Snake oil or broadband panacea? DIDO.

useruser · 12-03-2014 12:13pm #1

Well this is interesting - DIDO from a start-up wireless company (albeit 10 years in the making) claims to have developed a new radio technology that could lead to 1000x (yes, 1000) increase in radio capacity. I'm no radio expert but the whitepaper here: http://www.rearden.com/DIDO/DIDO_White_Paper_110727.pdf seems legit. More here: http://www.artemis.com/pcell

If this is real then there are profound implications for broadband delivery.

watty · 12-03-2014 6:45pm

Fantasy.
I can't believe the column inches this is getting.

It's the "radio" equivalent of people claiming Perpetual motion Machines and / or Free Energy.

Almost all the "so-called" tech journalism on this is uncritically recycling the press releases.

useruser · 12-03-2014 7:02pm

Here's a pretty detailed analysis of Artemis, I certainly don't know enough about radio to critique it but as a casual reader it seems thorough and is certainly an endorsement of Artemis' claims: http://akbars.net/how-steve-perlmans-revolutionary-wireless-technology-works-and-why-its-a-bigger-deal-than-anyone-realizes.html Perlman himself has a very long track record and seems unlikely to have spent 10 years developing something that ultimately cannot deliver.

What say you?

watty wrote: »

Fantasy.

More recent presentation by Perlman here: http://www.youtube.com/watch?v=5bO0tjAdOIw He would be risking a whole lot of egg on his face if those live demos are faked.

Here's a much more skeptical view from Dave Burstein (DSL Prime): http://fastnetnews.com/dslprime/42-d/5044-ny-times-latest-breakthrough-48-megabits-over-100-megabit-network

watty · 13-03-2014 10:19am

It's still nonsense though.

useruser · 13-03-2014 10:24am

watty wrote: »

It's still nonsense though.

Why?

watty · 13-03-2014 8:31pm

It's mostly really just MIMO.
And "useful" traffic gain with MIMO is poor in the real world:

Maybe an average 1.3x to 1.6x increase in traffic / capacity compared to x10 to x32 using FIXED wireless and roof top aerials
Short Range: No effect for most of cell area
Many expensive aerials: Space & size issue as well as cost
Many expensive "radio sets" and processing
Drastically Limited by size of a USB dongle or Phone.

MIMO is around for many years. It's expensive and limited in the improvement. It's used by Vendors as magic pixie dust to make the systems sound good on paper and make Mobile operators spend more.

There can be x2 to x3 average improvement though by making all the phone masts, channels and bands be part of a single Wholesale network which the retail phone / mobile companies then sell access to. The idea of splitting the bands to four separate operators on the simplistic and false idea that this creates competition and thus lower consumer prices and thus benefits consumer is absolute rubbish. In reality this is done to maximise licence revenue on auction. The "Wholesale network" should get the spectrum free but have severe conditions on quality and coverage. The Government then get their money via VAT.

So this is nonsense. You can do pretty impressive stuff in a controlled environment. Real life is limited by the laws of Thermodynamics which Shannon - Hartley and Shannon-Nyquist is derived from. There is no free lunch. We are already at the "Shannon Limit".

Summed up:
There are no free lunches.
If your lunch is free someone else is conned into paying for it.
If you even think about a free lunch you'll probably pay more.

Frequent claims surface about "super efficient" schemes.

No free lunch
http://en.wikipedia.org/wiki/A_Mathematical_Theory_of_Communication
http://en.wikipedia.org/wiki/Information_entropy

I've seen a lot of fake snake oil out there.

useruser · 13-03-2014 11:05pm

watty, I think you are making 3 points:

1. They are just using MIMO, nothing new here.
2. They are claiming to exceed the Shannon limit, which would obviously just be total bunkum.
3. This is just a demo in a controlled environment and this would never work in a real-world deployment.

On point 1, this seems to be "Cooperative MIMO" or "Network MIMO" rather than the more straightforward antenna arrays we are used to thinking of as MIMO. Perhaps this is just a (demo-able) implementation of well-known wireless technology ideas? In particular they insist that there are no antenna arrays at the base station, each one uses only a single antenna (as does the receiver). That they have got this working at all, let alone demonstrating (supposedly) more throughput than is currently possible in a normal LTE channel is interesting.
Perlman talks a bit about MIMO here although not in much detail: http://venturebeat.com/2014/02/20/steve-perlman-pcell-is-real-and-it-will-change-the-world-interview

2. Perlman states that this is not the case, they do not claim to be breaking any physical laws (Shannon limit in particular).

3. This would seem to be the real problem - coordinating transmissions for a handful of devices that aren't moving is one thing but handling 100s or 1000s of devices in the real world is quite another. In particular there are hints that a system like this would require enormous numbers of basestations, even if these are as small and simple as they claim that could still render the whole scheme unworkable. Add to that the (presumably) complex computing required and the (also presumably) low latency requirements for getting those computed signals to each basestation.

Some good (skeptical) discussion in the comments to this IEEE article on the subject: http://spectrum.ieee.org/tech-talk/telecom/wireless/5g-service-on-your-4g-phone

Perlman claims that a forthcoming whitepaper will explain everything, let's see...

watty · 14-03-2014 1:20pm

We are already close to Shannon Limit on 2G EDGE, 3G HSPA and 4G. They have ever higher speed for one user by using
200kHz or 2 x 200KHz (EDGE)
5MHz (3G vs 1.5MHz approx of older CDMA1)
Up to 20MHz channel (LTE, but most Irish LTE is 5MHz channels).
With more users 3G is inefficient due to CDMA, so with 10 Users (rubbish speed on any Mobile system), LTE or EDGE has twice performance of 3G. EDGE is typically 1/20th peak speed of 3G/HSPA simply because it's about typically 1/20th of spectrum! If there are enough channels (say 20) then for 10 Users EDGE (on 2G) bizarrely can give same speed as 4G with 10 simultaneous users on 5MHz and twice speed of 3G (about 1Mbps).

Without more aerials (Space Diversity, MIMO etc) at both ends it's not possible to get more speed in the same spectrum with same power. That is what Hartley's Law (based ultimately on Thermodynamics) is about. Shannon then derived a way to say what the actual error rate can be for a given channel's characteristics.

This is total bunk.

The ONLY way for more speed is spectrum. You can't use more power except on Satellite and Point to Point Links. Any Cellular (which means re-use of same frequencies so they don't overlap) system the power is set by cell size. More power isn't an option on Mobile. It is on Roof top aerial fixed systems as they can be highly directional. This is why in the SAME band and channel size that Fixed Wireless can be over x16 more capacity. Mobile can't even use Polarisation diversity. Satellite and Point to Point Terrestrial, Point to Many Fixed Wireless can use two polarisations (L & R or H & V depending on band) to double capacity per spectrum.

Unless the laws of physics change any capacity / speed improvement is only by using more channels / more spectrum.

I put some nice graphs here:
http://www.techtir.ie/comms/signal-loss-limits

So it's nonsense.

This is quite old and prepared only just after HSPA+ rolled out in Ireland.
http://www.radioway.info/comparewireless/

But it's not going to change. "5G" or "6G" doesn't really matter. It correctly forecast LTE, Wimax 4G performance. Actually I looked at dedicated Fixed Wireless DOCSIS and Flarion Flash-OFDMA, which since they are both native IP can be regarded as 4G. Really 4G has come to mean simply Native TCP/IP & UDP IP networks. 2G EDGE and 3G HSPA+ carry IP, but are not native data networks. The underlying frame structure etc is for efficient carriage of voice calls. 4G has no native voice so is ironically much less efficient (more costly!) for voice and SMS calls.

useruser · 14-03-2014 1:47pm

And yet, they claim to be demonstrating more IP throughput in 5MHz of bandwidth than should actually be possible. Of course there remains the possibility that this has just been faked for the cameras (which would be trivial to do) but you would have to wonder why they would go to that trouble? Either this is an elaborate hoax for unknown publicity reasons ala Steorn or they really have something. If what they have demo'd can be verified then the question still remains whether this is actually deployable.

Their claim is that they can replicate multiple individual channels all using the same spectrum (so-called "personal cell") by manipulating transmissions from many simple low-power base-stations in such a way that each receiver receives an optimal signal through constructive interference. If (and that's a very big if) this is true then of course your points about shared spectrum and power etc no longer apply and that's why this is so interesting.

watty · 14-03-2014 2:18pm

useruser wrote: »

points about shared spectrum and power etc no longer apply and that's why this is so interesting.

That's why it's nonsense,"my points" aren't opinion but based on immutable mathematics and physics.

There are many reasons and motives for such product claims. The "headline" Marketing of every Mobile Infrastructure and "Irish" Mobile operator is nonsense. Three for instance have no Broadband products. The "21 Mbps" and especially "42 Mbps" claimed by 3G HSPA+ is totally misleading. Can really only be achieved on a Femto cell in a shielded premises with Line of sight and only one user! But my really ancient WiFi in an 12 year old Laptop (WiFi upgraded about 7 years ago) is 108 Mbps in same circumstances. But 4G /LTE, 3G HSPA, EDGE and 250 Mbps 802.11n WiFi can easily be 0.25Mbps per connection.

For a Mobile Network with 15MHz + 15MHz spectrum and a profitable number of users the average speeds will be about 1Mbps for 3G and 2Mbps for 4G at PEAK times. Off peak daytime about 3Mbps average for both. If you had all 3 channels "bonded" (technically available) and only one user and perfect signal then you can have 60 Mbps on it. I can do that in a big field or a hillside, never mind a lab.

People buying networks, whatever misleading marketing they promote later use "KPI" Key performance indicators. They don't care about the theoretical peak performance as it's not relevant to how many people on average can watch YouTube at the same time, or average speed in a cell, number of dropped calls, number of masts they need for a given customer base (for twice as many customers at same speed you need more than three times as many masts because as you make cells smaller the probability is higher that some are under utilised. Users randomly clump. Scatter a bowl of sugar or 100s & 1000s over a very large table with coarse grid and fine grid.). Obviously you need smaller cells in Suburban/Urban than Rural, but that can be simulated by marking areas for Urban/Suburban on your scatter table and scattering x5 or what ever in those areas (you need a very very big table for this).

There are computer simulations. For simpler background of modeling how many masts / cell sizes for number of customers and a given traffic (measured in Erlang) see
http://en.wikipedia.org/wiki/Buffon%27s_needle

Erlang
http://en.wikipedia.org/wiki/Erlang_%28unit%29

see also
http://en.wikipedia.org/wiki/Stochastic

There is no free lunch. Mobile can never be economically Broadband. In fact a decent nationwide 100% coverage 3Mbps minimum at peak time 4G network is FAR more expensive than 1000Mbps fibre to every home and premises and would use maybe 20x as much electricity.

Clue in name. Mobile, i.e. "on the go" intermittent usage. Selling it as anything more makes the performance rubbish. Complimentary to real Broadband.

Even if the peak speed is 1000 Mbps on Mobile any affordable system with profitable number of customers isn't going to be much better than "off peak" 3G HSPA today when off peak and about x2 better at peak times.

useruser · 14-03-2014 2:45pm

watty wrote: »

There are many reasons and motives for such product claims.

Care to speculate what are the top 3 reasons/motives? Assuming that this is all lies, why would a respected entrepreneur risk his reputation and waste 10 years on something that is just bunkum? Hoax? Delusion? Marketing ploy?

The co-founder of Artemis has a Ph.d. in electrical engineering and wrote his dissertation on MIMO: "Antenna and Algorithm Design in MIMO Communication Systems Exploiting the Spatial Selectivity of Wireless Channels." His "principal architect" colleague has a Ph.d. in Electronics and Communications. Would such qualified individuals really be associated with a hoax in their own field of study? Of course this proves nothing and one could easily imagine a sinister plot of some sort - perhaps to fraudulently attract investment?

Do you think that the demonstrations show nothing at all?

watty · 14-03-2014 3:22pm

I wouldn't speculate.

As I said I have the expertise and experience to demo quite amazing radio systems without cheating. But it would be meaningless.

Ph.d. in Electronics and Communications

All that may mean is that he put in the time.

It doesn't need to be a sinister plot.
Google
FCC satellite GPS lightsquared

A tale of incompetence. They should never have got a "nod" from FCC in the first place. It was never going to work. The problem wasn't GPS receiver quality but high power in adjoining spectrum where there isn't meant to be any. Try tuning more than one station on an AM Radio near an AM transmitter.

Similarly "White space" Radio as originally proposed can't work at all. Google
Hidden transmitter syndrome
Known since maybe 1940s.
So they add a database.
Except that will only work if EVERY setbox & TV is real time connected to the Internet database too! So called "Whitespace" will 100% definitely destroy or degrade reception for 10s of thousands of TV viewers who will never know what is causing it or who to complain to.

Homeplug / Ethernet over the Mains at home if tested properly would never get a certification. They are wide band 1 .. 30MHz or 1 to 200MHz transmitters. They achieve certification by only plugging in one, or both with no data. They can't pass CE or CSA or FCC when actually transferring data.

Many LED, CFL lamps and "plug top" chargers in reality are not compliant with intention of laws.

So someone spending lots of time and money on something that's nonsense isn't unusual. It doesn't validate it. Edison was convinced Cylinder music players and Iron batteries for cars was the way to go.

useruser · 14-03-2014 4:19pm

watty wrote: »

As I said I have the expertise and experience to demo quite amazing radio systems without cheating. But it would be meaningless.

Well, that's interesting too - are you (in a roundabout way) saying that their demonstration is real yet the entire premise is flawed because it's impractical outside of a controlled demo?

watty · 14-03-2014 5:22pm

I can't say if the demonstration is real or not. Does it much matter either way? Probably not.

useruser · 14-03-2014 6:18pm

watty wrote: »

I can't say if the demonstration is real or not. Does it much matter either way? Probably not.

OK, well I guess we'll have to wait and see if anything ever comes of these guys. I won't bet against them just yet. If you don't think that a demonstration of something that should supposedly be impossible matters then I suspect the discussion is finished for now.

useruser · 06-05-2014 9:46am

A bit more detail on Artemis pCell - presentation by founder at Stanford: https://www.youtube.com/watch?v=wGAnDQEQJ_s

Claims that SE Asian operators are interested in deploying native pCell, bypassing LTE.

useruser · 24-02-2015 10:43am

Artemis publish a more detailed white-paper about their pCell product, currently building a network to prove the tech works in SF:

https://gigaom.com/2015/02/23/artemis-is-building-a-4g-network-in-sf-to-prove-its-pcell-tech-works/
http://www.rearden.com/artemis/An-Introduction-to-pCell-White-Paper-150224.pdf

Amazing claim of 59bps/Hz (vs. LTE which approaches just 2bps/Hz in ideal conditions).

Whitepaper excerpt:

pCell has been a journey of over a decade from its first conception to the announcement of its
first commercial deployment, coinciding with the release of this white paper. Despite the fact
that enormous gains and advantages of pCell have been demonstrable during most of its
development, it was such a huge leapfrog beyond conventional wireless that few people who
saw pCell working recognized that what appeared to be “too good to be true” not only was very
real, it wasn’t a research project; it was practical as a commercial offering.

And a demo: https://www.youtube.com/watch?v=UxNntxVWC1g
200mbps in 5MHz - if it's real it's amazing.

watty · 28-02-2015 3:32pm

useruser wrote: »

Amazing claim of 59bps/Hz (vs. LTE which approaches just 2bps/Hz in ideal conditions).

Obviously nonsense, or in very special signal to noise conditions as even 3G and EDGE can get within 6dB of the Shannon Limit.

On a coax you can do x4 to x8 better than LTE, fibre about x16 better as there is less noise. 59bps would need maybe 60W power on the user's device. Or a massive directional aerial / dish on the roof with line of sight.

In real life average signal conditions the speed of LTE is about 1/8th to 1/16th of peak. That will apply to ANYTHING. The Inverse Square Law is inescapable!

"200MHz in 5MHz" needs a very big dish, or massive power, or 2m distance or coax cable!

Snake oil ...

useruser · 28-02-2015 3:54pm

watty wrote: »

Obviously nonsense.

And yet here we have a mature start-up that has managed to convince a spectrum owner to hand over a chunk of their spectrum for a city-scale test. Of course this doesn't prove anything but I would have expected some due-diligence to have been completed prior to agreeing a roll-out.

"200mbps in 5MHz" needs a very big dish, or massive power, or 2m distance or coax cable!

We have (supposed) video evidence demonstrating that they can achieve this with a bunch of standard LTE handsets. Of course this could all be faked but to what end? Nobody is going to give them the time of day if this is snake oil and Perlman would have a lot to lose.
Also, they do not claim to be able to do 200mb to each device, this is 200mb (and 59bps/Hz) in aggregate across all devices. As they are just emulating standard LTE the same limits per device must presumably apply.

I take it that you completely reject their claim of using computed interference from many base stations to create an individual cell per subscriber?

watty · 28-02-2015 5:25pm

You can't break the laws of Thermodynamics.

Qualcomm for instance recently claimed their LTE-U on 5GHz doesn't interfere with WiFi on same band on "heavy loading". This is nonsense as basically it's claiming the spectrum can have twice as much traffic.

If you have N equal users fully utilising spectrum to the Shannon Limit, then they at BEST get 1/Nth performance/speed each compared with a single user.

You're up against basic physics and mathematics.

useruser · 01-03-2015 8:35am

watty wrote: »

You're up against basic physics and mathematics.

Which of course leaves us with the question: what are we seeing demonstrated here? is it just fraud? Surely a live trial in a real environment will quickly uncover whether or not this is real and viable - I just don't see why they would be pursuing it to such a large scale if it were nonsense.

If what they are demonstrating is true then even if it only works in a lab environment it is still remarkable.

bealtine · 01-03-2015 11:14am

useruser wrote: »

Which of course leaves us with the question: what are we seeing demonstrated here? is it just fraud? Surely a live trial in a real environment will quickly uncover whether or not this is real and viable

Ah yes the old perfect environment trick....mobile companies employ this all the time for their "demonstrations"

watty · 01-03-2015 4:09pm

useruser wrote: »

Which of course leaves us with the question: what are we seeing demonstrated here?

It's impossible to know exactly what is being demo'ed, but it's very like the flavour of all the discredited Cold Fusion demos.

useruser wrote: »

Surely a live trial in a real environment will quickly uncover whether or not this is real and viable - I just don't see why they would be pursuing it to such a large scale if it were nonsense.

No, it won't as it's not sensibly audited by experts that are not vested interests.
Money & Ego.

useruser wrote: »

... even if it only works in a lab environment it is still remarkable.

No, it's not. You can do stuff in a Lab any stage Magician would be proud of!

useruser · 02-03-2015 11:20am

watty wrote: »

It's impossible to know exactly what is being demo'ed, but it's very like the flavour of all the discredited Cold Fusion demos.

OK, so your assumption is that this is simply fraud. Fair enough, I certainly can't argue otherwise at the moment.

No, it's not [remarkable]. You can do stuff in a Lab any stage Magician would be proud of!

If (and I admit it's a big if) what they are showing in the lab is real (i.e. 200mb of LTE throughput in 5MHz) then this is of course something remarkable, even if it only worked in a lab environment. I am fully aware that this could easily be staged but again I have to ask: to what end?

useruser · 02-12-2015 11:28am

Artemis' pCell continues to grow legs.

Here's a detailed technical presentation (as given to the IEEE) by the chief scientist. Again re-iterating the claim of 200mb in 5MHz of spectrum or nearly 60bps/Hz for the entire system. https://www.youtube.com/watch?v=2ETMzxkyTv8&feature=youtu.be

The IEEE paper is here: http://files.artemis.com/151109-Achieving+Large+Multiplexing+Gain+with+pCell-IEEE+Asilomar+2015.pdf

Also of note is that Nokia Networks has recently announced that they will be testing pCell with some existing operator customers in 2016: http://files.artemis.com/Nokia-Networks-MoU.pdf It will be very interesting to hear how (if) this performs in real-world conditions.

Snake oil or broadband panacea? DIDO.

Comments