'Green' Broadband - How much power do the various technologies consume, anyhow?

trekkypj

I was just thinking:

Is it possible to do a comparison in terms of the units of energy per year used to provide 5,000 houses in a typical rural area with:

(i) 3G Mobile Internet (i.e. the energy used to produce the signals at the mast);
(ii) DSL over copper from the exchange to the home;
(iii) Fibre Optic Cable from the exchange to the home.

I suspect that using fibre is more energy efficient, but it would be interesting to know just what energy would be needed.

After all, we all want to stop climate change.

watty

Fibre likely uses more power than DSL, but on average is 10 to 100 times faster for that.

3G is not just signal at mast. But Modems too.

Up to 5W per modem. A mast is usually three sectors. The Average sustained throughput for even distribution of users of a cell is about 2Mbps a sector. Power needed for that for all Base station processing could be easily more than 200W per user.

It's technically possible to do ADSL2+ at about 20W per user or less!

So 3G/HSPA uses at least x10 power to deliver 1/5th performance of DSL.

maybe x4 power (not sure) to deliver 1/100th of Fibre performance.

trekkypj

watty wrote: »

Fibre likely uses more power than DSL, but on average is 10 to 100 times faster for that.

3G is not just signal at mast. But Modems too.

Up to 5W per modem. A mast is usually three sectors. The Average sustained throughput for even distribution of users of a cell is about 2Mbps a sector. Power needed for that for all Base station processing could be easily more than 200W per user.

It's technically possible to do ADSL2+ at about 20W per user or less!

So 3G/HSPA uses at least x10 power to deliver 1/5th performance of DSL.

maybe x4 power (not sure) to deliver 1/100th of Fibre performance.

Thanks watty, that's pretty much what I was looking for.

watty

Only "off the top of my head".

trekkypj

watty wrote: »

Only "off the top of my head".

*grin*

Good enough to base an initial hypothesis on, though.

It'd be an interesting policy paper to do up. Might be worth looking further into it .

watty

Actually the 3G may be much worse...
http://www.telematica.polito.it/ciullo/papers/ciullo_wgreen08.pdf

Basically,the UTRAN consists of two elements: the NodeB, that is the base transceiver station, and the Radio Network Controller(RNC) , that controls one or more NodeBs. A complete UTRAN with one Node B is about 6kW

A typical BS (base station) has 3 x Node B. These are 800W each

So total BS for 3G/HSPA is 7.6kW approx for three sectors, running for say 10 people streaming. You can support far more people just web browsing, but more than 2 to 3 people per sector streaming Web video is a struggle.

So about 760W per person. Excluding backhaul and backoffice etc.

So about 4x worse for 3G than I thought.

See also bar chart here http://www.huawei.com/innovations/new_generation_node_b/simpleres.do?id=401&type=technology

For a UMTS network with 5000 Node Bs, the power consumption of a traditional Node B with 3×1 configuration is 1,500 W, while that of a new generation Node B is 760 W

That's JUST the Node B, i.e. the Radio Set for one of three sectors of a mast.

So the first reference is 800W rather than 760W quoted by Huawei for "newer" UMTS basestations. So older Mast might be nearer 9kW to 10kW, i.e. 1000W per user.

Note we are assumimg a sustained Average speed of 1Mbps and on average 10 Users simultaneously connected.

Average speed on DSL is 3Mbps.
http://www.greentelecomlive.com/2008/05/21/zte-dslam-wins-best-green-innovation-award/
A single 2U shelf of DSLAM is what DSL needs for 96 users, that can be simultaneous.
(The BaseStation for 3G is a full rack!).

http://www.rfnetech.com/downloads/IP-6348-ADSL2+DSLAM-brochure.pdf
This one supports 48 lines and uses 85W.

So if only 1 user on line out of 48 and there is no power saving, the consumption is 85W per user, 1/10th of 3G. But actually consumption per user would only be 1/500th of 3G if everyone was online. In practice DSLAM can save power by not transmitting when there is no traffic.

Home -Modem

---

phoneline

---

DSLAM-Exchange--fibre backhaul ---

FTTC basically sticks a ADSL2+ (24Mbps per user) or VDSL2(100Mbps per user) in a cabinet in your street and fibre backhaul. Since no exchange is needed, then logically FTTC is better energy, greener than DSL and Exchange and gives minimum 20Mbps speed.

FTTC makes standard DSL look stupid as you get about x8 to x20 speed on average and uses less infrastructure/energy (close all the Exchanges).

DSL and FTTC make 3G/HSPA look mindless.

The ONLY value to Mobile, is for people needing Mobility. It makes no economic or strategic sense at all for Fixed Access.

Making Mobile companies charge real prices for mobile Data (not effectively 1/100th to 1/300th of cost and subsidizing as at present) would make Fixed Internet more attractive to invest in. No-one will while Mobile is charging €10 & €20 for Data packages (if real costs used and same margin as voice and not subsidized by voice then Mobile Data would be more like €80 to €200 for 1G package.). Then speeds would be x2 to x5 better for the real Mobile users. See http://irelandoffline.org/2009/08/is-mobile-midband-in-ireland-destroying-broadband-infrastructure/

Conclusion
Data Usage on Mobile is 100x to 400x the data rate per person of voice usage on Mobile and is 500W to 2kW per user. A mast can usefully support about 10 simultaneous transfers at "entry level" Broadband speed (1Mbps), so power consumption assumes 10 users. The Up To speed can only be reached by having a single user in a sector with perfect signal. With all three sectors used by just three users the power consumption is over 2kW per user. Minimum speed is "no connect" or 0.05Mbps

Data usage on DSL or FTTC is about 2W (48 user per DLSAM) to 85W per person (1 user per 48 user DSLAM). FTTC is more efficient if you have universal BB and close the exchanges. Phone would be by integrated phone port on Modem as Digiweb, Magent and UPC provide and secure QOS managed VLAN to the ISP gateway switch ensuring ISDN voice quality. Basic DSL gives average of 3Mbps, ADSL2+ 7MBps (both 1Mbps minimum) and proper FTTC minimum of 20Mbps.

See also http://www.techtir.ie/comms/mobile-vs-fixed

Chaz

FTTH Model using GPON.

OLT - around 1800Watts per 64 port. Each port is then fed into the passive network and pending on the design can have something like 48 clients per port.

You then need to add items such as servers, switches and backhaul devices.

A design for 50 000 homes passed requires around 30KW, around 18KW is load and around 12KW is cooling (Air Handling Units).

PON ports take power, irrespective of the actual client numbers and therefore once a PON port is lit, it uses a very similar power whether there are 1 or 10 clients on a port, certainly this is the case with the hardware that I have experience on.

watty

Thanks
We didn't consider backhaul, backoffice servers etc on the other solutions. Nor did we consider that Mobile Modem may be 2W to 5W and DSL Modem Router is maybe 10W, because if you add Firewall/Router/WiFi to 3GModem you have similar consumption, which is dwarfed by consumption of laptop/PC.


So 1800W is 64 x 48 clients = 3072 users..
So compared to 48 User DSLAM = 85, 48 users is 28W, or if all 3072 users active about 0.6W.

Conclusion:
For very light utilisation (i.e. 1:48 users only ) the GPON FTTH/FTTP may be about 28W and for full utilisation it's only 0.6W per user.

Comparing 480 users "passed" or potentially serviced:
A mast that has 48:1 contention and about 1Mbps average throughput for 10 users is "passing" 480 users for 7.6kW. (30kW for 50,000 homes passed).
Per 480 users "passed" the GPON/ FTTH is about 0.3 kW, 100Mbps minimum per user
DSL or DSLAM in FTTC for 480 users passed is 0.85 kW, 1 to 24Mbps, 3Mbps Average DSL, approx 20Mbps FTTC

Even if no-one is using it. Assuming User Modem/Equipment/PC/Laptop consumption is similar and items such as servers, switches and backhaul devices etc are similar, i.e. just considering the delivery mechanism.

Chaz

Of course there is one other major 'green' cost to consider and thats the 'laying' of copper or fibre whereas for wireless, its virtually nil ...

watty

Yes, I mentioned elsewhere that it one attraction of FTTC over FTTH/FTTP that you use the exsisting copper pair in the user's street, only replacing the copper from exchange to shared cabinet for the street.

watty wrote: »

FTTH/FTTP might be greener per Mbps.

The study didn't look at figures for FTTH/FTTP. However the power consumption, no matter if more or less than FTTC, is likely to be same order as DSL and FTTC. The Mobile is really bad technology for Fixed Broadband use.

Also FTTC has lower install greenhouse costs than FTTH as it uses the existing cabinet locations and existing copper pairs to the home and can use existing modem.

In terms of performance obviously FTTH is the way to go, but FTTC can do 20Mbps minimum and can be a rapid upgrade to x6 performance for DSL on average and x20 performance for worst DSL. In contrast only the upgrades of 3G from 0.35Mbps to 1.8Mbps HSDPA, then 3.6Mbps 3G/HSDPA gave significant improvement. Upgrade to 7.2Mbps is about 10% to 20% improvement. But Upgrades to 14.4, 21 and 42Mbps 3G are Marketing. These speeds or indeed more than 6Mbps are rarely seen in the real world and due to the method of modulation used, each upgrade only applies to 1/4 of the cell area previously seeing an upgrade.

Upgrade of FTTC
There is no upgrade path for FTTC other than replacing the cabinet DSLAMS and last street length of copper to home /office by fibre, giving FTTH/FTTP

The choice between DSL from exchange, FTTC and FTTH/FTTP is not about "green" but about performance, infrastructure, roll out costs vs revenue and benefit to the whole economy. Since the benefit of FTTH/FTTP is far beyond the revenue stream from it, it should not be left to eircom. That's another story.

However studies have shown that Fixed Wireless wireless has about x8 capacity of Mobile, even on same kind of system if higher gain polarised & directional aerials are used (the low gain omni aerial limits cell capacity). Obviously the "Green" cost is a once off extra due to install of "out door Radio". So Fixed Wireless is easily x8 better power consumption per user than Mobile. If you consider a system without the complexity of 3G's RNC, such as DOCSIS over Wireless (no mobility, no live hand over), then the power saving is perhaps x12 to x16 or more than that of 3G Mobile per user.

HSPA speed vs percentage of cell locations

Green flat line is DOCSIS over Wireless normalised to 5MHz FDD channel
Right Edge is 50% of locations.

Graph from http://www.techtir.ie/comms/fixed-wireless-broadband-better

This shows how the "up to" maximum Speed of Mobile applies to a tinier part of cell area as "up to" speed is increased. Due to the low gain and Omni-directional nature the "link budget" is very poor at edge of cell or else a further off cell (not the neighbour) using the same channel would have interference. If you have N=9 rather than UK & Ireland N=3 for 3G/HSDPA, then you can run higher power and have lower intercell interference and get more speed. But you have used 3x as much spectrum!

Capt'n Midnight

most copper/fibre routers are on 24/7 (and some will have an AP as well) whereas most USB dongle things are only on when used.

watty

This is not arguing against Mobile systems for Mobile Use.

We considered the Mast vs exchange or Cabinet only as it's a "Fixed Broadband" comparison.

The figures are for "reasonable" numbers of users on a mast.
The Mast is on all the time.

If you compare like with like, add PCs, Laptops, Xboxes etc.. not just the Dongle.

Some people switch off the router / modem when not used. It's about 10W, compared with 30W to 600W per netbook/laptop/Gamer PC.

If people are using their "3G" / NBS as a Broadband replacement (which is what this exercise is about, not Mobile use of Internet), they increasingly are adding a router/AP that supports 3G dongle. This is typically same power as regular router/AP with added 1W to 5W of the 3G dongle.

If someone is promoting Mobile masts as a replacement instead of rolling out Fixed Broadband, as the NBS does, then this is a reasonable comparison.