DIDO promises wireless breakthroughs

7/28/2011 11:29 AM EDT

If you are a wireless tech expert, please provide your thoughts on the white paper.

7/28/2011 1:10 PM EDT

I'm not an expert on wireless technology, but it looks like this requires a massive amount of processing power in order to be feasible. Using the term "Data Center" for the basestation corroborates with this as well.

This might be tough to implement in fast-changing environments, as you constantly need to re-characterize the environment in order to determine how to form the beam. I would imagine that this technology is difficult to implement when the client is moving, since you have to constantly re-calculate the waveforms as the client enters and leaves different environments.

Another question is what does the client's antenna configuration look like? It might require a lot of antennas in order for each transmitter on the client to successfully create a localized beam. Having all of these antennas and transmitters could seriously increase the power consumption of mobile devices to the point of infeasibility.

7/28/2011 1:24 PM EDT

This seems to me like this guy is borrowing the work of Whitespace spec and call it his invention. Go look at IEEE 802.22 standard. All this sounds awfully similar to this, as there envisioned use case data canter to people using the sub Gigahertz freq band.

7/28/2011 1:43 PM EDT

I have been working on MIMO wireless for the past 7 years and I see that this is same as Multi-User MIMO. Just calling MU-MIMO as DIDO doesn't change anything. Even the un-initiated can get a quick overview of MU-MIMO from the following link:
http://en.wikipedia.org/wiki/Multi-user_MIMO

It would be interesting for everyone to know that 802.22, 802.11ac, LTE-Advanced, 802.16e, 802.16m and many other drafts and completed standards already use MU-MIMO extensively.
In fact many chips that use this technology are already out and consumers would find products with this technology in their hands this year.

I feel it is not correct to re-name a well-known technology developed over many years by multilple groups around the world and try to own it up as a major breakthrough done by one individual or organization!

7/28/2011 3:46 PM EDT

I've been working in MIMO and multiple antenna systems for the last 16 years, and yes, it is not LIKE MU-MIMO, it IS MU-MIMO. You will also see that they omitted any references in their white paper. Indeed LTE and 802.16x use this technology already. Pitty that someone can make the claim that they "discovered" things that a whole industry has been working on long before his patent was filed. I also saw a youtube video (http://www.youtube.com/watch?v=1QxrQnJCXKo ) that SP presented at Colombia where he claims that he proved Shannon wrong... The sad thing about the US patent system is that anyone can get a patent on virtually anything - regardless if it is old news or not. The US patent office needs serious overall - otherwise we will have to continue to put up with this kind of self indulgence.

7/28/2011 4:41 PM EDT

My mother taught us if it sounds to good to be true...

7/28/2011 5:44 PM EDT

quoting from the article...
"I do not see that much new in this patent that was not known," Rabaey added. "Actually, LTE-Advanced does use similar approaches," he said"
what's wrong with this picture? - if we assume as stated that there's nothing unknown in this patent, why was it granted? Looks like we either have to believe there's something seriously wrong with the US patent system OR we have to accept that Information Theory and Shannon are wrong (not in my experience) - the theory shows the absolute limits for error-free communication in the presence of noise, and the limits of channel capacity, and modern systems can be within a few dB of these limits, yet this fellow claims a hundred-fold or even a thousand-fold (30dB!) improvement? Believe it when you see it...

7/28/2011 5:58 PM EDT

I 2nd Luis comments that we should not rush to judge. Actually, they provide enough information for me that what they are doing makes a TON of sense. The complexity lies in the accuracy needed between the different APs. But I totally see the beauty in the approach!

Since it might not able all that clear to others, some of the concepts that they more than likely employ are no different than pre-distorting the data that you feed to a DAC to account the Zero-Order Hold effect impacting the frequency response of the DAC's output. However, now apply this same concept to multiple APs, i.e. the time-ing and power-levels between APs have to be known apriori, which is already loaded into the DIDO Data Center to thus allow the appropriately unique patterns to be generated for each AP. The extra hardware on the user side would involve performing the properly signal power level adjustments before presenting a final signal to the user's wireless protocol handling device; be it WiFi, WiMax or LTE. So yes it can get VERY complicated, but from my experiences it is very viable and it is TOTALLY different than the 802.22 and the other standards as they involve dynamics allocation of different and unused frequencies, etc. When is also rather complex and has not made any major progress toward commercialization.

The folks at Rearden have my support with their effort to get DIDO rolling! Finally some real innovation and not patches to old ways of thinking!!! ;)

7/29/2011 5:04 AM EDT

"In a fully deployed DIDO system, the antennas are far closer to the user on average than in a cellular system where, for example, many users are on the fringes of the system, he explained."

and

"The average power is easily one-tenth that of cellular, and in many cases its one one-hundredth"

Hmmm, I think I can see how this might work .. but thats nothing new. What I don't understand is how he has dramatically reduced the latency - maybe there is not much FEC coding?

7/29/2011 10:24 AM EDT

The only problem I have is the name: DIDO.
The name is a little too suggestive.

They should call it DINDOUT instead.

I know this has absolutely nothing to do with the technical merits but I can hear some of my more comic friends say it now:

"Yeah, my cell uses DIDO inside. hehe."

7/29/2011 12:22 PM EDT

I am copying three comments from Steve Perlman I received in email this morning: DIDO is not at all related to 802.22. 802.22 is a conventional cognitive radio system that avoids interference at the cost of data rate per user, because it is limited by Shannon's Law.

7/29/2011 12:23 PM EDT

Here is the second comment from Steve Perlam's email: DIDO also is not MU-MIMO. MU-MIMO (and MIMO) achieve perhaps 3x-4x the Shannon limit. This (3x-4x) limitation comes from the fact that they rely on multiple antennas close to each other at a base station, which does not provide enough diversity between the transmit antennas. There are no examples of general-purpose MIMO systems that achieve 10X we have working now with DIDO.

7/29/2011 12:24 PM EDT

Here is the third and final comment Steve Perlman submitted via email: A key reason DIDO is able to achieve such high multiples of 10x [Shannon's limit] and more is because the antennas are widely distributed and achieve unprecedented diversity which is important for creating multiple simultaneous channels. There are no published reports in the literature that show 10x the Shannon limit in a general-purpose wireless system where a dense group of users can be anywhere. DIDO is the first to achieve this result, which is why we waited until we had that working to be sure we had developed something distinct from any previous wireless system.

7/29/2011 1:14 PM EDT

My comment is: For example if you increase the number of cell towers ten fold then you'll get a ten fold increase in capacity. It seems the "inventor" is suggesting this type of arrangement. I see no proof for the claims. The skeptic in me sees this as a "100MPG carburetor" claim.

7/29/2011 4:08 PM EDT

Replace every base station with a data center, and maybe this could fly.

Now if only the technology existed to put a modern data center in a small box and mount it to a traffic light...

7/29/2011 4:58 PM EDT

Skeptic that you can reliably share same channel (frequency,space) and recover multiple signals without penalty ( w/o limit bonudaries in sight).
Some level of processing may help some trivial discrete number (single digit) but no boundaries in sight per white paper ( more than 10x ???).

7/30/2011 9:52 AM EDT

I am not an expert at all in this wireless technology and standard, 802.xyz. Somehow I could follow datacenter concept in this article. I don’t get how those packets will be filtered inside of handset. If it will be implemented as inventor claimed, it will be another milestone in wireless world. I encourage author to go through steps of other saying.

Step 1. It is impossible.
Step 2. It is not much different than it has been.
Step 3. It was my idea.

Good Luck!

7/30/2011 9:54 AM EDT

I am not an expert at all in this wireless technology and standard, 802.xyz. Somehow I could follow datacenter concept in this article. I don’t get how those packets will be filtered inside of handset. If it will be implemented as inventor claimed, it will be another milestone in wireless world. I encourage author to go through steps of other saying.

Step 1. It is impossible.
Step 2. It is not much different than it has been.
Step 3. It was my idea.

Good Luck!

7/30/2011 2:12 PM EDT

Ok, guys. Here is a challenge. I will proof to you that the US patent system is a farce. I will, with your help and a patent lawyer friend of mine, take a well known idea/technology and get it patented. In fact I will fast track the patent to be granted within a 2 year period in order to not torture all of us with a 10 year wait for the patent to be granted (I have done this successfully in the past). The idea is this:
1) If I get 10 people to reply to this post that this is a good/worthy cause, I will open a Facebook page soliciting ideas (known technical ideas) for the patent from the public.
2) I will narrow the ideas down to the top 5 "ideas" and file a patent on one of the ideas. I will describe the idea with new terms, e.g. a piece of toast will be a "slightly browned organic compound" or even, MU-MIMO being DIDO etc. The reason for choosing one of the five ideas and not disclosing which one before the patent is granted is to ensure that the USPTO does not get wise on our scheme and decline the patent.
3) I will obviously keep all the posts and ideas etc and when the patent is granted, we can compare.
4) Unfortunately this will cost money - about $10,000, so if we can get 1,000 engineers/scientist to contribute $10 each to the experiment, we can make an historic point that could well change patent law, patent litigation and the USPTO (and DIDO) forever.

So, let me know if you think this is a worthy cause.

7/30/2011 11:25 PM EDT

@JackWin: You get my $10 bucks. But I would already cocede the point that you can get a patent examiner, many of them fairly junior people, to approve something they should not approve.

7/31/2011 7:31 PM EDT

The referenced White Paper is interesting, but serves best as a brief gateway for diving into the patents: While it doesn't permit the technology to be analyzed, it does provide some useful insights.

The relevant patents are here: http://www.google.com/search?tbo=p&tbm=pts&hl=en&q=ininventor:%22Antonio+Forenza%22

Each of the above patents references a massive number of other patents and publications (papers and articles).

The most accessible patent seems to be this one: http://www.google.com/patents?id=etPwAAAAEBAJ&printsec=abstract&source=gbs_overview_r&cad=0#v=onepage&q&f=false After the first page, the main text resumes on page 113.

-BobC

7/31/2011 8:12 PM EDT

I'm still working through the white paper, but on the patent system side of this the question of whether this is all new or rehash of old will happen in the patent challenge process rather than as an assumption of the original patent award. Current wireless systems are a jungle of intertwined patents held by companies with very deep pockets. I hope these guys are prepared for prolonged challenges.

Larry M.

8/1/2011 3:04 AM EDT

I think it's too early to poo-poo DIDO, but they brought this onto themselves by the way the white paper is marketed.

Really, if the patent/white paper is reflective, we're talking about large-scale multi-user MIMO in very large cells enabled through over-the-horizon links. The underlying technology is not new, but it hasn't been studied much on HF spectrum or with the dimensionality he seems to be claiming.

Perhaps he's got a very specific application that fits perfectly under these parameters, which would be great, but taking the revolutionary angle really does undermine that.

8/1/2011 11:50 AM EDT

Perlman has been on the leading edge of modems for 20 years, so if DIDO is real it's no surprise that he discovered it. The wide bandwidth path is one-way. If users want to send data upstream, the bandwidth has the same constraints as current technology. However, Internet data flow is predominately downstream.

8/1/2011 8:37 PM EDT

Steve Perlman also granted an interview with "Bloomberg Business Week" (Aug 1-7, 2011 pages 62-67 as well as a prepublication release of the white paper so there may be some glimmerings of explanation available there to augment these discussions.

8/4/2011 1:10 PM EDT

Rick, if you can get this question answered from Steve Perlman, I think it could be instructive: Does the performance degrade based on the nature of the data streams being sent to different clients (e.g. with *random* data)?

For example, take two clients C1 and C2, that are within interference range of each other. They download live streams from two websites, S1 and S2, respectively. S1 and S2 each are streaming uniquely different random data, generated similarly to how the "million random digit challenge" file is generated:

http://marknelson.us/2006/06/20/million-digit-challenge/

If the S1 and S2 data streams can be downloaded at 100% accuracy at the same rate when C1 and C2 are within interference range, as when they are not, then that could point to a major breakthrough and I'll be looking forward to hearing more from the experts on just how it works.

On the other hand, if performance does indeed degrade with random data, then it can be characterized and analyzed like a compression algorithm. The question there will be how much better is this form of "interference-based compression" as compared with other compression techniques used in the various wireless standards.

8/4/2011 3:01 PM EDT

Basically, DIDO is exactly the same as distributed multiuser MIMO (MU-MIMO).

In centralized MU-MIMO, a AP (base station) has M antennas, and there are M stations, each of them has only 1 antenna. The AP requests channel estimation on each station so that it can obtain all Channel State Information(CSI) of all its M antennas to that station. Once the AP has all M-by-M CSI, by simple linear algebra (matrix inversion), AP can pre-filter the signals so that each user can receive signal with free of interference from others.

In the distributed version (DIDO), the only difference is that now we have M APs and each one has only 1 antenna. Basically, M APs are cooperated so that they work like centralized MU-User. The cooperation is achieve by asking all the APs sending the CSI to the centralized server (cloud). Then the cloud will do the linear algebra and request the APs to send appropriate waveform to out to the air.

The DIDO idea definitely works. But there are at least two caveats (discussed in my next comments):

8/4/2011 3:07 PM EDT

1) Obtaining CSI takes time and it scales up with number of stations and the change of environment (e.g. moving objects around the station which change the CSI). Also, while the channel estimations are required to be done in interference-free channels. Why? while the Channel is being estimated, AP doesn't know the station's CSI. Hence, AP or cloud has no information to do the linear algebra. So clean channel is needed. So, if there are people/ objects moving, sorry, DIDO also faces the same situations that traditional wireless tech is facing.
\n
2) Cell edge effect. It's reasonable to assume that no AP can cover the whole Earth. There will be some areas that a AP can cover while the other AP cannot cover as they are placed in two locations. In this case, even though a station is in the coverage area that two APs are not overlapping, the DIDO concept still request two APs to work together to give the best signal to that station. Therefore, even though the DIDO system may just want to cover a small area, it still need to handle the interference coming from the areas that are out of its intended coverage. The complexity in the cloud server is extremely high then. Think of how many users AT&T has, say 2 millions. So if DIDO want to replace AT&T, they need to do a 2M x 2M matrix inversion every mini second. Not to mention, you need to silence other 2M-1 users when your channel is being estimated as I discussed in point 1. \n

8/11/2011 10:32 AM EDT

I have worked with wireless networks for the past 15 years, both in research and in startups. I just want to make a simple statement that may shed some some light on whether this can work and more importantly how novel this invention is. Then you make up your mind (mine is made):

Go to DIDO website and downoad their whitepaper. The way you can eliminate interference in wireless channels is explained there: You take many potentially interfering WiFi access points (AP) (as many as you want , 10, 100) and you link them together with wires, so they become all connected to a shared node (DIDO calls it a central "data center"). Once these 10 or 100 AP antennas are connected, what you have effectively built is a single virtual transmitting device with 10 or 100 transmit antennas. Such a device is well known to the wireless industry as "Distributed Antenna System".
Now send linear combinations of all the data streams expected by the 10 or 100 users to all the antennas, but carefully select the coefficient of the combinations such that a given data stream only reaches its intended user why not interfering with the other 9 or 99 users. Simple algebra using orthogonality principle will do the trick This process is well known to the wireless industry as "transmit beamforming". Once interference is gone, you can do the same process with as many users as you like, just link more AP together (this has a cost by the way).

So none of the key components of the invention are new.

Now you may say, but wait Distributed Antenna Systems are known , beamforming is known, but combining these two concepts must be novel. Problem: This precisely was proposed 5 years ago by Alcatel-Lucent under the name of NETWORK MIMO. Network MIMO (also called as "CoMP") is an advanced version of Multiuser MIMO and is currently being discussed for standardization for LTE cellular networks and beyond as a way to resolve interference in cellular networks.

8/20/2011 6:47 PM EDT

I like innovative ideas, but not hype. What's wrong with saying that this is MIMO, with a slight twist? That's what it is. You spread out the MIMO antennas, locating each closer to a user.

A couple of other thoughts. No one is violating Shannon's law here, or in MIMO. Shannon's law applies completely. In MIMO or DIDO, you treat each propagation path like a separate channel. That's all. And you use clever filtering to make sure the channels stay separate. If the same-frequency channels were separated by large distances, no one would be BSing about violating Shannon's limits, right? This is exactly the same thing. The separation is not by distance, but by propagation path. So what?

Furthermore, you certainly can achieve 10X multiples using MIMO. Search on 16X16 MIMO schemes, for example. It's been done and demonstrated in the field.

11/6/2011 3:24 AM EST

I think this DIDO technology is another variant of spatial diversity multiple access scheme.
TDMA FDMA CDMA SDMA... I wonder what will be the next big thing in wireless technology in the coming 20 years?

2/26/2012 1:39 AM EST

I got the impression that DIDO was being designed for large-scale distribution. "It is important to note that DIDO was designed to be a mass-market product"



If so,
The data center infrastructure that will have to exist would be understated as MASSIVE.



Additionally,
If I'm reading the white paper correctly there will be a need to have end points to perform the actual transmissions.
"having many distributed antennas" once again, MASSIVE deployment


Ever been around something thats transmitting at the higher end wavelengths?
"frequencies from 1 MHz to 1 GHz" the energy required for 100MHz and greater, not to mention the environmental implications = wow


The math that would be involved, not just at the DataCenter level, but at the recipient level to achieve receipt, goes way beyond simple tower hopping.



remember even a medium size city could contain ###xMillions of nodes

and none of this takes into consideration the possibilities of advanced devices being commercially available in the next 3-5-7 years



This may have small enterprise applications, but I cant see it as being a progressive solution.

"Merger of cloud computing, cloud gaming and cloud wireless"

Now "clouds" with VM's floating /drool

I may get schooled as a result of this post, but it's the first impressions I got from reading the White paper, and some of the posts.