Steve Perlman claims to have a new approach to revolutionize wireless networks w pCell/Artemis

[MrZorbatron]

You can dramatically increase wifi capacity by having multiple (overlapping) access points on different channels, each with dynamic transmit power control. Basically, a heavily loaded cell contracts while unloaded cells expand. I really think this technology is nothing more than this, except using cellular phone technologies instead of wifi.

[dnwk]

I think they just build a showcase in SF and hoping tier 1 Carrier will buy them out

[S4GRU]

You can dramatically increase wifi capacity by having multiple (overlapping) access points on different channels, each with dynamic transmit power control. Basically, a heavily loaded cell contracts while unloaded cells expand. I really think this technology is nothing more than this, except using cellular phone technologies instead of wifi.

Which leads to interference and performance degradation with current network technology. So if there is a way to eliminate the interference performance degradation, that would be a leap forward in progress in itself. However, I'm still not sure I get this whole program from even a big picture level, let alone in the details.

 

I won't say I'm a skeptic. I'm just saying that as I read any information, it either is confusing, contradicts what I know to be true, or leads to more questions. So I just continue to sit on the sidelines and watch it unfold.

[lou99/maximus1987]

Is there any indication this only works in Tdd?

 

 

Sent from my iPhone using Tapatalk

[lou99/maximus1987]

He says 5mhz - probably b39 Tdd - can do 12.5mbps

 

 

 

 

Sent from my iPhone using Tapatalk

[lou99/maximus1987]

I wonder:

For N people, how many pcell modules M do you need to continuously N pcells?

Even if M pcell modules can only synthesize D<N pcells continually, if the ratio of N/D - duty cycle effectively - is high enough, the data speeds should still massively improve

 

 

Sent from my iPhone using Tapatalk

[MrZorbatron]

Which leads to interference and performance degradation with current network technology. So if there is a way to eliminate the interference performance degradation, that would be a leap forward in progress in itself. However, I'm still not sure I get this whole program from even a big picture level, let alone in the details.

 

I won't say I'm a skeptic. I'm just saying that as I read any information, it either is confusing, contradicts what I know to be true, or leads to more questions. So I just continue to sit on the sidelines and watch it unfold.

I do understand what he is claiming to com and the Methods/technologies he is claiming to use to accomplish it.

 

as for my described wifi setup, there is little to no interference due to lack of frequency reuse by adjacent cells.  Overlap is carefully managed in two to three layers, so cell sizes ancan be actively managed.

[WiWavelength]

I wonder:

For N people, how many pcell modules M do you need to continuously N pcells?

Even if M pcell modules can only synthesize D<N pcells continually, if the ratio of N/D - duty cycle effectively - is high enough, the data speeds should still massively improve

 

If Train A leaves Bugtussel at 1pm traveling 50 mph and Train B leaves Hooterville at 2pm traveling 70 mph...

 

AJ

[MrZorbatron]

If Train A leaves Bugtussel at 1pm traveling 50 mph and Train B leaves Hooterville at 2pm traveling 70 mph...

 

AJ

Exactly

[MrZorbatron]

I wonder:

For N people, how many pcell modules M do you need to continuously N pcells?

Even if M pcell modules can only synthesize D<N pcells continually, if the ratio of N/D - duty cycle effectively - is high enough, the data speeds should still massively improve

 

 

Sent from my iPhone using Tapatalk

Theoretically, you just need the overlap of two or three.

[lou99/maximus1987]

Theoretically, you just need the overlap of two or three.

 

I don't think you could create thousands of pcells with just 3 pmodules.

[lou99/maximus1987]

Someone else postulated that this solution probably can't work with fdd.

Their reason is that you need feedback for both the downlink rf and uplink rf channel but in fdd, downlink rf channel can never be used to transmit so there's no way to estimate anything about the downlink rf channel.

[bigsnake49]

Someone else postulated that this solution probably can't work with fdd.

Their reason is that you need feedback for both the downlink rf and uplink rf channel but in fdd, downlink rf channel can never be used to transmit so there's no way to estimate anything about the downlink rf channel.

it will work fine with both fdd and tdd. There is info about the state of both the uplink and downlink being exchanged between the BTS and the handset.

[MrZorbatron]

I don't think you could create thousands of pcells with just 3 pmodules.

On paper, it would, since it basically leverages crosstalk.  More would still be better, up to a point, though.

 

The biggest problem is still that any remotely high noise floor would totally annihilate its performance, since it can't account for any signals that it itself is not producing.  Also, this technology would not lend itself to close to the same level of improvement on the reverse link.

 

Again, this would be functional in a lab, but in the real world, it won't work.  Certain people here seem to be so taken with this idea what they look straight past its obvious flaws and shortcomings, to just wholeheartedly believe in it.  Perlman's proven to be full of crap more than once before, and he has successes as well.  On this one, I'm leaning so far toward crap that I can barely stand up anymore.

 

The whole point of this system is to use the RF energy that is already out there, produced by the various radios, in the location of the subscriber.  Through technologies like phased antennas, beam forming, and a high degree of electronic processing, it purports to use all of its P-Cell antennas in an area in combination, in order to precisely control exactly what signal the subscriber unit sees.

 

If the system really works as it should, the bottleneck will be the data center processing the information, since all the P-Cell radios will need will be a link sufficient for one cell site's full load on whatever bandwidth they have, at maximum spectral efficiency.  If you can feed the data center fast enough, every user will have basically full speed.  The more P-Cell units are installed, the smaller and more finely grained the individual pseudo cell environments can be, and the higher and higher the processing load at the computing center will be (exponential basically).

 

By using a combination of constructive and destructive interfering signals to turn signal intended for one subscriber unit, which would normally be simple interference to all others, into signal useful to others.  This level of fine-grained, area-by-area, tight control over the RF spectrum basically requires a high resolution, real time understanding of exactly the signal present in every location at the same time.  This level of understanding is not possible, as some interfering signals will not be produced by the P-Cell units, and therefore not known to them.  The fact that this is supposed to work with existing LTE phones tells me that there is no monitoring and reporting done by the client hardware.

 

This is why the system is bullshit.  The level of processing is a fantasy in today's world.  There is no accounting for stray interference.  What we will see is a multiple small cell solution, with carefully managed, mapped, and balanced sizes.  The sizes may be irregular through beam forming etc.

 

Do I think the system will work?  Yes.  Do I think it will work as well as Perlman says it will, by the mechanisms he says it will, not at all.

[bcltoys]

On paper, it would, since it basically leverages crosstalk.  More would still be better, up to a point, though.

 

The biggest problem is still that any remotely high noise floor would totally annihilate its performance, since it can't account for any signals that it itself is not producing.  Also, this technology would not lend itself to close to the same level of improvement on the reverse link.

 

Again, this would be functional in a lab, but in the real world, it won't work.  Certain people here seem to be so taken with this idea what they look straight past its obvious flaws and shortcomings, to just wholeheartedly believe in it.  Perlman's proven to be full of crap more than once before, and he has successes as well.  On this one, I'm leaning so far toward crap that I can barely stand up anymore.

 

The whole point of this system is to use the RF energy that is already out there, produced by the various radios, in the location of the subscriber.  Through technologies like phased antennas, beam forming, and a high degree of electronic processing, it purports to use all of its P-Cell antennas in an area in combination, in order to precisely control exactly what signal the subscriber unit sees.

 

If the system really works as it should, the bottleneck will be the data center processing the information, since all the P-Cell radios will need will be a link sufficient for one cell site's full load on whatever bandwidth they have, at maximum spectral efficiency.  If you can feed the data center fast enough, every user will have basically full speed.  The more P-Cell units are installed, the smaller and more finely grained the individual pseudo cell environments can be, and the higher and higher the processing load at the computing center will be (exponential basically).

 

By using a combination of constructive and destructive interfering signals to turn signal intended for one subscriber unit, which would normally be simple interference to all others, into signal useful to others.  This level of fine-grained, area-by-area, tight control over the RF spectrum basically requires a high resolution, real time understanding of exactly the signal present in every location at the same time.  This level of understanding is not possible, as some interfering signals will not be produced by the P-Cell units, and therefore not known to them.  The fact that this is supposed to work with existing LTE phones tells me that there is no monitoring and reporting done by the client hardware.

 

This is why the system is bullshit.  The level of processing is a fantasy in today's world.  There is no accounting for stray interference.  What we will see is a multiple small cell solution, with carefully managed, mapped, and balanced sizes.  The sizes may be irregular through beam forming etc.

 

Do I think the system will work?  Yes.  Do I think it will work as well as Perlman says it will, by the mechanisms he says it will, not at all.

So you think he has pulled the wool over the eyes of  a multi-billon dollar company. Dish. Does sound to good to be true 

[WiWavelength]

Someone else postulated that this solution probably can't work with fdd.

Their reason is that you need feedback for both the downlink rf and uplink rf channel but in fdd, downlink rf channel can never be used to transmit so there's no way to estimate anything about the downlink rf channel.

 

For link estimation and power control, TDD is generally superior.

 

It can use an open loop approach, since both uplink and downlink are the same frequency and experience the same path loss -- this, of course, assumes zero or low mobility.  But if we use 5 ms as a frame of reference, even a vehicle traveling 60 mph moves only about 5 inches during that time.

 

In other circumstances, FDD and closed loop power control can be more effective.

 

AJ

[MrZorbatron]

So you think he has pulled the wool over the eyes of  a multi-billon dollar company. Dish. Does sound to good to be true 

Yes, I really do.

[Trip]

I thought he was leasing the spectrum from Dish, which is to say, Dish is getting paid for this deployment and thus doesn't care either way what is on it or whether or not it works.  Or did I miss something and Dish invested in it?

 

- Trip

[bcltoys]

I thought he was leasing the spectrum from Dish, which is to say, Dish is getting paid for this deployment and thus doesn't care either way what is on it or whether or not it works.  Or did I miss something and Dish invested in it?

 

- Trip

I think you are right about that.

[Nickel]

 

[Fraydog]

I suspect that if interference could be used to create personal cells, that Ericsson/DoCoMo/Qualcomm/Nokia/Huawei/3GPP and the other heavy hitters would be over that for 5G. The 5G standardization process should be a collaborative process and if Perlman wants to be part of that, then the 3GPP should welcome him with open arms. 

 

What my suspicion is, is that this is a beamforming solution that aggressively uses Cloud RAN... a technology Nokia and other providers are already pushing forward. Now the aspect of using interference to aim beamforms is one I have to see in the real world. The issue I see is getting the processing power to be able to track and accurately aim beamforms to be able to more consistently deliver bandwidth. I hope Perlman has a server farm that would be able to handle that. Otherwise the experience is going to fall apart. 

 

Honestly I'd be less skeptical if Perlman didn't make this sound like it was 5G happening right damn now but I think he should have taken a more measured marketing approach. 

[lou99/maximus1987]

I suspect that if interference could be used to create personal cells, that Ericsson/DoCoMo/Qualcomm/Nokia/Huawei/3GPP and the other heavy hitters would be over that for 5G. The 5G standardization process should be a collaborative process and if Perlman wants to be part of that, then the 3GPP should welcome him with open arms.

 

What my suspicion is, is that this is a beamforming solution that aggressively uses Cloud RAN... a technology Nokia and other providers are already pushing forward. Now the aspect of using interference to aim beamforms is one I have to see in the real world. The issue I see is getting the processing power to be able to track and accurately aim beamforms to be able to more consistently deliver bandwidth. I hope Perlman has a server farm that would be able to handle that. Otherwise the experience is going to fall apart.

 

Honestly I'd be less skeptical if Perlman didn't make this sound like it was 5G happening right damn now but I think he should have taken a more measured marketing approach.

He's offering real hardware now to carriers to test. Shouldn't take long to see if it works. Just start a continuous speedtest on bunch of phones in same sector.

 

 

Sent from my iPhone using Tapatalk

[Fraydog]

He's offering real hardware now to carriers to test. Shouldn't take long to see if it works. Just start a continuous speedtest on bunch of phones in same sector.

 

 

Sent from my iPhone using Tapatalk

But Nokia is already doing a lot of the same shit. Focus your eyes on this:

 

 

 

 

 

Sent from my iPhone using Tapatalk

[MrZorbatron]

 

 

This model doesn't work.

[Nickel]

This model doesn't work.

I understood what you getting at, I was just making a dumb joke about it being over my head.