Irish Rural Link and the NBS

bealtine

Irish Rural Link is a non-profit organisation, which represents rural community groups and associations at a national and international level.

They have done an excellent analysis of the NBS. You can find it here:

http://www.irishrurallink.ie/publications/IRL%20analysis%20of%20NBS.pdf

The comments from the DECNR are hilarious, worthy of an episode of "Yes Minister".

Has cracking comments like :
DCENR’s Comments on Quality:
The Department is satisfied that the NBS will provide speeds comparable with products
available in the market for urban area and greatly assist in bridging the digital divide.
Businesses currently without any broadband service will benefit by being able to send and
receive emails and files, surf the web, and update their own websites. Under the NBS
contract the broadband products will be upgraded to higher specifications (speeds,
contention and data caps) in July 2010 and again in October 2012 without any increase in
the monthly recurring charge.

also

The speeds and quality of broadband provided under the scheme allow individuals to
telework, bank and shop online. Mobile broadband is used by many organisations today,
particularly those with mobile workforces.

Have they ever tried to use it?

Sponge Bob

bealtine wrote: »

The comments from the DECNR are hilarious, worthy of an episode of "Yes Minister".

Has cracking comments like :
DCENR’s Comments on Quality:
The Department is satisfied that the NBS will provide speeds comparable with products
available in the market for urban area and greatly assist in bridging the digital divide.

The comments are true in a sense that Sir Humphrey would admire.

1. 3g Broadband will be ****e in rural areas as it has been in urban areas ...and therefore "comparable" .

2. Ripwave is available in urban areas and is not 3g . It is , however, ****e too.

The NBS is about spreading the pain !

cowboy1981

bealtine wrote: »

They have done an excellent analysis of the NBS. You can find it here:

http://www.irishrurallink.ie/publications/IRL%20analysis%20of%20NBS.pdf

Very interesting document Bealtine, although seriously scary how the DCENR appear to have fallen for a bunch of lies in awarding this contract.

"DCENR’s Comments:
The network supports VoIP - 3 offers a VoIP service on its existing network using Skype and further information available at http://www.3ireland.ie/on3/skype-faq.htm"

This reference to Skype as "VOIP" is complete rubbish. The existing Skype service on 3, referred to on this link, is based on mobile phones, not dongles. It uses a standard voice channel to set up a call to a gateway in the UK. This is not over IP at all - it is a normal voice call. At the gateway, it is converted to Skype. The 3 midband dongle is not capable of supporting this as it isn't a phone and doesn't use voice channels.

Has anyone succeeded in getting any kind of VOIP to work reliably on 3's existing dongles?

Is the NBS going to give free mobile phones to everyone in the NBS areas so that they can use 3's Skype service? They have been very explicit that the NBS service (including VOIP) is only going to cost 19.99 per month. Does this include a free mobile phone for Skype?

It would be interesting to make a list of other services that don't work on 3 Dongles. On boards.ie, many users complain that they share an IP address with all other users. Presumably this mean that VPNs, fixed IP addresses and other services do not work? It also means that if a single user gets blocked for illegal spamming, music download etc, all 3 subscribers will also lose service.

watty

cowboy1981 wrote: »

"DCENR’s Comments:
The network supports VoIP - 3 offers a VoIP service on its existing network using Skype and further information available at http://www.3ireland.ie/on3/skype-faq.htm"

This reference to Skype as "VOIP" is complete rubbish. The existing Skype service on 3, referred to on this link, is based on mobile phones, not dongles. It uses a standard voice channel to set up a call to a gateway in the UK. This is not over IP at all - it is a normal voice call. At the gateway, it is converted to Skype. The 3 midband dongle is not capable of supporting this as it isn't a phone and doesn't use voice channels.

Has anyone succeeded in getting any kind of VOIP to work reliably on 3's existing dongles?

Yes. Actually it's a sensible way to connect to PC Skype users, but it's not VOIP. iSkoot is a similar idea. iSkoot works on phones with no SIP as it uses data to connect the call and then a regular 3G voice call for the talking.

Fring uses built in SIP VOIP on your phone and a proxy to do a true data VOIP connection to Skype as there is NO actual Skype Client for any S60/Phone.

VOIP and 3G
Forgetting 3 and "I-HSPA" which is really still 3G W-CDMA transmission, but slightly better environment for IP, if a mast has 50Mbps zero packet loss 1:1 contention backhaul.

VOIP comes in two flavours basically, SIP based and Skype:
SIP
SIP based can use a G.729 codec, but while that can work at a low bitrate, it doesn't work if there is excessive jitter or packet loss. It's poor even on many DSL circuits. G.711 is an uncompressed codec so copes better with jitter and packet loss. However it needs nearly 100kbps upload and download. Very often that is not available consistently on the upload.

3G disconnects and reconnects quite often. SIP based VOIP will loose the session and you have to "redial".

On a mast sector with < 5 users and a good signal, SIP using G.711 will work, though delay can be excessive at up to 0.5 second. Echo cancellation may not work. Typically a sector could support up to five VOIP sessions if the signals are good and there is no other traffic. About 70% plus of rural cell users would have too poor a signal.

Skype
Skype is an adaptive codec. It can even work on an analogue dialup! But Dialup the speed does not change during a session. At higher compressions (lower speed) Skype copes badly with packet loss.

It does not cope well with jitter. Mobile is Bursty traffic and dramatic changes of speed from 3Mbps to 50Kbps during a session.
Skype works very much worse than SIP with G.711, even though Skype can run at a lower data rate as it is NOT designed for a mobile system.
Skype is peer to peer. It will lose the call if the session is dropped.

Mobile Characteristics that Dialup and real Broadband (even Fixed Wireless) doesn't have, that hurts VOIP:

1. Excessive jitter: latency can vary from 100ms to 2000ms during a session, and vary by 300ms within a second
2. Packet loss: Compressing codec that need less upload speed are poor at coping with packet loss. On G.711 it causes silence.
3. Speed Variation: Speed can rapidly change from 5Mbps to 50Kbps. Adaptive codecs can't respond to speed changes quickly enough and many use the speed at the call setup time
4. Frequent random dropping of session

.

Unsurprisingly given the fact that it has fixed speed, no packet loss, no jitter and doesn't drop sessions, ISDN supports VOIP perfectly. In fact G.711 is the native codec for 64k ISDN voice calls

What does I-HSPA improve?

• Minimum latency improved to 80ms from 120ms
• Speed is higher but only very close to mast. An Urban feature for small cells
• Upload is improved to HSUPA, much better upload than HSDPA.

What does I-HSPA not change?

• Speed for most of the cell beyond the 7.2Mbps band, Cell edge same 50kbps loaded.
• Dramatic bursty changes in speed
• Huge amount of jitter.
• Highest latency still near 2000ms
• Packet loss unchanged
• Capacity for VOIP calls beyond 200m from mast.
• Random dropping of session loses call.

Conclusion
VOIP using SIP or Skype is possible on a 3G dongle. I-HSPA doesn't help much as it only is a mechanism to add 20% to 25% capacity in a small Urban cell. The inherent HSUPA feature of I-HSPA/HSPA+ does help G.711 VOIP compared with basic HSDPA in a few cases.

Generally the experience will be poor, or unusable. Tests with true SIP phones using SIP directly and Fring over SIP to access Skype broadband users on PC where similar to fixed broadband when using WiFi and erratic using a 3G connection.

VOIP with QOS on LLU, Fibre, Metro or Cable can be good enough for faxing and not distinguishable from a regular phone call. Third party VOIP on Bitstream DSL often of GSM quality and depends on the interleaving of the Bitstream.

VOIP on a 3G modem on Laptop varies between poor to unusable. Capacity is limited to 2,000 calls roughly if there is NO other data traffic for the whole NBS network supposed to serve 223,000 buildings. If on average four users per sector are active then 500 VOIP calls (which are typically poorer than GSM) for the whole network is possible.

The likelihood of dropping a call at peak times is very high. VOIP support in I-HSPA is poor to non-existent, especially for a regular modem/dongle.

RedLedbetter

The IRL document is fantastic - we've actually got clear hard proof on the DCENRs perception and expectations of the solution Three are implementing on their behalf.

At least we know there can be no case of the DCENR coming out with "we weren't aware", "we didn't read the solution document" etc. It's all out there now. The question is, have the DCENR internally failed to correctly evaluate the solution presented by Three, or did Three exaggerate the potential of the solution when presented to the DCENR - in either case (without genuinely trying to appoint blame) how was the final solution evaluated within the DCENR and how did they come to the conclusion that the solution was appropriate? **Note to self - "appropriate" isn't that what ComReg currently call a funtional dial-up connection?**

If Min. Ryan is as progressive as we hope he appears, then it can never be too late to re-evalute, before this ship has sailed. Ideally, what I'd love to see now is a round table forum between Three, IoffL, the DCENR (solution architects) and Min. Ryan. A pipe dream I know, but I'd be gladly be willing to settle for a face to face between IoffL and the minister/DCENR, not to chastise anyone, just to deal with the facts and ensure everyone is completely aware of the repercussions. I just hope the DCENR are man enough and willing to utilise the resources being offered to them.

Sponge Bob

robertno wrote: »

Have the DCENR internally failed to correctly evaluate the solution presented by Three, or did Three exaggerate the potential of the solution when presented to the DCENR

Both . The DCENR are incapable of any form of technical analysis and therefore 3 were able to pull the wool over their eyes

bealtine

Sponge Bob wrote: »

Both . The DCENR are incapable of any form of technical analysis and therefore 3 were able to pull the wool over their eyes

Well the glossy brochures looked nice and the promise of being "able to update IHSPA to LTE" in a short time was alluring, a bit like a nice girl in a bikini selling printer cartridges. Of course the technical issues were totally overlooked (too busy watching the bikini?)

watty

If you get new modems, x8 amount of new spectrum, about 1500 more basestations and swap all the RF gear in the existing ones you can "upgrade" to LTE.

The Nokia brochure maybe mentions once in passing that "LTE is a different Air Interface", technical speak for "totally different and incompatible".

All the improvements of I-HSPA or HSPA+ only apply to about 1/4 of service area in a small cell in a dense urban /city network. They are all marginal in a Rural setting over 3.6Mbps HSDPA. Even 7.2Mbps is only for maybe 15% to 20% of urban cell area. The 14.4Mbps and 21Mbps sector speeds are at WiFi hotspot type urban distances, almost non-existent on a Rural cell. Rural cells it's hard to get more than 2.5Mbps sector throughput shared to ALL users. Ten simultaneous users thus get on average 200kbps.

It can't emphasised enough the misleading and totally hyped nature of quoting Mast speed without the number or location of users. The oft quoted Mobile Speeds are only real if the Mobile companies practically give each user their own mast sector.

Fibre to the home rurally is a fraction of the cost of a true 1Mbps per user Mobile system! See how much 2000 masts would cost, 3 Billion Euro.


You can't beat the Inverse Square law of physics.

cowboy1981

watty wrote: »

It can't emphasised enough the misleading and totally hyped nature of quoting Mast speed without the number or location of users. The oft quoted Mobile Speeds are only real if the Mobile companies practically give each user their own mast sector.

Presumably 3 will allow mobile phone users and international roamers to hog capacity on the Govt. subsidised base stations in rural areas. They may even give priority to this kind of traffic, as it will earn them more revenue? Imagine what it will be like for locals if there is a international conference in the local hotel, GAA match or other gathering. Such gatherings will tend to occur at the centre of population, near the base station (maybe not even an NBS area?), but will impact on the residents at the edge of the cell. The population of many rural areas swells during the Summer, with holiday homes and Gaeltacht students etc. Is the NBS designed to cater for seasonal traffic? There is nothing in any of the published documentation to suggest that the NBS investment will be exclusively for use of NBS residents.

watty

It's simply 79M toward the 220M rollout that 3's licence requires them to do.

I don't beleive there are any dedicated NBS basestations. Voice calls always get priority on Mobile masts, as there is maybe x400 revenue.

compare cost of 12Gbyte data (€20 approx) for 1month 1 user data allowance with cost of over 2000 hrs of phone calls (same amount of data)
That's the €29 euro 150min price plan for 800 people per month
http://www.three.ie/priceplans/paymonthly.htm
= €23,200 for phone calls vs €20 for data.

So comparing NBS data price and 3 pay monthly 150minute package the voice calls are
€23,000 extra revenue or 1,160 times revenue per data subscription.

With voice they have to pay the termination charge if it goes off their network (but they get similar amount for all the INCOMING calls to you) so if they are lucky the calls you make and the calls you receive cancel costs.

The backhaul is the same as we are comparing the same 12GBytes of traffic.

The Internet traffic of their network they have to pay peering charges.

So in case of 3 the ratio of voice to data is about 1,000 times. I think for some Vodaphone closer to 400 times. You can see why vodafone only has 5Gbyte cap.

Does anyone disbelieve that voice calls always have priority over data and that Mobile Internet charges are artifically low compared with fixed wireless/DSL/Cable broadband which has much lower costs?

You can give EVERYONE 100Mbps fibre to home for the full cost of a 100% coverage National 3G/HSPA mobile network.

The only more expensive way to deliver FIXED internet connectivity is Satellite. Mobile Phone Network is designed for MOBILE use, it's not efficient in performance or cost for Fixed Data networking. It's primarily designed for many voice calls a day, not an always on Broadband connection.

BTW, at continuous 1Mbps download you use up a 12Gbyte cap in about 26 hours.

You can sell a sector to about 100 data users before it's totally brain dead. It could support about 170 simultaneous phone conversations. 150mins is about 1/288th of a month. If half the calls are incoming then you could maybe support 170 x 144 phone subscriptions on a sector = 24,000 approx.
That's about 74,000 customers on a mast.
even if customers use 600mins each on average and all the same 1/4rd of the day = 170 x (144/(4x4)) = 1,530 customers at €79 each = about 120,000 Euro revenue

For a mast more than 300 fixed Internet data customers is crazy. = 300 x 20 Euro = 6,000 Euro

You see why over Cap data can be €250 a Gigabyte?

Since the 160 masts covers 400,000 people, there is no difficulty with supporting the voice calls.

This all also explains why some operators will offer a data-plan cheaper for a phone that can't be used as a laptop modem. Are you going to eat 1Gbyte never mind 10GB on a 2.1" screen browser & T9 text emails?

cowboy1981

watty wrote: »

It's simply 79M toward the 220M rollout that 3's licence requires them to do.

The NBS contract will probably not cost the state anything like as much as €39M. Even if it ever gets built and the state is stupid enough to pay out for a contract that cannot possibly meet the requirements, the State is set to earn a significant amount of revenue from the 160 new 3Gbase-stations. 3 are also likely to require additional microwave repeater sites since they don't have any fibre backhaul.

In the rural areas where NBS is focused, the main land and infrastructure owners with suitable sites for 3G masts are State bodies - eg Coillte, ESB, councils etc. The state should re-coup a significant proportion of the subsidy, and the site rental will likely continue long after the 5 year subsidy runs out.

This lucrative source of revenue for the State may have influenced the bizarre choice of 3 as vendor. Presumably the State would not have received anything like as much income if eircom had won, since eircom already own or have use of a wide variety of suitable infrastructure throughout the country (eg exchange buildings, masts for FWA/Meteor/Tetra, phone boxes and telegraph poles).

watty

eircom sold most masts to Threefold. The return to State directly would be very similar on eircom tender, but 3 was a lot cheaper. eircom are looking for planning permission for a lot of Tetra masts.

The state gets about €70M a year from Telcos via Comreg surplus.

There is estimated that about 3,500 or more (capped at about 17,000) would get Satellite. The NBS subsidizes the install and running cost which over 2 years would be about 10 to 15 Million on the eircom tender. I suspect the costs 3 quoted are much less on that as they are using Avanti who are about cheapest.

3 has to roll out most of those mast sites anyway, so can quote as little as they want on the rest of the tender. No surprise 3 is a lot cheaper.

Remember eircom was going to use Meteor Edge/3G/HSDPA for as much as possible and only four entrants. BT & Motorola/IFA withdrew as they could not economically meet the NBS tender as specified.

3 is not the basic issue. The basic issues were the way the NBS was designed and the decision to award based on price apparently without due dillengence on technical performance and changed Mapping vs effect on existing WISP coverage. Given the solutions offered by the two remaining bidders the tender should not have been awarded at all but the scheme redesigned for true 100% real Broadband and solving competitive issues.

I don't think the tender was award on basis of "getting money back" via 3's infrastructure rollout, that will happen anyway even if 3 didn't get the NBS.

See ARPU and 120% penetration.
3 want a cut of those 400,000 people in the NBS areas as mobile phone users

Telefonica O2 Ireland reported a drop of 3.4 % to €234 million in service revenue for the three months to the end of September [2008] compared with the same period last year.

http://wirelessfederation.com/news/o2-ireland-arpu-drops-due-to-extra-value-for-customers/

The 160 news masts will help 3 (the smallest Mobile company) take share from O2, Meteor and Vodafone. There are no new customers to get. The €79M grant and €19.95 pm and NBS endorsement just distorts competition with the existing Better service unsubsidised Wireless ISPs in many of these areas and the other Mobile Companies.

All the Mobile companies heavily subsidize the data modem packages via voice call sales.

cowboy1981

watty wrote: »

eircom sold most masts to Threefold. The return to State directly would be very similar on eircom tender, but 3 was a lot cheaper.

Threefold isn't owned by the State - so my point still stands - eircom would not have to rent as many sites from the State as 3 would. The State will get a lot more money out of 3 than they would have from eircom. What makes you believe that 3 was a lot cheaper? Have some figures been published for the different bids?

watty

Anyone can rent the threefold sites. I'd be sceptical that much money would come to State via site rental due to 3 rather than Meteor.

There are plenty of issues with NBS, but this is not one of them.