Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie

Musings on the future of mobile internet

Options
  • 07-02-2009 9:18pm
    #1
    watty
    Registered Users Posts: 32,417 ✭✭✭✭


    blast05 wrote: »
    Not necessarily - current technologies being developed in R&D labs (next generation ADSL) will allow for downlink of up to 80 Mbps using existing copper wire infrastructure - will allow high definition IPTV to be broadcast ..... so while all the network nodes would need to be upgraded, the nationwide copper wire infrastucture connecting to the home would not so therefore no need for fibre optic networks or any other thing like that. In any case, mobile broadband is where the real future is

    Fantasy I

    We've had 200Mbps on copper phone wires over 5 years ago. The distance is the problem. You can't break the laws of physics.

    IPTV, esp. VOD needs 1:1 backhaul contention and massive servers. Even on fibre or short DSL with 20Mbps it's a poor copy of Cable and hugely expensive for broadcast compared with Satellite.

    Fantasy II
    100Mbps LTE translates to 200kbps loaded at cell edge. The future of mobile is LTE. The future of Broadband is faster fibre. Fibre BB is in most countries already. Slovakia has loads of FTTH.



    RBOS has maybe biggest exposure of banks around here. None of the Irish ones.


Comments

  • Options
    #2
    thebman
    Closed Accounts Posts: 10,012 ✭✭✭✭thebman


    blast05 wrote: »
    Not necessarily - current technologies being developed in R&D labs (next generation ADSL) will allow for downlink of up to 80 Mbps using existing copper wire infrastructure - will allow high definition IPTV to be broadcast ..... so while all the network nodes would need to be upgraded, the nationwide copper wire infrastucture connecting to the home would not so therefore no need for fibre optic networks or any other thing like that. In any case, mobile broadband is where the real future is

    I can't honestly believe someone said this :(

    Just visit the broadband forum and Midband forum please, at least once before posting something like that.


  • Options
    #3
    PogMoThoin
    Closed Accounts Posts: 13,874 ✭✭✭✭PogMoThoin


    blast05 wrote: »
    In any case, mobile broadband is where the real future is

    :rolleyes: Dumbest quote of the day. FTTH is where the future of broadband is, we've not even starting about our new network yet while some other countries are completing theirs


  • Options
    #4
    blast05
    Closed Accounts Posts: 256 ✭✭blast05


    PogMoThoin wrote: »
    :rolleyes: Dumbest quote of the day. FTTH is where the future of broadband is, we've not even starting about our new network yet while some other countries are completing theirs

    Hmm, dumbest quote of the day i see.
    Anyway, i am talking 10 years down the line when we will have hundreds of Mbps from LTE. Already we have HSDPA pushing the boundaries to 42 Mbps - to be demonstrated at Mobile World Conference in Spain tomorrow. So why pay billions for a a complete nationwide FTTH rollout (which certainly won't happen in the next 3 to 5 years anyway) when in 5 years time it will almost certainly be possible to achieve as fast a download rate as you require** if a true nationwide advanced LTE network were rolled out.

    **Note of course Bill Gates once said that he could not see the need for more than 64Kb per second of memory on a home PC so suggesting that an end user would not need more than a couple of hundred Mbps might be a bit placed .... although by the scale that Gates was out !!


  • Options
    #5
    bealtine
    Registered Users Posts: 4,051 ✭✭✭bealtine


    blast05 wrote: »
    Hmm, dumbest quote of the day i see.
    Anyway, i am talking 10 years down the line when we will have hundreds of Mbps from LTE. Already we have HSDPA pushing the boundaries to 42 Mbps - to be demonstrated at Mobile World Conference in Spain tomorrow. So why pay billions for a a complete nationwide FTTH rollout (which certainly won't happen in the next 3 to 5 years anyway) when in 5 years time it will almost certainly be possible to achieve as fast a download rate as you require** if a true nationwide advanced LTE network were rolled out.

    LTE as a delivery platform is wishful thinking. As is HSPA+ etc.
    Fibre is the only realistic way of rolling out decent broadband backhaul.
    Last mile delivery can be achieved in many different ways.

    However no CDMA system will be delivering any time soon


  • Options
    #6
    PogMoThoin
    Closed Accounts Posts: 13,874 ✭✭✭✭PogMoThoin


    blast05 wrote: »
    Hmm, dumbest quote of the day i see.
    Anyway, i am talking 10 years down the line when we will have hundreds of Mbps from LTE. Already we have HSDPA pushing the boundaries to 42 Mbps - to be demonstrated at Mobile World Conference in Spain tomorrow. So why pay billions for a a complete nationwide FTTH rollout (which certainly won't happen in the next 3 to 5 years anyway) when in 5 years time it will almost certainly be possible to achieve as fast a download rate as you require** if a true nationwide advanced LTE network were rolled out.

    Ireland doesn't have any spectrum for LTE, the 2.6GHz spectrum is currently used for TV distribution, and will continue to do so until at least 2014
    blast05 wrote: »
    **Note of course Bill Gates once said that he could not see the need for more than 64Kb per second of memory on a home PC so suggesting that an end user would not need more than a couple of hundred Mbps might be a bit placed .... although by the scale that Gates was out !!

    Bill Gates said we should never need more than 64K of RAM


  • Advertisement
  • Options
    #7
    watty
    Registered Users Posts: 32,417 ✭✭✭✭watty


    200Mbps LTE needs huge spectrum (40MHz +40MHz).
    This SHARED. It's 400kbps typical for over 40% of rural coverage area.

    Typically LTE will be 100Mbps peak sector speed (320MHz + 20MHz) and about 20Mbps average sector throughput, shared among 5 to 50 users, so speeds 400kbps to 4Mbps at peak times on small urband cells. But could be 100kbps on rural cell edge.

    LTE is much better than HSDPA, but not competition for cable, ADSL2+, VDSL or even Fixed Wireless Metro LAST YEAR!

    LTE is a bit like Wimax (including Hype), but designed for Telcos by Telcos and about 5x faster. Gets rid of CDAM on downlink, so al new gear at base and new modem.

    In 10 years time LTE won't deliver what decent xDSL, Cable and premium Fixed Wireless services delivered in 2008. Because it's designed for MOBILE.


  • Options
    #8
    blast05
    Closed Accounts Posts: 256 ✭✭blast05


    Bill Gates said we should never need more than 64K of RAM

    D'oh, yeah. Can't even get my quote right !

    Anyway, 10 years, 20 years, whatever - can anyone even envisage a scenario where download rates on mobile broadband will be sufficient re replace all fixed infrastructure ? A nationwide network of LTE Advanced or even whatever comes after that. LTE Advanced has already shown 5 Gbps on the downlink in a trial by NttDoCoMO in Japan 2 years ago.


  • Options
    #9
    bealtine
    Registered Users Posts: 4,051 ✭✭✭bealtine


    blast05 wrote: »
    D'oh, yeah. Can't even get my quote right !

    Anyway, 10 years, 20 years, whatever - can anyone even envisage a scenario where download rates on mobile broadband will be sufficient re replace all fixed infrastructure ? A nationwide network of LTE Advanced or even whatever comes after that. LTE Advanced has already shown 5 Gbps on the downlink in a trial by NttDoCoMO in Japan 2 years ago.

    Personally I can't ever see it happening, radio spectrum is expensive and the spectrum requirements are enormous. New fibre techniques continuously push out fibre speeds to speeds of over a terabit per second (theoretical).
    No CDMA technology will catch that:)

    The big problem with these mobile speed "trials" is that they are controlled and rather contrived, perhaps only one user was using the cell site at the time of the trial. That is often the case with these "fictional trials".


  • Options
    #10
    Thraktor
    Closed Accounts Posts: 287 ✭✭Thraktor


    blast05 wrote: »
    D'oh, yeah. Can't even get my quote right !

    Anyway, 10 years, 20 years, whatever - can anyone even envisage a scenario where download rates on mobile broadband will be sufficient re replace all fixed infrastructure ? A nationwide network of LTE Advanced or even whatever comes after that. LTE Advanced has already shown 5 Gbps on the downlink in a trial by NttDoCoMO in Japan 2 years ago.

    It won't happen. Basically, fiber-optic and wireless speeds both increase at roughly the same rate (I think they go up by a factor of ten every five years, but don't quote me on that). However, achievable fiber-optic speeds are already far in excess of those available with wireless technologies (to counter your 5Gbps wireless trial, I'll point out that people have achieved over 640Gbps on fiber-optic). Furthermore, wireless signals through air have to pass through considerably more electromagnetic interference than signals traveling along a fiber-optic cable. Shannon's law states that there is a physical limit as to how much information you can convey through a medium, which is inversely proportional to the amount of noise on that medium. Hence, with a medium (air) that has a high amount of noise, eventually physical limits will be hit by wireless technologies and (like xDSL tech now) higher speeds will only be achievable at shorter and shorter distances. Conversely, due to very low line noise, a single fiber-optic cable is estimated to be capable of handling up to 300Tbps, which would be well in excess of the physical limit of any reasonable wireless infrastructure.

    Furthermore, fixed-line technologies provide the added advantages of a much more reliable speed and service quality, and far lower latency than wireless offerings.


  • Options
    #11
    blast05
    Closed Accounts Posts: 256 ✭✭blast05


    Thraktor wrote: »
    It won't happen. Basically, fiber-optic and wireless speeds both increase at roughly the same rate (I think they go up by a factor of ten every five years, but don't quote me on that). However, achievable fiber-optic speeds are already far in excess of those available with wireless technologies (to counter your 5Gbps wireless trial, I'll point out that people have achieved over 640Gbps on fiber-optic). Furthermore, wireless signals through air have to pass through considerably more electromagnetic interference than signals traveling along a fiber-optic cable. Shannon's law states that there is a physical limit as to how much information you can convey through a medium, which is inversely proportional to the amount of noise on that medium. Hence, with a medium (air) that has a high amount of noise, eventually physical limits will be hit by wireless technologies and (like xDSL tech now) higher speeds will only be achievable at shorter and shorter distances. Conversely, due to very low line noise, a single fiber-optic cable is estimated to be capable of handling up to 300Tbps, which would be well in excess of the physical limit of any reasonable wireless infrastructure.

    Furthermore, fixed-line technologies provide the added advantages of a much more reliable speed and service quality, and far lower latency than wireless offerings.

    I think i understand everything ! but it brings me back to the question of how far we actually need to go. Its like who cares that fibre optic has acheived 640 Gbps if there is never going to be an end-user need for that speed, given also that compression algorithms among other things are continuously improving meaning that lesser speeds are requried to transmit a given data size compared to earlier generations.

    Or maybe i am falling into the Bill Gates trap again ! although i see there is a column on the BBC that questions the need for even 100 Mbps . http://www.bbc.co.uk/blogs/technology/2009/02/100mbps_broadband_who_needs_it.html


  • Advertisement
  • Options
    #12
    Thraktor
    Closed Accounts Posts: 287 ✭✭Thraktor


    blast05 wrote: »
    I think i understand everything ! but it brings me back to the question of how far we actually need to go. Its like who cares that fibre optic has acheived 640 Gbps if there is never going to be an end-user need for that speed, given also that compression algorithms among other things are continuously improving meaning that lesser speeds are requried to transmit a given data size compared to earlier generations.

    Or maybe i am falling into the Bill Gates trap again ! although i see there is a column on the BBC that questions the need for even 100 Mbps . http://www.bbc.co.uk/blogs/technology/2009/02/100mbps_broadband_who_needs_it.html

    Shannon, the ever inventive guy that he was, also showed that compression has a physical limit, based on a quality of the data known as entropy (roughly speaking how much the data can be described in terms of patterns). So advancing compression can't be expected to counter the diminishing returns of the underlying technology.

    I also don't really see any immediate use of a 640Gbps internet connection. But, then again, when the first 10Mbps Ethernet standard was created, barely anyone would have made use of the bandwidth. Today, 100Mbps is the bare minimum for a home network, let alone a business, so I'm just going to assume that, so long as there's bandwidth there to be used, we're going to find ways to use it.


  • Options
    #13
    jmcc
    Registered Users Posts: 7,323 ✭✭✭jmcc


    PogMoThoin wrote: »
    Bill Gates said we should never need more than 64K of RAM
    Wasn't it 640K of RAM? Then again he denied saying it. He did have this in his book though:
    "The obvious mathematical breakthrough would be development of an easy way to factor large prime numbers." Surprising that he didn't invest in Eircom with that kind of insight. :)

    Regards...jmcc


  • Options
    #14
    watty
    Registered Users Posts: 32,417 ✭✭✭✭watty


    On wireless we are within a percentage or so of the Shannon limit. For over 5 years you have only been able to get faster wireless by more bandwidth.

    i.e.
    100bps = 20MHz.
    1Gbps = 200MHz.

    If you have MIMO and are very close to a mast then 40MHz/200MBps could have "virtual sectors" of same frequency within the normal single channel per sector. This makes peak SECTOR speed upto 1.6Gbps for a peak user speed of 200MBps and only 8 users, within 150m of mast.

    Mathematically there is nothing left to give more real Mbps/MHz.

    To compare wireless systems you must know
    Mbps / MHz modulation density
    Power/Aerial gain
    Range
    Frequency Band

    For a given power as you increase range the Mbps/MHz has to drop. This is why MOBILE wireless varies over 1/8 or 1/64 speed for one user with distance. They ALWAYS quote the best case and the widest MHz of spectrum used.

    Fixed Wireless uses more power as the user end is mains driven, not on batteries and uses a Fixed Roof aerial giving up to 32dB more gain than a Mobile system.
    This allows a Fixed wireless system to engineered to give the same speed (or maybe two speeds) over the whole cell to everyone. Further away people can be given big 16dBi aerials (UHF/L-Band or 40dBi dish if it's 10GHz LOS wireless) and close up to mast fixed users can be installed with 6dBi flat panels ( UHF/L-Bandor 18dBi panels if it's 10GHz LOS wireless).

    DOCSIS on Wireless is 10 years old or more. That delivers to the WHOLE cell about 5Mbps/MHz maximum using fixed 64QAM. On Cable 256QAM is used.

    OFDM based systems like DVB-t, DAB, Flash-OFDM, WiMax and LTE can only get to about 5 or 10% less at the same QAM due to Symbol Guard time required to remove multipath.

    Mobile system can use up to 256QAM very close to mast (20Mbps/MHz) but drop to QPSK with distatnce (about 1MBps per MHz at best) and to get more range drop modulation clock rate and increase FEC using 5/6ths of the data to correct errors dropping speed to about 0.1Mbps per MHz.

    Thus a 14Mbps peak speed Mobile system may have only 1Mbps average cell throughput if you make the cell too big.



    Non-Wireless
    2048 QAM on cable and fibre has been tested, that's 32x faster or about 150Mbps/MHz.

    Shannon:
    Noise is basically the limit to increasing QAM or more MBps/MHz. Noise makes it impossible for the receiver to decide what symbol has been sent and thus an error occurs. As you increase the Data Rate for given bandwidth of frequency, with a given Transmit power, path loss (range) and background noise (band), you reach the Mathematical Shannon Limit. This is why quieter higher power cables can use higher data rate than Wireless, possibly 100x for Coax. Also wireless channel is unlikely to be more than 20MHz wide. Regular coax cable has nearly 1000MHz wide and fibre potentially 1000,000,MHz wide. This means coax today has upto 1500x to 15,000x the capacity of typical Mobile systems operating at Shannon Limit on 20MHz channel and fibre 1,5000,000x to 15,000,000x the capacity of 20MHz wide Mobile Wireless.

    Note that whatever clever trick you use for Wireless, can 10mins later be put on coax or fibre and in neither case can you break the Shannon limit, which we have already reached.

    Newer wireless systems are about using the spectrum with same efficiency when 100 people rather than one is using it. CDMA (3G/HSDPA) is only at full efficiency with a single user. Also the newer systems can use wider channels.
    GSM = 200kHz per channel
    CDMA-One / EVDO A = 1.25MHz
    Flash-OFDMA = 1.25MHz
    IPW, FOMA = 4MHz to 5MHz
    W-CDMA (3G/HSDPA/UMTS) = 5MHz
    LTE = 5MHz, 10MHz, 20MHz
    WiMax = 5MHz, 10MHz

    MIMO does not give any real speed increase. It is spacial reuse of the same spectrum, it's very limited in range.

    At 10GHz, or better still 22GHz you can get nearly perfect reuse of same spectrum and double capacity by using two polarisations. It's not feasible with more than about 100m range at 900MHz to 2.1GHz as the signals change with air layers, terrain etc.

    Mobile "Broadband" is NEVER EVER going to deliver real broadband unless every lamp-post and pole in Ireland is a base station mast.

    Fixed Wireless can for low density Rural locations deliver real broadband at the bottom end of the scale (3Mbps to 10MBps per user).

    Real 100Mbps Broadband isn't in the future, the deliver systems are mature. It's just we haven't got much of it
    FTTH
    FTTC with VDSL/ADSL2+ on less than 900m
    HFC with DOCSIS3 and 900MHz capacity coax

    All deliver real 100Mbps per user today at low or no contention. Mobile wireless can't ever do this unless every cell is less than 100m radius


  • Options
    #15
    crawler
    Registered Users Posts: 1,660 ✭✭✭crawler


    Can we also all remember the very scientific rule of telco i.e. "things always get worse"

    More distance - things get worse
    More users - things gets worse
    More noise - things get worse
    More modulation - things get worse
    More gain - things get worse
    More interference - things get worse

    I think you will get the picture :D


  • Options
    #16
    watty
    Registered Users Posts: 32,417 ✭✭✭✭watty


    A good executive summary of my technical ramblings. Thanks Crawler :)


  • Options
    #17
    Thraktor
    Closed Accounts Posts: 287 ✭✭Thraktor


    Thanks for the technical info, Watty, it's very informative.
    jmcc wrote: »
    Wasn't it 640K of RAM? Then again he denied saying it.

    64K would have made sense, as that's the largest amount of memory that can be directly addressed by a 16bit processor.


  • Options
    #18
    watty
    Registered Users Posts: 32,417 ✭✭✭✭watty


    Going very off topic, shoot me.

    8088 and 8086 were the so called 16 bit processors. They can address 1Mbyte. However the Original PC and MSDOS right up to version 7 was never a true 16 bit OS, but a segmented hybrid 8/16 bit OS. 640K was set as limit of RAM and most I/O memory mapped, including graphics RAM between the 640K and 1024K

    However addressing range is not directly related to CPU main databus width or majority of registers, but to details of CPU design. Most 8Bit CPUs have a 16 bit address bus and thus can address 64K. Some can access 1Mbyte. The 8086 16bit data bus CPU has 20 bit address bus, but is really an extension of the 8085 to make porting 8bit CP/M software easy. So you are sort of correct, as it could only access 64k segements at a time.

    The 80286 unlike the 8088 and 8086 is a "real" 16bit CPU. having a 24-bit address bus, the 286 was able to address up to 16 MB of RAM, in contrast to 1 MB that the 8086 could directly access.

    Sadly only a few versions of UNIX at the time took advantage of the linear addressing, DOS indeed used a very broken segementated 16bit 1MByte via 64k segements.

    Segmentation is why MSDOS and 8088/8086 is really and extended form of CP/M and the 8bit 8080/8085/Z80 with 64K.

    I believe he said 640K and given the stupid architecture and need to leave space for Memory Mapped I/O and graphics, 64k Segmentation and 1024Kbyte limit the stupid design was a reasonable compromise. IBM should never have used the 8088/8086. There were cost effective real 16bit CPUs in 1981. The CPU choice, not the 640k of itself held back desktop computing for nearly 15 years (1996 with release of NT4.0 was when buisness started turning from Win9x on DOS 8/16/32 kludge platform to a true OS, though UNIX was on 286 from about 1985 and NT3.1 released in 1993).


  • Options
    #19
    crawler
    Registered Users Posts: 1,660 ✭✭✭crawler


    This thread has "gone like SO off topic" :)


  • Options
    #20
    watty
    Registered Users Posts: 32,417 ✭✭✭✭watty


    I used to think finance was boring, so the whole Babcock&Brown thing would be so like... BORING...

    But economics & Finance and eircom etc is starting to look like a Bruce Willis film. The kind where with nothing but a dirty white T-shirt he saves the day.

    I never thought I'd be playing a walk on bit-part in one of the films. But really this thread had never much to do with IoffL.


  • Options
    #21
    watty
    Registered Users Posts: 32,417 ✭✭✭✭watty


    blast05 wrote: »
    Already we have HSDPA pushing the boundaries to 42 Mbps - to be demonstrated at Mobile World Conference in Spain tomorrow.

    Note this is a gimmick. It uses TWO whole 5MHz channels. Each at 21Mbps. It's still good ole W-CDMA, but at higher QAM, so only closer to mast. Maybe 150m in real terrain.

    So yes. If a Mobile Operator is stupidly generous enough to dedicate 2/3rds of a mast to single household, you can have 42Mbps.

    In real life it makes <3% difference to rural cell capacity and <15% to urban capacity and only 42 peak instead of 21 if an operator can deploy twice the spectrum
    It's virtually a con trick to get Operators to upgrade on HSDPA spectrum now before spending money all over again on LTE.


  • Advertisement
Advertisement