( Register Here | Lost Password ) Search Today's Posts FAQ
Page 1 of 2 12 LastLast
Results 1 to 10 of 16
  1. #1
    Join Date
    Oct 2010
    Location
    Hampshire/Surrey bdr
    Posts
    319
    PM
    Send Message

    Default Local broadband project - tech specs

    I'm at the final stages of summarising our local campaign for a broadband network for our area and want to compare the relative performance of three things: Wi-Fi, Fibre to the Cabinet and FTTP.

    Not much to say about the latter, that's the preferred option, but I'd like to know which of the other two will be better for the majority.

    FTTC

    Two cabs both on the edges of the village. Fibre ducting is already in place to the village (previous separate project).

    Speed depends on D-side lengths and qualities.

    D-sides ranging from maybe 50m in the kms. The 50m line might expect to see the top speed, but I can have a go at working out what each house will get or taking some point averages.

    However while I can work out the line lengths, I have no idea of the metal or the gauge. As an indicator our 3680m long line managed an IP profile of 1750 with attenuation in the high 50s/low 60s when we had it, and a noise margin of 15 or 18 I think, it's so long ago.

    I have stats for quite a few of the lines (some are hilarious, they're 0.5Meg down). Any way of spotting lines which are obviously decrepit/aliminium? Is it the high attenuation for the length that indicates it, or the noise margin? (Line performance round here swings wildly even on similar length lines)

    As, without knowing what lines are what, if I don't have stats to try and work it out, the only thing I can see to do is to calculate the speeds based on (0.5mm?) copper and then just downgrade them by an abitrary estimate. I just don't know what that estimate ought to be. I'd also be making the assumption that the D-side is aliminium too, rather than the E-side (which would be bypassed and is therefore not relevant) or a mix of the two. It can only be a best guess.

    Wi-Fi

    Church off centre of village, but at one of the highest points and near the fibre ducting. Greatest distance for signal to travel - perhaps 7km. No obstructions at all on one side, a hill on the other so might need a repeater, but might not.

    --

    To my questions...

    FTTC: Can anyone point me at likely speeds versus distance for the newer profile? I know it's only an estimate, I've seen 600m lines do 20Meg down and others do 40. I need for both aliminium and copper. Am I right in thinking that most lines which are copper, will be 0.5mm copper as they're all very old? Plus, most charts I've seen top out at around 900m line length (convenient). I need a chart that goes up to say 5km.

    WiFi: What's the speed versus distance computation? Any other factors? Size of antenna? Size of cell?

    Finally, what sort of backhaul would be needed for perhaps 100 users (not simultaneous) aiming to deliver a true 25Mbps+ downstream, what is the "normal" contention ratio to work to?

    That's the key question - I want to know if either of these technologies gets us anywhere near that, so if the FTTP option doesn't come to bear, we know which of the two alternatives will get the most speeds to the most people.

    The other factor is that if we went for any option other that FTTC I'm sure BT will pop along to fibre the cabs the next day, so I want to know what speeds that will achieve by comparison.

    Thanks

  2. #2
    Join Date
    Oct 1999
    Location
    London
    Posts
    8,530
    PM
    Send Message

    Default

    I saw some quite speculative estimates for FTTC on TB and Original Broadband recently.

    http://www.thinkbroadband.com/news/5...tc-run-at.html

    http://www.origin-broadband.co.uk/ou...es/origin-max/

    If you ask me the bottom half of TB's figures should be about 15% less (from experience with three FTTC lines) but this is variable and in some cases it could be 5% or so higher; there's a wide margin of error. At least with ADSL2+ you could get your bearings through attenuation, in fact you probably still can but only if you compared before-and-after between ADSL2+ and FTTC (we still don't have enough data to make this work accurately in a prediction).

    WiFi is even more tedious because it depends upon so many different factors, not least your hardware (antennas, routers, transmitters etc.) and the environment. I'm afraid the only real way to be sure is to run a trial or talk with an existing wireless ISP (for commercial reasons they may or may not want to share their experiences with you). A proper fixed wifi network is quite different from a home solution.

    As for contention, well that really depends upon your budget and some basic math . A "true" 25Mbps service would need 25Mbps on a 1:1 contention but that's dedicated business grade stuff. I've seen quite a few wireless ISPs offering home solutions with something close to 20:1 or 30:1 at the upper end, though others have made it work with lower contentions than that.

    To some degree this will be dictated by the usage in your particular area and you will need to conduct a trial first. Remember on a new service you should be cautious with the investment and spend later to expand. Don't put all the cash into max capacity right away, try to adapt organically.
    Mark 'Winter' Jackson
    Editor-in-Chief - ISPreview.co.uk

  3. #3
    Join Date
    Oct 2010
    Location
    Hampshire/Surrey bdr
    Posts
    319
    PM
    Send Message

    Default

    Thanks Mark.

    The Digital Region project seems to have trouble (as reported on here) estimating speeds for VDSL based services (believe the quote was that some subscribers get only half the speed they were promised).

    This isn't surprising based on what I'm seeing here. I'm trying to get to a better than "the performance will be almost completely random" estimate.

    This partly since something East Hants quoted at me intrigued me - that if we had FTTC to our cab, we'd see 31Mbps down. That's with a D-side of 1180m on a line which only manages 42% of its theoretical ADSL1 sync.

    I've taken ten sample points which are all reasonably central so these aren't the longest D-sides. I'm restricted as I don't have survey results for everyone so have used what I have to pick some differing sample points. And the data makes little sense.

    Three out of the ten lines perform at between 97% and 102% of their theoretical ADSL1 sync e.g. if it should be 4Mbps it syncs at about 4Mbps (using figures from Kitz calculator)

    Those three are not near each other or even on the same cab.

    The other lines perform at between 16% and 71% of their theoretical ADSL sync with one exception which actually manages over 100% which makes me wonder if some lines to some cabs were 1mm when the cab is further away/stock was available/the engineer was in a good mood/the D or E side was so decrepit it got replaced with a brand new one

    The one which does over 100% of its theoretical max sync, is only 250m from the one which manages 20% of its max sync.

    Noise margins are all over the place, only one is 6 and most are 12 or higher.

    The shortest D-side is 50m and the average from my sample (bearing in mind this is mostly the central properties) is 713m, the average actual ADSL sync is 3.4Mbps (2.7Mbps throughput) and the average total line length is 3400m (if you reverse calculate the line length based on the attenuation, this rises to 3800m, I don't think my estimates are far out)

    So, based on all that, making the extrapolation that a line which is performing appallingly now won't be that great for VDSL (Aliminium? dodgy oxidised joints and repair boxes on D-sides? Can't imagine it's severe crosstalk!) and if every fibre cab were enabled: to start, the average speed rises from 2.7Mbps (now, actual) to 33.7Mbps (optimal estimated sync).

    If you then downgrade that performance by the same as ADSL per line, the average speed would be 21Mbps with a range of speeds in the sample from 49.5Mbps down to 1.6Mbps (the line that's so appalling it probably wouldn't hold VDSL if the poor quality is on the 1650m D-side). Ours, endearingly, would rise from 1.7Mbps ADSL1 to be 8.1Mbps VDSL. Perhaps 5 properties would actually be in line for something close to the 80Mbps top speed down.

    30% of the lines would be "superfast broadband" (25Mbps+ down or better) if *both* cabs were done. One of them only feeds about 50 properties from what I can see. It's also the one with the one property which would see the fastest top speed (the pub, oddly enough) but where most of the others are over 300m away (and also business premises).

    But this does not include any of the more remote areas of the village - as mentioned these are all pretty central. So the actual average is likely to be a bit lower than that.

    This little bit of research doesn't surprise me at all, having lived exclusively in houses with dodgy or long phone lines in my past :-)

    If you take a trip over to our neighbouring village, FTTC would provide 50Mbps down to something like 75% of the village since it's clustered and so one cab in the right place would be a pretty good option.

    This is all of course estimates, but then I don't really think BT have got a clue either and the knowledge about line quality seems to live with the engineer(s) responsible for that "patch", coming back to the Digital Region project it's only when a line is enabled for a rate adaptive DSL service that anyone can have any idea what it will do, partly because of the mixed metal and guages in the network.

    So if FTTC can bring superfast broadband to a third of the village, and that involves hassle with SLU to get the fibre running down the phone lines as a distribution mechanism (this means running fibre "out" to the cabs, then the connectivity comes "back in" to the houses) I'm thinking Wi-Fi is a serious contender to match or better the performance.

    And then I see local Wi-Fi projects offering speeds of up to 4Mbps or 6Mbps and wonder. Perhaps that's your point - start small, grow later - but then it's the same single fibre pipe to the mast and the cost of more backhaul is not linear relative to the performance.

    I think it can't be that hard to get 25Mbps+ to every home with line of sight (and possibly some good upstream performance, too) via Wi-Fi, but it doesn't seem to be done.

    We've approached a FTTP company and I'll now go and approach all the Wi-Fi companies who mailed me in response to the project.
    Last edited by DTMark; 29-04-2012 at 11:25 AM.

  4. #4
    Join Date
    Oct 1999
    Location
    London
    Posts
    8,530
    PM
    Send Message

    Default

    Long post of the year award there DTMark but I did read it all .

    If you have a good line-of-sight then that's a huge plus in WiFi's favour I'd say because LoS is a big issue for most places. Sadly the only WiFi networks I've ever run or built have been home/office style interior ones that are a different ball game to exterior services, which require special antennas on the roof and transmitters (some used for boosting) etc.
    Mark 'Winter' Jackson
    Editor-in-Chief - ISPreview.co.uk

  5. #5
    Join Date
    Mar 2011
    Location
    Essex
    Posts
    16
    PM
    Send Message

    Default

    As Admin point out the chief handicap is the absence of performance stats. So, in case it is helpful I will add my experience as both a FTTC (VDSL) and Fixed Wireless user.

    I live in a village served by a major suburban exchange, some 3.4km from the exchange and 600m by estate road from the cab. There are few dwellings between the exchange and the village, but the underground line through the estate, built in the late 1960s, has repair joints at almost every dwelling. On ADSL2+ the maximum speed sync locally from any provider was around 4Mbps, and throughputs up to 3.5Mbps, just over half of what we could expect on a good line with my attenuation reported at 47, and SNR up to 10.

    When my speeds declined to sub 2Mbps on LLU (DSLAM problems I suspect, it was Orange) I resorted to Fixed Wireless through
    an "averagish" local WISP. For comparison I measure my stats using JD's tester, which runs automatically set to a 5hr 40m frequency, so I get a range of stats through the day. The WISP service is based on a 25Mbps symmetrical service (12.5 each way) - you should bear in mind that although wireless can provide speeds up to and in excess of 100 Mbps these are normally higher price services than the base service advertised (more than double the cost, usually). I connect direct to their main transmitter (not through a relay) which might compensate a bit for the distance, which is served by a dedicated Easynet backhaul connection (over the Sky network).

    Over 800 tests, the Fixed Wireless (I am 5km or so from the transmitter, though with good line of sight across a river valley) manages at a peak nigh on 12Mbps down and over 10Mbps up throughput measured by the speedtester, the average is 9 down/6 up, the "normal" lowest being about 7/5, usually during daytime when congestion I suspect is at its most noticeable (due to business users). Speed is not noticeably affected by weather, and some of the highest speeds are during inclement winter weather - though I did loose connection for the first weather-related time during a recent electrical storm (when the VDSL2 also reported a doubling of its highest error rate). So the Wireless normally and reliably manages more than 75% of its maximum download speed (50% upload) on my connection.

    I had little hope for VDSL, but in practice sync is the maximum 38.71 on the 40 cap, and throughputs are consistent in the range 36-38Mbps, mostly at the higher end of the range, since January this year, now that I am using a non-BT modem to gain access to stats. The maximum attainable sync (if my service were not capped) is reported at just over 40 Mbps, and attenuation "high" at 20, SNR as low as it will normally go at 6/7. I suspect that because of the poor quality line power levels may be reduced at the cab (which I what I believe affects achievable speeds more than anything - to reduce interference with ADSL services) and may be shown by the fact that download tones disappear almost completely temporarily at around 1200kHz, and I get no benefit from the higher frequencies above 8500kHz introduced by the 17a profile (so no third "band"). This looks to be normal for those on poorer/longer lines. The modem also reports that the maximum attainable upload speed is limited at 7.2Mbps, although the IP profile is 10Mbps. The JDast speedtester reports are 36-38Mbps download and 4.8-5.4Mbps upload, consistently with no time of day variation, the BT speedtester is consistent though a bit higher on the upload, up to 6Mbps. The modem reports syncs consistently at 39992/6812 - there are no comparable stats from the cable modem on the Fixed Wireless service. So with my poor history on ADSL, and I believe you could add effectively up to 20% on the cab distance for the effect of the poor line, I still achieve a max sync which is around 80% of the expected best projected on line distance (40+/48), though the upload suffers more (due to a greater effect on the tones, up to a 60% loss if you take the profile - though some with better lines have greater loss). There is no noticeable effect of congestion on the local loop.

    Latency on both are comparable at best at around 16ms, though the FTTC never varies (8ms to the cab) but the Fixed Wireless service can rise to 50ms and higher, though not normally over 100ms. There is some packet loss and jitter on the Fixed Wireless, effectively none on the VDSL2. The noticeable difference between the two services is the twice a month average outages on the BT Wholesale backhaul for VDSL2, usually overnight for maintenance, with hardly any problems over the Easynet backhaul - where latency is normally only a few ms checked over the TB tester.

    Hope this rambling gibberish helps . . .possibly.

    PS The other thing that occurs to me is the normal lack of competition on Fixed Wireless, particularly from the more competitive ISPs: BT retail, Talk Talk and Sky as they introduce value-added services to the consumer. These matter to quite a few people.
    Last edited by dragoneast; 30-04-2012 at 12:29 PM. Reason: add comment on competition

  6. #6
    Join Date
    Oct 2010
    Location
    Hampshire/Surrey bdr
    Posts
    319
    PM
    Send Message

    Default

    Was a bit long, wasn't it :-)

    I was trying to summarise a table of results. The upshot, in brief, is that nobody really knows how FTTC would perform; I'd be prepared to take the "typical VDSL estimates" for the three lines which "look right" from an ADSL perspective e.g. they perform as they ought to. 30% of the lines are "spot on".

    For the others, there could be a multitude of reasons why they're so poor, but in each case I'd be very cynical about any estimate.

    So the average village speed with FTTC would be somewhere between about 20 and 35 meg which sounds very good - impressive for a rural area. But, it's headlines. It masks enormous variations from nearly 80Meg at the top, to practically nothing at the bottom end (e.g. it won't work, or if it did it would be very poor) and I'd hate to see one of the project champions eagerly order their new fibre connection to find it runs at 5 meg or not at all, and then their ISP choice is restricted to those who just don't go for the low-hanging fruit. We can only spend what we have once.

    As I sit here looking out of the window I can see the high points of the village and if you walk up the hill behind it, you can see very nearly everything in one go. It's such a small area (think of an I shape with the phone cabs at the left edges of the symbol, which is why they're not so performant - cabs are away from the bulk of the properties) that it must, surely, be possible to deliver 25Mbps+ down to everyone over Wi-Fi with almost everyone having good line of sight. He says.

    In the longer term Wi-Fi could/would become FTTP so there's an upgrade path and fibre already here, so if the maximum a cell can offer out is about 300Meg (which I'd read) then that can't provide the 25Meg to more than 12 people at once if my maths is right. FTTC could provide 25Meg to everyone who could get 25Meg, but that's no use if you're one of those who wants 25Meg and has a naff phone line that can't do 25Meg.

    There are only 240 premises. Based on a 30% take up that's 80 lots of 25Mbps needed which is 2Gbps based on 1:1 and the Wi-Fi is then the bottleneck (the fibre can deliver the 2Gbps fine) but then it wouldn't be a 1:1 service. If someone wanted something really quick, perhaps they could have a fixed line spur from the cell.

    The network you're decscribing above sounds like one where contention is the issue if it really can deliver the 12Meg at least some of the time (like 3G here - can theoretically deliver 7.2Meg but rarely punches above 6.1Meg with 3/5 bar strength and has been down as slow as 4.3Meg lately at times). Wonder why the ceiling to the service (25Meg symmetrical?) Perhaps the contention? Relative to size of area covered? We're only looking at a tiny area.

    Latency doesn't sound great on wireless, though. I'm guessing that's a factor of signal strength and possibly contention?

    The value-added service people seem to want is TV, but that's still in its infancy, I don't think we're going to get cable anytime soon and there's no easy "add on platform" or "package" that compares with cable (interaction, choice, VOD, it's just installed for you and it just works etc) which is what I suspect people would like. The broadband facilitates getting it when it exists though I suppose. But there's something nagging at me as regards streaming HDTV with Wi-Fi...

    There will be a mix from people who will use a mobile and get shot of their landlines pretty quickly (who admit they barely use them anyway and want a broadband service, and who can't wait to get shot of BT) to those who will wait in perpetuity for their BT PSTN line to become useful one day and who won't be dragged away.

    Survey results indicate people *are* prepared to pay a bit more for a decent service but then since they're the ones interested enough to respond thus far, that's perhaps not surprising or representative.

    600m by road and 40Meg max sounds about right for FTTC e.g. the D-side performs as it ought to so all the dodgy wiring looks like it's been skipped over. To be fair, the quality issues could be with the E-sides round here, it's impossible to know though I'd presumed the D-sides get damaged/replaced more often since they're partly overhead.

    One other aspect is DACS boxes. A number of the poles have DACS boxes (like our one - not sure how many, have seen others in the village so there is, or has been, a pair shortfall in the past) so there's no spare PSTN E-side. Does this prevent VDSL being installed e.g. can you have a DACS box in a VDSL circuit?

  7. #7
    Join Date
    Mar 2011
    Location
    Essex
    Posts
    16
    PM
    Send Message

    Default

    Just to say I could get 50 or 100 meg wireless if I was prepared to pay for it: 50 is 24 pm here double the price of the 25 Meg (both symmetrical but that's not necessary).

  8. #8
    Join Date
    Oct 2010
    Location
    Hampshire/Surrey bdr
    Posts
    319
    PM
    Send Message

    Default

    I wonder if that's related to the contention - the best I could manage based on some assumptions about take-up was in the region of 30 per month for a shared 2Gbps backhaul distributed via Wi-Fi based around 25Meg down per user, plus the cost of the antenna.

    Kijoma's offer of 30/10 is 80+VAT per month. I'd pay it, seems perfectly reasonable to me, but not sure how many would.

    http://www.kijoma.net/tiki-view_articles.php

    Surely, though, with Wi-Fi especially when over larger areas the operator has to be confident about the vagaries of contention in order to be able to have a split offering where you can pay more to get more. It implies there is "more" available. Different frequency set?

    Shame their network ends about 12 miles away from us. At the moment.

  9. #9
    Join Date
    Oct 1999
    Location
    London
    Posts
    8,530
    PM
    Send Message

    Default

    What kind of contention is that 30/10 offer? Does it factor in usage costs?
    Mark 'Winter' Jackson
    Editor-in-Chief - ISPreview.co.uk

  10. #10
    Join Date
    Oct 2010
    Location
    Hampshire/Surrey bdr
    Posts
    319
    PM
    Send Message

    Default

    Can't see anything about contention levels on the site. The "up to" explanation page reads like a shorter version of my posts above :-)

    "This graph below represents a typical ADSL2 sync speed versus the length of the cable back to your exchange. It assumes your cable is copper , not aluminium and there isn't a load of joins and discontinuities in the line. It also discounts any effects of interference and crosstalk."

    http://www.kijoma.net/tiki-index.php?page=up+to&bl

    From the same page:

    Contention on our network

    As with all networks there is a finite capacity feed to each access node on our system, this is always being increased as we ride the technology wave but it will always have a top limit as with ALL systems. Fortunately we do have Quality of Service management on our networks (QoS) and have had since day one. Way before most other ISP's had considered it worthwhile.

    QoS is not "throttling" as some may think it means, its balancing the load and is a beneficial feature. Could you imagine what the road system would be like without traffic lights, flow control and other "balancing" methods? . it would turn into gridlock and chaos.. Some would get a fast speed, others maybe stationary.. A completely "unfair" situation.

    This is not to say of course that some ISP's may use it to throttle the speed of high users, or to put them on a "naughty" profile which has low priority.. We do not do this. Which is why we may not have as high a usage limit as many offer but unlike many of those, we do not them throttle you via stealth instead.


    Service priority level

    The different Kijoma services have different priority profiles. We believe that if you pay more for one of our business tariffs then you deserve a higher priority than say our home Light service. As such each tariff has a different overall priority on the QoS system. This is fair as a 14 home service should not hinder a business 27 one if a contention issue was to exist. Likewise for Business plus vs any lower tariff etc..

Page 1 of 2 12 LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  

Copyright © 1999 to Present - ISPreview.co.uk - All Rights Reserved (Terms, Privacy Policy, Links (.), Live Chat & Website Rules).