soon.

ePMP  802.11ac is in 2016 roadmap. All current ePMP SM will have interoperability with future ePMP 802.11ac , that means they will able to work in multipoint together with future SM 802.11ac  .

The real question is on the AC platform, can you take advantage of the AC upgrades in terms of speed even with old N clients on the sector.  Has cambium discussed this with anyone? If not, I'll bring it up in my meeting with them Monday and see if I can share back any of the data.


@Vyacheslav wrote:

Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse. 


Not sure where Vyacheslav is getting his information as none of this has been announced from Cambium and I haven't been told any of this through my sources. I think it's more a wish list. In regards to ePMP 2000, there were some very basic specifications announced. It's an advanced AP utilizing a 'next generation' 802.11n chipset, it features a filtering technology called HyPure (similar to 450i's dynamic interference filtering), it's completely backwards compatible with ePMP 1000, and it features an optional actively driven beam steering/beam forming antenna. To sum it up, ePMP 2000 is being designed to squeeze as much performance out of 802.11n as possible while providing superior interference mitigation.

1 Like

@Eric Ozrelic wrote:
In regards to ePMP 2000, there were some very basic specifications announced. It's an advanced AP utilizing a 'next generation' 802.11n chipset, it features a filtering technology called HyPure (similar to 450i's dynamic interference filtering), it's completely backwards compatible with ePMP 1000, and it features an optional actively driven beam steering/beam forming antenna. To sum it up, ePMP 2000 is being designed to squeeze as much performance out of 802.11n as possible while providing superior interference mitigation.

I also may add that ePMP 2000 has superheterodyne receiver that provides good spectral mask (   low Out of Band Emission - OOBE )  and  high  Adjacent Channel Interference  rejection -ACR . Together with GPS  sync these features provide  excellent ePMP 2000  multisectors RF collocation.

Hi

as per my understanding, for WISP, in PtMP, it seems of not much interest the .ac as 256QAM requires from 5-7 dB more in CINR over the 64QAM. Being this logaritmic, it requires much more signal over noise. Please note that extended bandwidth (40, 60 80 MHz channels) are of no interested as usually you can't find such spectrum hole on any business interesting area, nowadays...

Only on very limited, SM near AP and quiet scenarios it may be of benefit. In other cases, using a mix of customers with a mix of modulations, it may not improve more than 5-10 % the overall performance.

Better to improve the UL (with filters, beamforming, mimo) SNR, improve the PPS, improve the timing of the frame (gap between tx/rx, gap between sm, etc..), improve the QoS mechanisms, etc...

Best regards

Antonio


@Vyacheslav wrote:

@ste wrote:

Yes. Only problem there is no .ac product line announced from cambium. The 11n chipsets are limited in what can be done.


Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse.

At the same time  current ePMP 1000 802.11n  point to multipoint  solution  outperforms  twice Ubnt 802.11ac with:

- max quantity  CPEs  per sectors

- max sector (  8+ active CPEs) throughput in 10/20/40 MHz channel bandwidth 


How do you figure that current ePMP1000 outperformes UBNT AC?

I've got UBNT gear wirh 30 and 40 CPEs per sector pushing easily 25-30megabits to clients.   There are no limits on the number of associated CPEs on UBNT gear.


@RFWaveRider wrote:

@Vyacheslav wrote:

@ste wrote:

Yes. Only problem there is no .ac product line announced from cambium. The 11n chipsets are limited in what can be done.


Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse.

At the same time  current ePMP 1000 802.11n  point to multipoint  solution  outperforms  twice Ubnt 802.11ac with:

- max quantity  CPEs  per sectors

- max sector (  8+ active CPEs) throughput in 10/20/40 MHz channel bandwidth 


How do you figure that current ePMP1000 outperformes UBNT AC?

I've got UBNT gear wirh 30 and 40 CPEs per sector pushing easily 25-30megabits to clients.   There are no limits on the number of associated CPEs on UBNT gear.

UBNT has a limit, they are not upfront about it.   cambium has always been one to advertise its limits rather than let customers find out the hardway, like us.           M gear seemed to be 1 CPE per MHZ to have a sector run decent and rarely could we break that rule.   I could only imagine the RSSI and SNRs you've got to be trying to maintain in order to see those speeds.   what kind of performance do you see cumulatively ?   we've got epmp sectors that sit over 100 mpbs during peak.   
we've got customers getting over 60 mpbs durring peak hours from EPMP
Screenshot_2015-10-20-21-29-24.jpg
1 Like

@Chris_Bay wrote:

@RFWaveRider wrote:

@Vyacheslav wrote:

@ste wrote:

Yes. Only problem there is no .ac product line announced from cambium. The 11n chipsets are limited in what can be done.


Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse.

At the same time  current ePMP 1000 802.11n  point to multipoint  solution  outperforms  twice Ubnt 802.11ac with:

- max quantity  CPEs  per sectors

- max sector (  8+ active CPEs) throughput in 10/20/40 MHz channel bandwidth 


How do you figure that current ePMP1000 outperformes UBNT AC?

I've got UBNT gear wirh 30 and 40 CPEs per sector pushing easily 25-30megabits to clients.   There are no limits on the number of associated CPEs on UBNT gear.

UBNT has a limit, they are not upfront about it.   cambium has always been one to advertise its limits rather than let customers find out the hardway, like us.           M gear seemed to be 1 CPE per MHZ to have a sector run decent and rarely could we break that rule.   I could only imagine the RSSI and SNRs you've got to be trying to maintain in order to see those speeds.   what kind of performance do you see cumulatively ?   we've got epmp sectors that sit over 100 mpbs during peak.   
we've got customers getting over 60 mpbs durring peak hours from EPMP
Screenshot_2015-10-20-21-29-24.jpg

For sure ePMP1000 is much better compared to UBNT M Gear. It is at the top of 11n gear. The question is what does UBNT AC gear.

At this stage AC Firmware for UBNT is still beta. Even the latest stable release needs nightly reboots to run along. So you cant rely on this to buildout larger networks now. With the beta tree they integrate M gear into a AC network.

Doing tests with a low number of subscribers show the potential of AC gear. You can see >100MBit single speed tcp on a 20MHz Channel. So AC is better than N for sure as N is better than A. It needs some time to mature. This was the same with first 11n gear. UBNT has built interesting stuff into their gear. They add some filtering to improve uplink, they integrate additional hardware to speed up TDMA protocol and they added a second wireless card to do a background spectrum scan. So there is a lot of potential for the future. Repeat: Still beta.

Some claim AC needs some additional signal strength / quality to make sense. This is true but it is doable. You get additional speed for part of your CPEs now and you might use higher gain CPEs in the future. I see stable 256 QAM on a 19db CPE connected to an omni 2km apart. So no magic there. Using sectors and small dishes you get a reasoable range for higher modulations.

1 Like

@ste wrote:

@Chris_Bay wrote:

@RFWaveRider wrote:

@Vyacheslav wrote:

@ste wrote:

Yes. Only problem there is no .ac product line announced from cambium. The 11n chipsets are limited in what can be done.


Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse.

At the same time  current ePMP 1000 802.11n  point to multipoint  solution  outperforms  twice Ubnt 802.11ac with:

- max quantity  CPEs  per sectors

- max sector (  8+ active CPEs) throughput in 10/20/40 MHz channel bandwidth 


How do you figure that current ePMP1000 outperformes UBNT AC?

I've got UBNT gear wirh 30 and 40 CPEs per sector pushing easily 25-30megabits to clients.   There are no limits on the number of associated CPEs on UBNT gear.

UBNT has a limit, they are not upfront about it.   cambium has always been one to advertise its limits rather than let customers find out the hardway, like us.           M gear seemed to be 1 CPE per MHZ to have a sector run decent and rarely could we break that rule.   I could only imagine the RSSI and SNRs you've got to be trying to maintain in order to see those speeds.   what kind of performance do you see cumulatively ?   we've got epmp sectors that sit over 100 mpbs during peak.   
we've got customers getting over 60 mpbs durring peak hours from EPMP

It needs some time to mature. This was the same with first 11n gear. UBNT has built interesting stuff into their gear. They add some filtering to improve uplink, they integrate additional hardware to speed up TDMA protocol and they added a second wireless card to do a background spectrum scan. So there is a lot of potential for the future. Repeat: Still beta.


 All new AC features - filters, background scan, hardware ( TDMA ???) accelerator  are not be considered  as  key competitive advantages.   Total Airmax redesign   is   required  for ubnt AC solution. It  must be proper  true TDMA implementation.   But  main  problem of ubnt AC solution is wrong way of product development. Current AC Wave 1 chip is EOL. Development  of AC wave 1  based  gear and provide  interoperability  with 802.11n  is useless,  it does not make sense, it is  dead-end.   


@Vyacheslav wrote:

@ste wrote:

@Chris_Bay wrote:

@RFWaveRider wrote:

@Vyacheslav wrote:

@ste wrote:

Yes. Only problem there is no .ac product line announced from cambium. The 11n chipsets are limited in what can be done.


Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse.

At the same time  current ePMP 1000 802.11n  point to multipoint  solution  outperforms  twice Ubnt 802.11ac with:

- max quantity  CPEs  per sectors

- max sector (  8+ active CPEs) throughput in 10/20/40 MHz channel bandwidth 


How do you figure that current ePMP1000 outperformes UBNT AC?

I've got UBNT gear wirh 30 and 40 CPEs per sector pushing easily 25-30megabits to clients.   There are no limits on the number of associated CPEs on UBNT gear.

UBNT has a limit, they are not upfront about it.   cambium has always been one to advertise its limits rather than let customers find out the hardway, like us.           M gear seemed to be 1 CPE per MHZ to have a sector run decent and rarely could we break that rule.   I could only imagine the RSSI and SNRs you've got to be trying to maintain in order to see those speeds.   what kind of performance do you see cumulatively ?   we've got epmp sectors that sit over 100 mpbs during peak.   
we've got customers getting over 60 mpbs durring peak hours from EPMP

It needs some time to mature. This was the same with first 11n gear. UBNT has built interesting stuff into their gear. They add some filtering to improve uplink, they integrate additional hardware to speed up TDMA protocol and they added a second wireless card to do a background spectrum scan. So there is a lot of potential for the future. Repeat: Still beta.


 All new AC features - filters, background scan, hardware ( TDMA ???) accelerator  are not be considered  as  key competitive advantages.   Total Airmax redesign   is   required  for ubnt AC solution. It  must be proper  true TDMA implementation.   But  main  problem of ubnt AC solution is wrong way of product development. Current AC Wave 1 chip is EOL. Development  of AC wave 1  based  gear and provide  interoperability  with 802.11n  is useless,  it does not make sense, it is  dead-end.   


You're talking without any techical background. "proper true TDMA implementation", "dead-end".

This is not UBNT forum where fanboys just cheer and everything other vendors do is crap. Background scan is a great feature I would like to see with epmp. And as you see with the slow gui there is some need for more processing power. Considering Wave 1 as dead you can consider every computer as dead as it is worse then next years computer for sure.

Just stay with the facts. UBNT AC is beta but has promising features. If they leave beta before ePMP AC arrives is guesswork. If UBNT ever make GPS work with their PTMP is guesswork. If ePMP AC will be non beta from the beginning is guesswork, too. Just let's hope.


@ste wrote:

@Vyacheslav wrote:

@ste wrote:

@Chris_Bay wrote:

@RFWaveRider wrote:

@Vyacheslav wrote:

@ste wrote:

Yes. Only problem there is no .ac product line announced from cambium. The 11n chipsets are limited in what can be done.


Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse.

At the same time  current ePMP 1000 802.11n  point to multipoint  solution  outperforms  twice Ubnt 802.11ac with:

- max quantity  CPEs  per sectors

- max sector (  8+ active CPEs) throughput in 10/20/40 MHz channel bandwidth 


How do you figure that current ePMP1000 outperformes UBNT AC?

I've got UBNT gear wirh 30 and 40 CPEs per sector pushing easily 25-30megabits to clients.   There are no limits on the number of associated CPEs on UBNT gear.

UBNT has a limit, they are not upfront about it.   cambium has always been one to advertise its limits rather than let customers find out the hardway, like us.           M gear seemed to be 1 CPE per MHZ to have a sector run decent and rarely could we break that rule.   I could only imagine the RSSI and SNRs you've got to be trying to maintain in order to see those speeds.   what kind of performance do you see cumulatively ?   we've got epmp sectors that sit over 100 mpbs during peak.   
we've got customers getting over 60 mpbs durring peak hours from EPMP

It needs some time to mature. This was the same with first 11n gear. UBNT has built interesting stuff into their gear. They add some filtering to improve uplink, they integrate additional hardware to speed up TDMA protocol and they added a second wireless card to do a background spectrum scan. So there is a lot of potential for the future. Repeat: Still beta.


 All new AC features - filters, background scan, hardware ( TDMA ???) accelerator  are not be considered  as  key competitive advantages.   Total Airmax redesign   is   required  for ubnt AC solution. It  must be proper  true TDMA implementation.   But  main  problem of ubnt AC solution is wrong way of product development. Current AC Wave 1 chip is EOL. Development  of AC wave 1  based  gear and provide  interoperability  with 802.11n  is useless,  it does not make sense, it is  dead-end.   


You're talking without any techical background. "proper true TDMA implementation", 


Ubnt Airmax  is asynchronous polling based   protocol of  wireless multiple access. It is the same type of access  as Nstreme/Nv2 Miktrotik, the same as Outdoor Router Protocol of  Proxim and many  others. It was  invented in 1996 by  Doug Karl  from  Karlnet , it was  called as Turbocell protocol, and  was used  instead  of CSMA/CA   in outdoor applications  of  802.11 ( those time non standard WLAN) systems.    Due to  it's asynchronous type GPS sync and Frequency Reuse  feature will never be made for Airmax products. Due to  asynchronous type  Airmax  ubnt  802.11n has in P2MP ( 8+ active CPEs) low sector capacity - total max throughput ( 25-30 Mbps)   and  low max served CPEs per sector ( less 30 ).   Airmax UBNT AC  is the same  as Airmax N. Why Airmax AC will work better then Airmax N? Due to hardware accelerator ? Which scheduler function does it accelerate? It need not be accelerated, it should be completely redesigned  and should become  synchronous protocol, as true TDMA  in ePMP or  Radwin ( both based on 802.11 chipset) ,  or   Readline  or  Cambium PMP450 ( both based on   proprietary chipset). Only  this protocol ( TDMA) of multiple access  is able to be synced and only TDMA  round-robin multiple access   scheduler is able to get   close  throughput in Point to Point  and P2MP   with 100+ CPEs scenarios. 

Test your ubnt sector AC with 30 CPEs when  8+ are active and  these CPEs have  ( before test)    >  5-10 Mbps  total  upload traffic . You will see real  ubnt 802.11AC throughput in P2MP, that even at 256QAM 5/6 is not much higher  then Airmax 802.11n.

What are   promising  functions in UBNT AC?  256QAM,  filters, accelerator, background  scan? All of them  are not able to provide  in 20 MHz  channel key competitive  advantages:  80-90 Mbps sector throughput with 80-120 CPEs on sector with Sync and FR , that  ePMP 802.11n already  has.  

1 Like

Not to mention when dealing with Cambium as a company you get actual support. Someone who answers the phone and knows the product day or night. RMAs that don't include games, anyone here tried to RMA an AirFiber 24 yet? They seem to cause you endless delays, demand endless testing before they even consider running an RMA on the product.  On the other hand, I had to RMA a  ePMP radio because I lost power when upgrading the firmware, clearly my fault. Cambium issued the RMA with a 10 minute phone call, and even followed up with me three times to make sure I would send it.

We have been running ePMP gear alongside Ubnt M gear, and we see much better sector density, much better aggregate bandwidth, and many more happy users. Our number one complaint from customers running the UBNT M gear is our internet gets really slow around 5PM. We load all our sectors the same, run the same 20 Mhz channels, similar gain antennas. ePMP gear has given us a new edge in reliability and aggregate sector capacity  that UBNT gear has never offered.

All that being said, I’m no fanboy here. If a company offers a superior product I’m always interested. We continually test products for our network deployments. But one thing that gives me confidence in the Cambium product line, their product teams ALWAYS think of the current customers, and current deployments when building new products. Not leaving me with a choice of replacing all my SMs to take advantage of a new upgrade. On the downside I think Cambium needs to look into making backhauls for 24Ghz, many companies already have…

2 Likes

@Vyacheslav wrote:

@ste wrote:

@Vyacheslav wrote:

@ste wrote:

@Chris_Bay wrote:

@RFWaveRider wrote:

@Vyacheslav wrote:

@ste wrote:

Yes. Only problem there is no .ac product line announced from cambium. The 11n chipsets are limited in what can be done.


Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse.

At the same time  current ePMP 1000 802.11n  point to multipoint  solution  outperforms  twice Ubnt 802.11ac with:

- max quantity  CPEs  per sectors

- max sector (  8+ active CPEs) throughput in 10/20/40 MHz channel bandwidth 


How do you figure that current ePMP1000 outperformes UBNT AC?

I've got UBNT gear wirh 30 and 40 CPEs per sector pushing easily 25-30megabits to clients.   There are no limits on the number of associated CPEs on UBNT gear.

UBNT has a limit, they are not upfront about it.   cambium has always been one to advertise its limits rather than let customers find out the hardway, like us.           M gear seemed to be 1 CPE per MHZ to have a sector run decent and rarely could we break that rule.   I could only imagine the RSSI and SNRs you've got to be trying to maintain in order to see those speeds.   what kind of performance do you see cumulatively ?   we've got epmp sectors that sit over 100 mpbs during peak.   
we've got customers getting over 60 mpbs durring peak hours from EPMP

It needs some time to mature. This was the same with first 11n gear. UBNT has built interesting stuff into their gear. They add some filtering to improve uplink, they integrate additional hardware to speed up TDMA protocol and they added a second wireless card to do a background spectrum scan. So there is a lot of potential for the future. Repeat: Still beta.


 All new AC features - filters, background scan, hardware ( TDMA ???) accelerator  are not be considered  as  key competitive advantages.   Total Airmax redesign   is   required  for ubnt AC solution. It  must be proper  true TDMA implementation.   But  main  problem of ubnt AC solution is wrong way of product development. Current AC Wave 1 chip is EOL. Development  of AC wave 1  based  gear and provide  interoperability  with 802.11n  is useless,  it does not make sense, it is  dead-end.   


You're talking without any techical background. "proper true TDMA implementation", 


Ubnt Airmax  is asynchronous polling based   protocol of  wireless multiple access. It is the same type of access  as Nstreme/Nv2 Miktrotik, the same as Outdoor Router Protocol of  Proxim and many  others. It was  invented in 1996 by  Doug Karl  from  Karlnet , it was  called as Turbocell protocol, and  was used  instead  of CSMA/CA   in outdoor applications  of  802.11 ( those time non standard WLAN) systems.    Due to  it's asynchronous type GPS sync and Frequency Reuse  feature will never be made for Airmax products. Due to  asynchronous type  Airmax  ubnt  802.11n has in P2MP ( 8+ active CPEs) low sector capacity - total max throughput ( 25-30 Mbps)   and  low max served CPEs per sector ( less 30 ).   Airmax UBNT AC  is the same  as Airmax N. Why Airmax AC will work better then Airmax N? Due to hardware accelerator ? Which scheduler function does it accelerate? It need not be accelerated, it should be completely redesigned  and should become  synchronous protocol, as true TDMA  in ePMP or  Radwin ( both based on 802.11 chipset) ,  or   Readline  or  Cambium PMP450 ( both based on   proprietary chipset). Only  this protocol ( TDMA) of multiple access  is able to be synced and only TDMA  round-robin multiple access   scheduler is able to get   close  throughput in Point to Point  and P2MP   with 100+ CPEs scenarios. 

Test your ubnt sector AC with 30 CPEs when  8+ are active and  these CPEs have  ( before test)    >  5-10 Mbps  total  upload traffic . You will see real  ubnt 802.11AC throughput in P2MP, that even at 256QAM 5/6 is not much higher  then Airmax 802.11n.

What are   promising  functions in UBNT AC?  256QAM,  filters, accelerator, background  scan? All of them  are not able to provide  in 20 MHz  channel key competitive  advantages:  80-90 Mbps sector throughput with 80-120 CPEs on sector with Sync and FR , that  ePMP 802.11n already  has.  


To make it clear I am no UBNT Fanboy and I do not use it in large scale. And yes. UBNT AC is not ready to use for large scale. My point is that there are some functions I want to see in ePMP. This is 256 QAM, background channel scanning and improved CPU/HW speed.

BTW: I dont know what they changed with Airmax AC. For sure it is changed as you've to update old N Gear to be compatible. I guess you dont know the internals, too. So it is guesswork if it will be syncable and if it will scale once finished. 

2 Likes

@Eric Ozrelic wrote:

@Vyacheslav wrote:

Cambium  will   make  PTP and PTMP ePMP,  based on 802.11ac chipset. It will have:

- 802.11 ac wave 2 with MU-MIMO 

- full  interoperability with ePMP 1000 ( 802.11n) in point to multipoint 

- TDMA protocol with GPS Sync and Frequency Reuse. 


Not sure where Vyacheslav is getting his information as none of this has been announced from Cambium and I haven't been told any of this through my sources. I think it's more a wish list. 


Add the similar informations about AC gears on a ePMP trainning, that Cambium was waiting for phase 2 of AC chipset but don't remember about MU-MIMO. The also said about interoperability with ePMP 1000.

2 Likes