Revisiting DIDO Wireless

A wireless icon
Image via Wikipedia

I’ve had some time to think about the DIDO wireless idea, and still think it has a very important part to play in the future – assuming the trial conducted of 10 user nodes is truthful. Before I explore the commercial benefits of this idea, I will first revisit the criticisms as some have merit, and will help scope a realistic business case.

Analysis

Weaknesses

  • One antenna per concurrent node – The trial used 10 antenna for 10 user nodes. Each antenna needs a fixed line or directional wireless backlink – this would imply poor scalability of infrastructure. [Update: This is likely so, but Artemis claim the placement of each antenna can be random – whatever is convienient]
  • Scalability of DIDO – We are told of scaling up to 100s of antenna in a given zone. I question the complexity of the calculations for spatial dependent coherence, I believe the complexity is exponential rather than linear or logarithmic. [Update: Artemis pCell website now claims it scales linearly]
  • Scalability of DIDO controller – Given the interdependence on signals, is the processing parellelisable? If not this also limits the scale of deployment. [Update: Artemis claim it scales linearly]
  • Shannon‘s Law not broken – The creators claim breaking the Shannon’s law barrier. This appears to be hyperbole. They are not increasing the spectrum efficiency, rather they are eliminating channel sharing. The performance claims are likely spot on, but invoking “Shannon‘s Law” was likely purely undertaken to generate hype. Which is actually needed in the end, to get enough exposure for such a revolutionary concept.

Neutral

Discussion surrounding neutralised claims which may be reignited, but are not considered weaknesses or strengths at this point in time.

  • BackhaulEven though the antenna appear to require dispersed positioning, I don’t believe that backhaul requirements to the central DIDO controller need to be considered a problem. They could be fixed line or directional wireless (point to point). [Update: This is not really a problem. Fibre is really cheap to lay in the end for backhaul, it’s most expensive for last-mile. Many Telcos have lots of dark fibre, not being used and Artemis is partnering with Telcos, rather than trying to compete with them]
  • DIDO Cloud Data Centre – I take this as marketing hyperbole. Realistically a DIDO system needs a local controller, all other layers above such a system are distractions from the raw technology in question. And as such, the communication links between the local controller and antenna need not be IP transport layer links, but would rather be link layer or even physical layer links.
  • Unlimited number of users – Appears to also be hyperbole, there is no technological explanation for such a sensational claim. We can hope, but not place as Pro until further information is provided. [Update: It does scale linearly, so this is a fair claim when compared to current Cell topology or if pCell was was limited to exponential processing load]
  • Moving User Nodes – Some may claim that a moving node would severely limit the performance of the system. However this pessimistically assumes a central serial CPU based system controls the system (a by-product of Reardens “Data Centre” claims). In reality I believe it’s possible for a sub-system to maintain a matrix of parameters for the main system to encode a given stream of data. And all systems may be optimised with ASIC implementation. Leaving this as a neutral but noteworthy point.
  • Size of Area of Coherence – Some may claim a problem with more than 1 person in an area of coherence, assumed to be around one half wavelength. How many people do you have 16cm away from you (900Mhz)? Ever noticed high density urbanisation in the country? (10-30Mhz for ionosphere reflection – <15M half wavelength) [Update: demonstrations have shown devices as close as 1cm away from each other – frequency may still be a limiting factor of course, but that is a good result]
  • DIDO is MIMO – No it’s very similar, but not the same and is likely inspired by MIMO. Generally MIMO is employed to reduce error, noise, multipath fading. DIDO is used to eliminate channel sharing. Two very different effects. MIMO Precoding creates higher signal power at a given node – this is not DIDO. MIMO Spatial multiplexing requires multiple antenna on both the transmitter and receiver, sending a larger bandwidth channel via several lower bandwidth channels – DIDO nodes only need one antenna – this is not DIDO. MIMO Diversity Coding is what it sounds like, diversifying the same information over different antenna to overcome wireless communication issues – this is not DIDO. [Update: Artemis and the industry and now standardising calling it a C-RAN technology]
  • 1000x Improvement – Would this require 1000 antenna? Is this an advantage given the amount of antenna required? MIMO is noted to choke with higher concurrency of uses. Current MIMO systems with 4 antenna can provide up to 4x improvement – such as in HSPDA+. Is MIMO limited in the order of 10s of antenna? Many many questions… [Update: This is likely so, but Artemis claim the placement of each antenna can be random – whatever is convenient]

Strengths

  • Contention – Once a user is connected to a DIDO channel, there is no contention for the channel and therefore improved latency and bandwidth.
  • Latency – Is a very important metric, perhaps as important as bandwidth. Latency is often a barrier to many innovations. Remember that light propagates through optical fibre at two-thirds the speed of light.
  • Coverage – It seems that DIDO will achieve coverage and field less black spots than what is achievable with even cellular femtocell. Using new whitespace spectrum, rural application of pCell would be very efficient, and if rebounding off the Ionosphere is still feasible, the answer to high speed, high coverage rural internet.
  • Distance – DIDO didn’t enable ionosphere radio communications, but it does make ionosphere high bandwidth data communication possible. Elimination of inter-cell interference and channel sharing make this very workable.
  • Physical Privacy – The area of coherence represents the only physical place the information intended for the user can be received and sent from. There would be potential attacks on this physical characteristic, by placing receivers adjacent to each DIDO antenna, and mathematically coalescing their signals for a given position. Of course encryption can still be layered over the top.
  • Bandwidth – The most obvious, but perhaps not the most important.
  • [New] Backward Compatibility – Works with existing LTE hardware in phones. Works better if using a native pCell modem with better latency performance particularly. Seamless handoff to cell networks, so it can co-operate.
  • [New] Wireless Power – Akbars (See Update below) suggested this technique could be used for very effective Wireless Power, working over much larger distances than current technology. This is huge!

Novel Strength

This strength needed particular attention.

  • Upstream Contention Scheduling – The name of this point can change if I find or hear of a better one. (TODO…)

Real World Problems

Unworkable Internet-Boost Solutions

I remember reading of a breakthrough where MEMS directional wireless was being considered as an internet boost. One would have a traditional internet connection and when downloading a large file or movie, the information would be sufficiently cached in a localised base station (to accommodate a slow backlink or source) and then forwarded to the user as quickly as possible. This burst would greatly improve download times and a single super speed directional system would be enough to service thousands of users given its’ extreme speed and consumers limited need for large transfers. Of course even such a directional solution is limited to line of sight, perhaps it would need to be mounted on a stationary blimp above a city…

Mobile Call Drop-outs

How often do you find yourself calling back someone because your call drops out? Perhaps it doesn’t happen to you often because you’re in a particularly good coverage area, but it does happen to many people all the time. The productivity loss and frustration is a real problem which needs a real solution.

Rural Service

It is very economical to provide high-speed communication to many customers in a small area, however when talking of rural customers the equations are reversed. Satellite communication is the preferred technology of choice, but it is considerably more expensive, is generally a lower bandwidth solution and subject to poor latency.

Real World Applications

The anticipated shortcomings of DIDO technology need not be considered as deal breakers for the technology. The technology still has potential to address real world problems. Primarily we must not forget the importance/dominence of wireless communications.

Application 1: A system could be built such that there may be 10 areas of coherence (or more), and can be used to boost current technology internet connections. One could use a modest speed ADSL2+ service of 5Mbps and easily browse the bulk of internet media {Text, Pictures} and then still download a feature-length movie at gigabit speeds when downloaded. This is a solution for the masses.

Application 2: DIDO allows one spectrum to be shared without contention, but that spectrum need not be a single large allocation of spectrum, it could mean a small (say 512Kbps) but super low latency connection. In a 10 antenna system, with 20Mhz of spectrum and LTE-like efficiency this could mean 6000 concurrent active areas of coherence. So it would enable very good quality mobile communication, with super low latency and practically no black-spots. It would also enable very effective video conferencing. All without cellular borders.

Applications 3 and 4: The same as Applications 1 and 2, but using a long-range ionosphere rural configuration.

Conclusions

We still don’t know too much about DIDO, the inventors have surrounded their idea with much marketing hype. People are entitled to be cautious, our history is littered with many shams and hoaxes, and as it stands the technology appears to have real limitations. But this doesn’t exclude the technology from the possibility of improving communication in the real world. We just need to see Rearden focus on finding a real world market for its’ technology.

UPDATE

  • [2017-08-31] I have noticed another news update on their site dated Jule 20th.
    • This mainly concerns the use of 600MHz spectrum once used by TV.
    • In my previous update they hinted at this band “Or in any band from 600 MHz to 6 GHz”
    • This is prime wireless spectrum which cuts through building walls.
    • But it’s also a spectrum to suffer most from overlapping-cell problems. So this is where pCell can shine the brightest, and they’ve probably realised this. They are playing their best cards.
    • It will be interesting to see how they position themselves as new Satellite services come to market. 600MHz is certainly capable of rural reach.
  • [2017-01-10] Finally, the hint text has disappeared completely, to be replaced with
    • “supports a different protocol to each device in the same spectrum concurrently” – following up on their last update
    • “support multiple current and future protocols at once.” – this is a great new insight. They have right up top, that pCell supports 5G, and future standards. So without considering the increased capacity, customers don’t need to keep redeploying new hardware into the field.
    • “In the future the same pWave Minis will also support IoT” – there are standards floating around, and what better way to implement security for IoT, than physically isolated wireless coherence zones, and perhaps very simplistic modulation.
    • “precise 3D positioning” – This confirms one of my predictions, pCell can supercharge the coming autopilot revolution
    • “and wireless power protocols” – as I always suspected. However, it still seems impractical. This is likely just a candy-bar/hype statement.
    • “Or in any band from 600 MHz to 6 GHz” – it’s interesting to learn this specification – the limits of typical operation of pCell. I note they have completely abandoned long-wave spectrum (for now at least).
    • “pWave radios can be deployed wherever cables can be deployed” – I still think fibre/coax is going to be necessary, wireless backhaul is unlikely to be scalable enough.
    • “Typically permit-free” – does this refer to the wireless signal I wonder? Very interesting if so. It could also refer carrier licensing, because you’re only carrying data, information is only deduced back at the data centre.
    • “can be daisy-chained into cables that look just like cable TV cables” (from Whitepaper) – so perhaps long segments of coax are permitted to a base-station, but that base-station would likely require fibre out.
    • “pCell technology is far less expensive to deploy or operate than conventional LTE technology” – they are pivoting away from their higher-capacity message, now trying to compete directly against Ericson, Huawei, and others.
  • [2016-02-25] pCell will unlock ALL spectrum for mobile wireless. No more spectrum reservations. pCell could open up the FULL wireless spectrum for everyone! I hope you can grasp the potential there. Yesterday I read a new section on their website: “pCell isn’t just LTE”. Each pCell can use a different frequency and wireless protocol. This means you can have an emergency communication and internet both using 600Mhz at the same time meters away! In 10 years, I can see the wireless reservations being removed, and we’ll have up to TERABITS per second of bandwidth available per person. I’m glad they thought of it, but this is going to be the most amazing technology revolution of this decade, and will make fibre to the home redundant.
  • [2015-10-03] It’s interesting that you can’t find Hint 1 on the Artemis site, even when looking back in history (Google), in fact the date of 2015-02-19 it reads “Feb 19, 2014 – {Hint 2: a pCell…”, which is strange given my last update date below. Anyway the newest Hint may reveal the surprise:
    • “Massless” – Goes anywhere with ease
    • “Mobile” – outside your home
    • “Self-Powered” – either Wireless Power (unlikely) or to wit that this pCell is like some sort of Sci-Fi vortex that persists without power from the user.
    • “Secure” – good for privacy conscious and/or business/government
    • “Supercomputing Instance” – I think this is the real clue, especially given Perlman’s history with a Cloud Gaming startup previously.
    • My best guesses at this stage in order of likelihood:
      • It’s pCell VR – already found in their documentation, and they just haven’t updated their homepage. VR leverages the positioning information from the pCell VRI (virtual radio instance) to help a VR platform both with orientation as well as rendering.
      • Car Assist – Picks up on “Secure” and the positioning information specified for VR. VR is an application of pCell to a growing market. Driverless is another growing market likely on their radar. Driverless cars have most trouble navigating in built up, busy environments and particularly round abouts. If pCell can help in any way, it’s by adding a extra absolute position information source this cannot be jammed. Of course the car could also gain great internet connectivity too, as well as tracking multiple vehicles centrally for more centralised coordination.
      • Broader thin-client computing, being beyond “just communications”, although one can argue against that – pCell is communications an enabler. This would include business and gaming.
      • Emergency Response. Even without subscription it would be feasible to track non-subscribers location.
  • [2015-02-19] Read this article for some quality analysis of the technology – http://archive.is/ZTRhf [Archive Link] – Old broken link: http://akbars.net/how-steve-perlmans-revolutionary-wireless-technology-works-and-why-its-a-bigger-deal-than-anyone-realizes.html
  • [2015-02-19] Artemis have on their website – “Stay tuned. We’ve only scratched the surface of a new era.…{Hint: pCell technology isn’t limited to just communications}’ – I’m gunning that this will be the Wireless Power which Akbars suggested in his blog article. [Update 2015-10-03 which could be great for electric cars, although efficiency would still be quite low]
  • [2016-06-02] Technical video from CTO of Artemis – https://www.youtube.com/watch?v=2ETMzxkyTv8
    • Better coverage – higher density of access points = less weak or blackspots
    • When there are more antenna than active users, quality may be enhanced
    • Typical internet usage is conducive for minimising number antenna for an area
    • pCell is not Massive MIMO
    • pCell is Multi User Spatial Processing – perhaps MU-MIMO [see Caire’03, Viswanath’03, Yu’04]
    • According to mathematical modelling, densely packed MIMO antenna cause a large radius of coherent volume. Distributed antenna minimises the radius of coherent volume. Which is intuitive.
    • see 4:56 – for a 3D visulasation of 10 coherent volumes [spatial channels with 16 antennas. Antenna are 50m away from users – quite realistic. Targetting 5dB sinr.
    • pCell Data Centre does most of the work – Fibre is pictured arriving at all pCell distribution sites.
    • 1mW power for pCell, compared to 100mW for WiFi. @ 25:20
Enhanced by Zemanta

DIDO – Communication history unfolding?

First of all I just want to say – THIS MAY BE HUGE!!

I read this article last night: http://www.gizmodo.com.au/2011/08/dido-tech-from-quicktime-creator-could-revolutionise-wireless-broadband/

In plain english, a company has discovered a way to dramatically improve mobile internet. It will be 5 – 10 years before it’s commercialised, however I believe it will happen sooner, with many realising just how revolutionary it will be, investing more money, attracting more resources to get it done sooner.

I am not a representative of the compnay, but have been involved in understanding and pondering wireless technology, even coming up with faster and more efficient wireless communication concepts, but none as ground-breaking as this one. I don’t claim to know all the details for certain, but having read the whitepaper I beleive I can quite accurately assume many details and future considerations. Anyway I feel it’s important for me to help everyone understand it.

How does it work (Analogy)?

Imagine walking down the street, everything is making noise, cars, people, the wind. It’s noisy, someone in the distance is trying to whisper to you. Suddenly all the noise dissappears, and all you can hear is that person – clearly. This is because someone has adjusted all the sounds around you to cancel out, leaving just that persons voice.

How does it work (Plain English)?

When made available in 5-10 years:

  • Rural users will have speeds as fast as in CBDs, receving signals from antennas as far as 400km away!
  • In cities there will need to be several antennas in within ~60km of you
    • today there are so many required for mobile internet, the number will be reduced…
    • the number of antennas per mobile phone towers and buildings will be reduced to just one.
  • there will be a central “server” performing the mathematical calculations necessary for the system.

The most technical part (let’s break it down):

  1. Unintended interference is bad (just to clarify and contrast)…
  2. DIDO uses intereference, but in a purposeful way
  3. DIDO uses multiple antennas, so that at a particular place (say your house), they interfere with each other in a controlled way, leaving a single channel intended for you.
  4. It’s similar to how this microphone can pick up a single voice in a noisy room – http://www.wired.com/gadgetlab/2010/10/super-microphone-picks-out-single-voice-in-a-crowded-stadium/
    but a little different…

How does it work (Technical)?

I have been interested in two related concepts recently:

  1. Isolating a single sound in a noisy environment – http://www.wired.com/gadgetlab/2010/10/super-microphone-picks-out-single-voice-in-a-crowded-stadium/
  2. I saw an interview with an ex-Australian spy who worked at a top secret facility in Australia in co-operation with the US. The guy was releasing a book revealing what he can. From this facility he spied on radio communications around the world. I wondered how and then figured they likely employ the “super microphone” method.

When I heard about this technology last night, I didn’t have time to look at the whitepaper, but assumed the receivers may have “super microphone” sort of technology. It turns out the inverse (not opposite) is true.

Scenario:

User A’s radio is surrounded by radios from DIDO. The DIDO server calculates what signals need to be generated from the various radios such that when converging on User A, they “interfere” as predicted to leave the required signal. When there are multiple users the mathematical equations take care of working out how to converge the signals. As a result, the wireless signal in the “area of coherence” for the user, is as if the user has the full spectrum 1:1 to an external wireless base station.

Implications for domestic backhaul

There would need to be fibre links to each of the antennas deployed, but beyond that remaining backhaul and dark fibre will rapidly become obsolete. DIDO can reach 400km in the rural mode, bouncing off the ionosphere and still maintaining better latency than LTE at 2-3ms.

Physical Security?

We hear about quantum communication and the impossibility to decipher the messages. I believe a similar concept of physical security can be achieved with DIDO. Effectively DIDO provisions areas of coherency. Areas in 3D space where the signals converge, cancelling out signal information intended for other people. So effectively you only physically receive a single signal on the common spectrum, you can’t physically see anyone else’s data, unless you are physically in the target area of coherency. This however, does not mean such a feature enables guaranteed privacy. By deploying a custom system of additional receivers that can sit outside the perimeter of your own area of coherency, you can sample the raw signals before they converge. Using complex mathematics and empowered with information of the exact location of the DIDO system antennas, one would be theoretically able to single out the individual raw signals from each antenna, and the time of origin and then calculate the converged signal at alternative areas of coherence. This is by no means a unique security threat. Of course one could simply employ encryption over their channel for secrecy.

This doesn’t break Shannon’s law?

As stated in their white paper, people incorrectly apply the law to spectrum rather than channel. Even before DIDO, one could use directional wireless from a central base station and achieve 1:1 channel contention (but that’s difficult to achieve practically). DIDO creates “areas of coherency” where all the receiving antenna picks up is a signal only intended for them.

Better than Australia’s NBN plan

I’ve already seen some people attempt to discredit this idea, and I believe they are both ignorant and too proud to give up their beloved NBN. I have maintained the whole time that wireless technology will exceed the NBN  believers interpretation of Shannon’s law. Remember Shannon’s law is about *channel*, not *spectrum*. DIDO is truly the superior option, gigabit speeds with no digging! And clearly a clear warning that governments should never be trusted with making technology decisions. Because DIDO doesn’t have to deal with channel access, the circuitry for the radios is immensely simplified. The bottleneck will likely be the ADC and DACs, of which NEC has 12bit 3.2Giga-sample devices (http://www.physorg.com/news193941421.html). So multi-terabit and beyond is no major problem as we wait for the electronic components to catch up to the potential of wireless!

CRITICISMS UPDATE:

  • One aspect to beware of is the potential need for 1:1 correlation of antennas from the base station and users. I can’t find any literature yet which either confirms or denies such a fixed correlation. But the tests for DIDO used 10 users and 10 antennas.
  • If there must be one antenna per user this idea isn’t as earth shattering as I would hope. However there would still be relevance. 1) It still achieves 100% spectrum reuse, 2) all the while avoiding the pitfalls of centralised directional systems with beam-forming where obstacles are an issue. 3) Not to mention the ability to leverage the ionosphere for rural applications – very enabling.
  • After reading the patent (2007) – I see no mention of the relationship between AP antennas and the number of users. However I did see that there is a practical limit of ~1000 antennas per AP. It should be noted that if this system does require one antenna per user, it would still be very useful as a boost system. That is, everyone has an LTE 4G link and then when downloading a video, get the bulkier data streamed very quickly via DIDO. (The amount of concurrent DIDO connections being limited by the number of AP antennas)
  • The basis for “interference nulling” discussed in 2003 by Agustin et al. http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.10.2535
  • Removed many  ! at the top, to symbolise the potential for disappointment.
  • Hey there’s a spell check!
  • Have a look here for whirlpool discussion: http://forums.whirlpool.net.au/forum-replies.cfm?t=1747566

Co-Operative LTE

Introduction

Australia’s on the way to getting “seperate” LTE networks, by Telstra, Optus and others. Each taking about 20Mhz of spectrum each. Each of these networks will overlap, representing a poor allocation of resources.

Why not build a government or industry body organisation to build a single network which will reach further and be able to provide faster speeds than any of the smaller individuals.

Imagine if a single wholesale network had all 100Mhz of spectrum! You could achieve gigabit speeds and would be very cost effective for customers.

A better picture

Consider the status quo:
+ provider A use 10 towers to cover a town with 95% coverage, 20Mhz of spectrum and up to 300Mbps at a cost of $2bn, and;
+ provider B use 10 towers to cover the same town with 95% coverage a seperate 20Mhz slice and up to 300Mbps at a cost of $2bn, and;
+ provider C use 10 towers to cover the same town with 95% coverage, another 20Mhz of spectrum and up to 300Mbps at a cost of $2bn.

In total you will have 30 towers, 95% coverage, top speeds of 300Mbps and a total cost of $6bn. The customers are the losers here.

If instead a wholesaler uses 15 towers to cover a town with 99% coverage, 60Mhz of spectrum and up to 900Mbps of speed at a cost of $3bn. Then there is:
+ a big saving of $3bn dollars.
+ lower cost for service
+ an improvement in coverage
+ an improvement in speed

A possible business plan

Remember this is only #one way# of acheiving this. Let’s call the wholesale company (WholeMobile):
+ Each of the providers (A-C) would invest $1bn
+ Each of the providers would provide information of their currently owned mobile tower sites and current backhaul arrangements WholeMobile.
+ WholeMobile would pay the appropriate providers to upgrade the best sites to LTE (selecting from sites from all three providers)
+ WholeMobile would receive free access to the LTE services and ownership of the LTE components.
+ The tower property could remain property of the providers.
+ Co-located equipment would remain property of the respective owners.
+ The spectrum would be owned by WholeMobile.
+ WholeMobile would pay the relevent providers to upgrade backhaul to the various sites, with all the additional bandwidth belonging to WholeMobile.
++ Where upgrading backhaul does not suffice, new links will be commissioned by WholeMobile and fully owned by WholeMobile.
+ Profit from WholeMobile is saved for future network upgrades. Any excess profit is returned as a profit to investors.
+ The government may make a small investment to help inspire the process.

Operationally:
+ WholeMobile represents the full $3bn investment, with ownership of the use of the LTE network
+ Only a few sites should be required, where coverage is to be improved
+ Rural customers now can have premier coverage and speed (no need for satellite), in fact coverage can potentially be 99.9999% of the population and roads.
+ All providers and retailers regardless of investment pay for access (No need for ACCC to regulate prices).
+ Example of access business model: Use a [per KB weighted model]. Traffic on a congested tower results in penalty accounting. Eg. 1KB from a maxed out tower accounts to 10KB for the customer instead. This ensures customers don’t try to use all 900Mb of bandwidth all the time.

The approaching revolution

By the way 300Mbps * 11.5 is 3.5Gbps not 3.5Tbps (my mistake). But keep in mind that the digital dividend is just the beginning. Take into consideration the 403-520Mhz spectrum, and existing licensed spectrum.

Digital technology has been exploding in the wireless domain {Digital TV, Digital Radio, 4G etc}. It is obvious that digital encoding is much more efficient than analog. If the whole spectrum was re-assessed and applied with mostly digital technology, most government, military, navel and air services could be greatly condensed, leaving huge chunks of spectrum for commercial purposes. We are only seeing the beginning of a wireless revolution.