Network Requirements for DSL systems, (ADSL through G.Fast) (A summarized view) Gilberto GG Guitarte, BB Connectivity Director TE Connectivity FTTH LATAM Chapter CHAIRMAN 2/24/2014 G.A.Guitarte 1
Executive summary 1)The higher the bandwidth, the more demanding the conditions on the copper loop will be. 2)The design rules for DSL are the same as for the DLC revised DR. 3)The SNR for a given bandwidth has to be greater than 30 db in order for that bandwidth to produce the maximum capabilty of it in terms o bps.(see slides 3 an 4) 4)The Attenuation above 250kHz follows increase with the Square Root of the Frequency increase.(see slides 5 and 6) 5)Excessive Attenuation, or excessive Noise or a combination of both will degrade SNR across the bandwidth for the DSL service being used and thus reduce the capacity od the Channel in bps, as expressed in Shannon s equation (see slides 12 to 20) 6)Bad insulation in the OSP Cables over copper wires may degrade the transmission performance of that cable..telcordia GR 421.Look for 30 to 40 minutes OIT minimum! (see slides 59 to 63) 7)Missing shields or missing grounds can favor Noise ingress in the copper channel and thus increase Outside sourced NOISE in the active pairs, decrease SNR, and decrease channel capacity C in bps. 8)Inside sourced noise ( SELF-FEXT) in the cable group of pairs, which degrades SNR, can be mitigated by vectoring. (At an extra cost 2/24/2014 G.A.Guitarte 2
Performance of QAM-n related to the SNR BER Non acceptable zone 10-7 10-10 SNR 2/24/2014 G.A.Guitarte 3 db
Signal / Noise related to QAM - n Bits/symbol QAM Signal/Noise ratio (db) for BER<10-7 4 QAM-16 21,8 6 QAM-64 27,8 8 QAM-256 33,8 9 QAM-512 36,8 10 QAM-1.024 39,9 12 QAM-4.096 45,9 14 QAM-16.384 51,9 2/24/2014 G.A.Guitarte 4
From ADSL to ADSL2+ to VDSL to VDSL2 (and now to G.Fast too.) Things get worse and more demanding requiring a very stable copper plant because In the realm of DSL, above 250 khz Att.(f2) = Att.(f1) f2/f1 f i >250kHz 2/24/2014 G.A.Guitarte 5 page 5 / February 24, 2014
ATTENUATION as a function of the Frequency of Operation From ADSL to ADSL2+ to VDSL to VDSL2 1)ADSL f1= f max = 1.1 MHz α1 =attenuation @ f1 2)ADSL2+ f2= f max = 2.2 MHz α2 = α1 2.2/1.1= α1 (1.4) 3)VDSL f3= f max = 12 MHz α3 = α1 12/1.1 = α1 (3.3) 4)VDSL2 f4= f max = 30 MHz α4 = α1 30/1.1 = α1 (5.2) 2/24/2014 G.A.Guitarte 6
POTS ADSL & ADSL 2 ADSL 2+ bandwidth 0 4kHz 1.1MHz 2.2MHz 2/24/2014 G.A.Guitarte 7
The Copper-Loop Plant Environment TP s are grouped into binders of 10,25, or 50 pr/group Design Rules were dictated for feeder and distribution cables (not including the drops) CO F1 F2 Feeder XC Cab Distrib Drop CPE Terminal 2/24/2014 G.A.Guitarte 8
The Rules POTS-ResistanceDesign (RD) Loop Resistance<1500Ω Coil when total loop length (including bridge taps > 15 kft. 88 mh coils @3kft. From CO and intervals of 6 kft. In loaded cables, total length (including bridge taps) beyond last coil <3 to 12 kft> No bridge taps between coils No loaded bridge taps DLC-CSA- Revised RD No coils Max 2 different gauges in one loop Total sum of bridge tap length < 2.5kft. Each bridge tap < 2.0 kft. 26AWG cable < 9kft. Including bridge taps Single gauge or multigauge cable with 19,22,or 24AWG, total cable length including bridge taps < 12 kft. If above includes 26AWG, total cable length < 12-3(L26) kft 9-Ltap 2/24/2014 G.A.Guitarte 9
The Rules DSL DLC-CSA-Revised RD No coils Max 2 different gauges in one loop Total sum of bridge tap length < 2.5kft. Each bridge tap < 2.0 kft. 26AWG cable < 9kft. Including bridge taps Single gauge or multigauge cable with 19,22,or 24AWG, total cable length including bridge taps < 12 kft. If above includes 26AWG, total cable length < 12-3(L26) kft 9-Ltap 2/24/2014 G.A.Guitarte 10
-10 Channel Capacities Transmit Signal -50-100 -140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 11
-10 9 kft 26AWG Insertion Loss Transmit Signal -50-100 -140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 12
-10 9 kft 26AWG Insertion Loss Channel Capacities Transmit Signal -50 Receive Signal -100-140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 13
-10 Channel Capacities -50 Receive Signal -100-140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 14
-10 Channel Capacities -50 Receive Signal -100 Noise -140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 15
-10-50 Channel Capacities S/N = SNR = 10 logps/pn 10 log Ps 10 log Pn = Signal (db) Noise (db) Receive Signal -100 Noise -140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 16
-10-50 Channel Capacities 40 (db) SNR 30 20 10 0 f(khz) Receive Signal -100 Noise -140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 17
-10-50 SIGNAL to NOISE RATIO (SNR) Shannon Capacity for a Channel C= log 1+SNR(f) dƒ 2 0 = C= (Bps) C, in Bps,is the Area between SIGNAL and NOISE ReceiveSIGNAL -100 ReceiveNOISE -140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 18
-10-50 SIGNAL to NOISE RATIO (SNR) Shannon Capacity for a Channel C= log 1+SNR(f) dƒ 2 0 Receive SIGNAL -100-140 Receive NOISE 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 19
GOOD! spectrum with good acceptable SNR > 30dB -10 Shannon Capacity SIGNAL to NOISE for RATIO a Channel (SNR) -50 C= log 1+SNR(f) dƒ 2 0 Receive Signal - 10 0 GOOD Receive Noise February 24, 2014 page 20 / Frequency (k Hz) page 20 / February 24, 2014
BETTER! More spectrum with same good acceptable SNR>30dB -10 Shannon Capacity SIGNAL to NOISE for RATIO a Channel (SNR) -50 C= log 1+SNR(f) dƒ 2 0 Receive Signal BETTER! - 10 0 Receive Noise February 24, 2014 page 21 / Frequency (k Hz) page 21 / February 24, 2014
-10 WORSE! More spectrum with SNR< 30dB Shannon Capacity SIGNAL to NOISE for RATIO a Channel (SNR) -50 C= log 1+SNR(f) dƒ 2 0 Receive Signal WORSE! - 10 0 Receive Noise February 24, 2014 page 22 / Frequency (k Hz) page 22 / February 24, 2014
Types of Cross-Talk Noise NEXT FEXT 2/24/2014 G.A.Guitarte 23
NEXT Signal Source Vj(t) @full power on TP j Tx Disturbing Near & Far End TP j NEXT Rx Xn(t) Disturbed Near End TP i Xf(t) Disturbed Far End 2/24/2014 G.A.Guitarte 24
FEXT Signal Source Vj(t) @full power on TP j Tx Disturbing Near & Far End TP j FEXT TP i Xn(t) Disturbed Near End Xf(t) Disturbed Far End Rx 2/24/2014 G.A.Guitarte 25
NEXT & FEXT Please refer to ANSI T1. 413 ANNEX B, page 165 Graphs B.2, B.3, B.4, and B.6 2/24/2014 G.A.Guitarte 26
-10 9 kft 26AWG Insertion Loss Channel Capacities Transmit Signal -50 Receive Signal -100-140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 27
Impulse Noise (I-Noise) Please see ANSI T1.413 Page 140 Point 11.2.2 2/24/2014 G.A.Guitarte 28
-10 Wide Band Noise (W-Noise) -50 Sun of all unwanted signals in the frequency Spectrum of interest (30kHz to 1.1 MHz) Receive Signal -100-140 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 Frequency ( Hz) 2/24/2014 G.A.Guitarte 29
From ADSL to ADSL 2+ 2/24/2014 G.A.Guitarte 30
-10 9 kft 26AWG Insertion Loss Transmit Signal -50-100 -140 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2/24/2014 Frequency G.A.Guitarte ( Hz) 31
ADSL 2+ Increase in downstream bandwidth up to 2.2 MHz POTS ADSL & ADSL 2 ADSL 2+ bandwidth 0 4kHz 1.1MHz 2.2MHz 2/24/2014 G.A.Guitarte 32
-10 b 9 kft 26AWG Insertion Loss Transmit Signal -50-100 ADSL2+ Rx signal power is too attenuated with 9 kft. AWG26! -140 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2/24/2014 G.A.Guitarte 33 Frequency ( Hz)
-10-50 1mi/24/0.6 1mi/26/0.4 2mi/24/0.6-100 3mi/24/0.6 2mi/26/0.4-140 3mi/26/0.4 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2/24/2014 Frequency G.A.Guitarte ( Hz) 34
-10-50 1mi/26/0.4-100 -140 10x ADSL 2+ Disturbers 4xT1 disturbers 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2/24/2014 Frequency G.A.Guitarte ( Hz) 35
-10-50 -100 1mi/26/0.4-140 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2/24/2014 Frequency G.A.Guitarte ( Hz) 36
-10-50 Portion of spectrum usable (at or abpove 30dB SNR) SNR(dB) 40 30 20 10 0-100 1mi/26/0.4-140 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2/24/2014 Frequency G.A.Guitarte ( Hz) 37
So What with noise on the line? Standards tend to define the value of 30 db for SNR as a lower limit for acceptance of a loop for both POTS and DSL NOISE SOURCES: NEXT/FEXT/Proximity to Power Lines/ Electromagnetic field in general/am Radio/FM Radio/Other POTS/ Other what you want to name it, it is out there, in the universe. NOISE EFFECT ENABLERS: >Poor Cable Shielding and shielding continuity >Poor Grounding and Poor Grounding practices >Poor IR between conductors of PIC >Poor Connectivity (High CR) between connectors and PIC. >PIC degradation (decreasing OIT) >Many users of DSL in same cable generating NEXT and FEXT 2/24/2014 G.A.Guitarte 38
The Highway for outside sourced Noise? 2/24/2014 G.A.Guitarte 39
The Highway for Noise? = Shielding 2/24/2014 G.A.Guitarte 40
Noise on the road again ( a great traveller) Shield cont... PE Al Mylar PE/PP/Foam-Skin Cu 2/24/2014 G.A.Guitarte 41
The Bus STOP for Noise? 2/24/2014 G.A.Guitarte 42
The Bus STOP for Noise? = Grounding 2/24/2014 G.A.Guitarte 43
The Bus STOP for outside sourced Noise? = Grounding 2/24/2014 G.A.Guitarte 44
How as Noise increases (outside generated or inside generated or both!) Two following slides show how an ADSL2+ loop which originally had 75% of the 2.2 MHz spectrum available producing a max of 18Mbps,degrades to 5% or 1.1Mbps max throughput as the total received Noise increases at the user end, produced by FEXT, and bad loop condition in terms of shielding, grounding, insulation, etc. 2/24/2014 G.A.Guitarte 45
-10-50 SNR(dB) 40 30 20 10 0-10 0 1mi/26/0.4 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 February 24, 2014 page 46 / Frequency ( khz)
-10-50 Usable SpectrumAfter Noise increases! SNR(dB) +40 +30 +20 +10 0-10 - 10 0 Recived NOISE Received SIGNAL 1mi/26/0.4 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 February 24, 2014 page 47 / Frequency ( khz)
As you can imagine, the conditions get tighter for VDSL, and even tighter for VDSL2, because of the ATTENUATION imposed by COPPER to the higher frequencies they try to use!!! and this is the reason why VDSL and VDSL2 have interleaved segments of UP and DOWN streams of spectrum because most likely, the higher segments of UP and DOWN are not usable! In next slides, DS = Downstreamk, and US = Upstream 2/24/2014 G.A.Guitarte 48
VDSL PSD (G.993.1, Annexes A to F) dbm/hz -30-40 -50-60 -70-60 to -56.5 dbm/hz Tx signal DS1 US1 DS2 US2 2kft. AWG26-80 -90 3kft. AWG26-100 -110 Rx signal -120-140 0 1 2 3 4 5 6 7 8 9 10 11 12 MHz February 24, 2014 page 49 /
dbm/hz -20-30 -40-50 VDSL2 PSD DS1 US1 DS2 US2 DS3 US3 200 m PIC filledawg26/ 0.4mm -60-70 -80-90 -100-110 -120 0.640 3.75 5.2 8.5 12 18.1 30 February 24, 2014 page 50 / MHz
dbm/hz -20-30 -40-50 VDSL2 DS1 US1 DS2 US2 DS3 US3 200 m PIC filledawg26/ 0.4mm -60-70 -80-90 -100-110 Noise test profile per ITU -120 0.640 3.75 5.2 8.5 12 18.1 30 February 24, 2014 page 51 / MHz
dbm/hz -20-30 -40-50 VDSL2 200 m PIC filled AWG26/0.4mm DS1 US1 DS2 US2 DS3 US3-60 -70-80 -90-100 -110-120 0 0.640 3.75 5.2 8.5 12 18.1 February 24, 2014 page 52 / MHz
And now, a couple of comments on G.Fast It will use 2 spectrum segments 106 MHz and 212 MHz It will use DMT (Discrete Multi-tones, like the rest of the DSL family It will use 12bits per DMT coupled with TDM It will use FEC with Trellis encoding and Reed Solomon encoding It is supposed to reach 250m on 0.5mm diameter copper pair (this, will happen on a good sunny day, no cloud in the sky, no disturbers, or one and only one customer on the whole copper cable using it, and at low speeds) 2/24/2014 G.A.Guitarte 53
Let us apply the basic formula for the same kind of copper usage Att.(f2) = Att.(f1) f2/f1 f i >250kHz HowmuchworsedoesG.Fasthaveitthan thehighestspectrumofvdsl2atits30mhzmax? 2/24/2014 G.A.Guitarte 54
G.Fast Att at 106 MHz =Att VDSL2 106/30 = 1.88 times Att VDSL2 G.Fast Att at 212 MHz = Att VDSL2 212/30 = 2.66 times AttVDSL2 This ratios of G.Fast max. attenuations over VDSL2 max. attenuation just point to the hard fact that.. G.Fast will ONLY Work in very short loops Loops MUST be in extremely good physical conditions Loops MUST be in extremly good electrical conditions AND with the help of VECTORING. Because noise is INTOLERABLE for G.Fast due to the extremely high frequencies used 2/24/2014 G.A.Guitarte 55
CONCLUSION Ifeverybodyagreesthatthehighestcostoftheaccess Networkisinvolvedinthecivilworksoftheconstruction, andyouwillhavetogetfiberwithin50moftheuserplus InvestontoponallofthehelpyoucangetoutofVECTORING, whynotuseftth? 2/24/2014 G.A.Guitarte 56