Transcript xDSL Overview Oct 08

xDSL
Technical Overview
Oct 08
1
DSL Market Drivers & Enablers
B-Box / SAI
VDSL over OSP
Twisted Pair
SAI
CO
NID/Splitter
POTS
Res.
Gateway
OR
IP DSLAM
NID/Splitter
ADSL over OSP
Twisted Pair
STB
Consumer drivers




IPTV
More upstream data
High-speed internet data
Consolidated billing
Service Provider Drivers


Telco's desire to compete with
Cable companies
Additional service(s) = revenue
STB
STB
Enablers



IC Technology advancements
Leverage ADSL and extend
frequency range/bitrate
ITU standards finalized
2
DMT
Discrete Multi-Tone
Each one is controlled by the DSL protocol based on actual line conditions.
BITS/TONE
15 Max
Vf Upstream
Downstream
MHz
4.3125 Khz
One sub-carrier, “tone” = 4.3125 Khz passband
DMT uses 256 “tones” to carry bits/data for ADSL, 4096 for VDSL2
Each “tone” can carry up to 15 bits (QAM)
3
Signal Attenuation
SNR is responsible for the performance
Transmitted Signal Power
Received Signal Power
Signal to Noise Ratio (SNR)
Received Noise Power
64 kHz
MHz
Frequency
4
Bits per Tone
With good SNR we got more Bits
Bits per tone
Transmitted Signal Power
15
As distance increases from the DSLAM,
signals attenuate on the copper loop reducing
difference between noise and the signal
restricting the number
of bits each
DMT carrier
Received
Signal
Power
can support.
Received Noise Power
. . .
0
64 kHz
MHz
Frequency
5
Standards
Family
ITU
Name
Ratified
Maximum
Speed capabilities
ADSL
G.992.1
G.dmt
1999
8 Mbps down
800 kbps up
ADSL2
G.992.3
G.dmt.bis
2002
8 Mb/s down
1 Mbps up
ADSL2plus
G.992.5
ADSL2plus
2003
24 Mbps down
(Amend 1, 29Mbps)
1 Mbps up
ADSL2-RE
G.992.3
Reach Extended
2003
8 Mbps down
1 Mbps up
VDSL
G.993.1
Very-high-data-rate
DSL
2004
55 Mbps down
15 Mbps up
VDSL1 -12 MHz
long reach
G.993.2
Very-high-data-rate
DSL 2
2005
55 Mbps down
30 Mbps up
VDSL2 - 30 MHz
Short reach
G.993.2
Very-high-data-rate
DSL 2
2005
100 Mbps up/down
6
Technology
ADSL - VDSL Frequency Ranges & Rates
VDSL2
VDSL
ADSL2+
ADSL
17.66MHz
25kHz 1.1MHz 2.2MHz
12MHz
Frequency
30MHz
Technology
Freq range
Max Rates
Max # of carriers and
Bin spacing
ADSL
25kHz – 1.1MHz
800kbps up
8Mbps down
256 with 4.3125kHz bins
ADSL2+
25kHz – 2.2MHz
1Mbps up
24Mbps down
512 with 4.3125kHz bins
Amend. 1 = 29 Mbps down
VDSL(1)
25kHz – 12MHz
15Mbps up
55Mbps down
2782 with 4.3125kHz bins
VDSL2
25khz – 30MHz
100Mbps up
100Mbps down
4096 with 4.3125kHz bins
3478 with 8.625kHz bins
7
What are VDSL2 – Key Features
– Improvements to initialization, including a Channel Discovery phase
and a Loop Diagnostics mode
– Improved framing based G.992.3 (ADSL2) with improved overhead
channel
– Support of Impulse Noise Protection (INP) up to 16 symbols
– Support for a MIB-Controlled PSD mask mechanism for in-band
spectral shaping
– Support for an optional extension of the USO band to 276 kHz
– Improved FEC capabilities, including a wider range of settings for the
Reed-Solomon encoder and the inter-leaver
8
ADSL2+/VDSL/VDSL2 - Rate versus Reach
250
DS ADSL2+ (2.2 MHz)
Symmetrical
100Mbit/s due to
30MHz spectrum
200
DS VDSL1 (12 MHz)
DS VDSL2 (30MHz)
Rate / MBit/s
AWGN/-140dBm/Hz/ANSI-TP1
150
Improved mid range performance
through Trellis coding and Generic
Convolutional Interleaver
100
ADSL-like long reach
performance due to
Trellis coding and Echo
Cancellation
50
0
Reach / m
Reach / ft*
0
500
1000
1500
2000
2500
3000
3500
1600
3300
4900
6600
8200
9900
11,500
9
VDSL Rate and Reach
AWG26, Gap = 12dB, 20-self, Tx PSD = -53 dBm/Hz
140
30 MHz
25 MHz
20 MHz
17.6 MHz
12 MHz
8.5 MHz
4.4 MHz
2.2 MHz
1.1 MHz
120
Rate (Mbits/s)
100
80
60
40
20
0
0
1
2
3
Loop Length (kft)
4
5
6
10
Bonded Service
 A way to increase rate and reach over single pair limitations
 Multiple physical pairs carrying a portion of the total bit
stream.
 Three approaches:
– G.998.1, ATM based
– G.998.2, Ethernet based
– G.998.3 Time –division Inverse Mux
 VDSL will use an Ethernet approach with “muxing” at the
TC layer with a new aggregation and rate matching
function.
 May not achieve double the rates due to VDSL cross talk in
the same binder group
11
Ham Radio Notches
Table 6-2/G.993.1 – Transmit notch bands
Band start
(kHz)
Band stop
(kHz)
1 800
2 000
3 500
4 000
7 000
7 300
10 100
10 150
14 000
14 350
18 068
18 168
21 000
21 450
24 890
24 990
28 000
29 700
12
Band Plans – VDSL
4-Band
3-Band
2-Band
Band Plan
(G.993.2, Annex A)
1-Band
D1
U0
4-25 kHz
U1
138-276 kHz 3.75 MHz
D2
U2
5.2 MHz
12 MHz
8.5 MHz
6-Band
5-Band
4-Band
3-Band
2-Band
Band Plan
(G.993.2, Annex C)
1-Band
D1
= Radio Notches
U1
640 kHz 3.75 MHz
D2
5.2 MHz
U2
8.5 MHz
D3
12 MHz
U3
17.7-18.1 MHz
30 MHz
U0 is used for VDSL Long Range Products (VLR)
13
Band Plans – VDSL
4-Band
3-Band
2-Band
Band Plan 998
1-Band
(G.993.2, Annex B)
U0
25 kHz
D1
U1
138-276 kHz 3.75 MHz
D2
5.2 MHz
U2
12 MHz
8.5 MHz
4-Band
3-Band
2-Band
Band Plan 997
1-Band
(G.993.2, Annex B)
U0
= Radio Notches
25 kHz
D1
U1
138-276 kHz 3.0 MHz
U2
D2
5.1 MHz
7.05 MHz
12 MHz
U0 is used for VDSL Long Range Products (VLR)
14
Band Plans for VDSL
Frequencyplans
Band-edge frequencies
(As defined in the generic band plan)
997
998
f0
f1
f2
f3
f4
f5
kHz
kHz
kHz
kHz
kHz
kHz
25
138
25
276
3000
5100
7050
12000
138
276
25
138
25
276
138
276
3750
5200
8500
12000
N/A
276
DS1
Opt
US1
DS2
US2
f(MHz)
f0
f1
f2
f3
f4
f5
T1544750-02
15
VDSL2 Profiles
•
•
•
Profiles are specified to allow transceivers to support a
subset of the allowed settings and still be compliant with
the recommendation.
The specification of multiple profiles allows vendors to
limit implementation complexity and develop
implementations that target specific service requirements.
The eight VDSL2 profiles (G.993.2):
8a, 8b, 8c, 8d, 12a, 12b, 71a, 30a,
define a set of configurations for transmit power and band plan.
•
Service Providers are now using these terms
16
VDSL2 Some Favored Profiles
Maximum aggregate downstream transmit power (dBm)
Maximum aggregate upstream transmit power (dBm)
Subcarrier spacing (kHz)
Minimum net aggregate data rate (Mbit/s)
Typical use case
8b
+20.5
+14.5
4.3125
50
CO
17a
+14.5
+14.5
4.3125
100
FTTN
Annex A,
Annex B (998):
1971
(8.5)
N/A
30a
+14.5
+14.5
8.625
200
FTTB
Japan
N/A
1205
(5.2)
N/A
N/A
1971
(8.5)
4095
(17.664)
2098
(18.1)
1205
(5.2)
2782
(12)
3478
(30)
Annex C
Index of highest supported
downstream data-bearing subcarrier
(upper band edge frequency in MHz
(Informative))
Index of highest supported upsteam
data-bearing subcarrier (upper band
edge frequency in MHz (informative))
Index of highest supported
downstream subcarrier (upper band
edge frequency in MHz (informative))
Index of highest supported upstream
subcarrier (upper band edge
frequency in MHz (informative))
Note: While Annex C is specified as for Japan, other regions are using those profiles
17
VDSL2 Spectrum Capability
•
•
For exchange deployment
– VDSL2 spectrally compatible with ADSL/ADSL2 (138kHz to
1.104MHz) and with ADSL2+ (138kHz to 2.208MHz)
For cabinet deployment
– VDSL2 spectrally compatible with cabinet-based ADSL2+
– Power control needed to ensure spectrum compatibility with
exchange based services (138kHz to 2.208MHz)
– Achieved by shaping the cabinet signals by a factor based on
the electrical distance between the exchange and cabinet
– Degree of shaping defined via MIB control (G.997.1)
– Enables VDSL2 to comply with regulatory requirements
– VDSL2 PSD shaping currently being investigated by various
European and Asian operators
18
VDSL2 PSD Shaping
 PSD shaping in VDSL2 facilitates coexistence between ADSL/2/2+ from
the CO with ADSL2 from the cabinet.
 PSD shaping functionality exists already in ADSL2+
– Compared to ADSL2+ VDSL2 has extended the parameter range
– Likely to be amended to ADSL2+ as well
 Different level of transmit power makes disturbance in the same binder
– need adjustment.
 One configuration example:
Crosstalk from VDSL
effecting ADSL:
PSD management
approach
Exchange:
OLT
Optical
Node
DSLAM
20 to 25 M bps
for VDSL
VDSL
Profile 17a
VDSL
Profile 8b
ADSL2+
19
OLR - Dual Latency (Fast and Interleaved Paths)
Dual Latency refers to bearer channels that can have different latency
treatments as defined by such things as interleave depth, INP settings
and FEC configurations.
 Fast path has low latency (<1ms).
– Good for voice traffic.
– People perceive delay negatively during a conversation.
– Losing (small amounts of) data is not critical. Most CODECs will
disguise lost data by replaying the previous audio.
 Interleaved path has more latency (up to 10ms) but has better immunity
to disturbers such as impulses.
– Guaranteed to correct errors due to impulses <250μs.
– Good for data and video.
– Data and video are tolerant of delay (not "delay variation" that's
jitter) but are not tolerant of lost data
20
On-Line Reconfiguration (OLR)

Reconfiguration takes four forms:
Bit Swapping (BS), Seamless Rate Adaptation (SRA). Dynamic Rate
Repartitioning (DRR) and Dynamic Spectrum management (DSM)

BS reallocates data and power (i.e. margin) among the allowed sub-carriers
without modification of the higher layer features of the physical layer. Bit
Swapping reconfigures the bits and fine gain parameters without changing any
other PMD or PMS-TC control parameters.
SRA is the ability to change data rates in real-time based on monitoring
changing line conditions and adjusting such things as bit swapping, DMT
symbol bit assignments and DMT bins in use without losing frame sync.
DRR is used to reconfigure the data rate allocation between multiple latency
paths by modifying the frame multiplexer control parameters. DRR can also
include modifications to the bits and fine gain parameters, reallocating bits
among the sub-carriers. DRR does not modify the total data rate, but does
modify the individual latency path data rates.
DSM enables transceivers to autonomously and dynamically optimize their
settings for both channel and neighboring systems, reducing crosstalk
significantly.



21
OLR - Seamless Rate Adaptation (SRA)
 SRA dynamically monitors line conditions and adjusts bit rates to take
advantage of improved conditions and reduces bit rates if necessary
without loss of sync.
 Parameters and their typical values used for SRA
– Downshift margin up = 3 dB
– Downshift interval up = 60 seconds
– Downshift margin down = 3 dB
– Downshift interval down = 60 seconds
– Upshift margin up = 3 dB
– Upshift interval up = 60 seconds
– Upshift margin down = 3 dB
– Upshift interval down = 6 seconds
 The effect is to increase bit rate performance
22
OLR - Dynamic Rate Repartitioning (DRR)
 DRR monitors the bandwidth on a connection and
reallocates the bandwidth per path allowing the available
bandwidth to be used more efficiently.
– It achieves this by modifying the framing parameters and by using
bit swapping.
– The reallocation of the bandwidth is done seamlessly without
disturbing the user’s applications (video stream, VoIP call, surfing
the net).
– The total delivered bandwidth is not changed. It will reallocate the
bandwidth assuring each application gets the highest possible QOS.
23
Dynamic Spectrum Management (DSM)
 Static Spectrum Management (SSM) setup as part of network
engineering guarantees that all of the DSL lines in binder are spectrally
compatible. Since services running on the DSL lines are dynamic, static
management typically wastes bandwidth.
 DSM takes advantage of dynamically changing conditions and improves
the wasted channel capacity left by SSM.
 The ultimate DSM solution requires monitoring of the line conditions
by a central processing unit as well as the individual modems
monitoring line conditions as well.
 The central DSM unit monitors:
– Line margin
– Tx Power Levels
– Bits/tone tables
– Insertion loss/tone
– Noise/tone
– Actual PSD levels/tone
– Errored seconds
– Known service items such as bridge taps, loop lengths, and binder
service area (so they know what other services are in the same
binder)
24
Dynamic Spectrum Management (DSM)
 There are 4 levels of DSM coordination between
multiple DSL lines
– Level 0 Static Spectrum Management (SSM)
– Level 1 Autonomous power allocation (Single –user)
– Level 2 Coordinated power allocation (Multi – user)
– Level 3 Multi-pair, multiple-input, multiple-output (MIMO)
25
DSM (The Four Levels) Level 0
 Level 0: The performance of one individual pair is
optimized without considering the other pairs in the
binder
– Rate Adaptive (RA) and Margin Adaptive (MA) modes of
operation.
• RA mode – All available power is used to maximize rate at the
required margin
• MA mode – All available power is used to maximize margin at a
fixed rate.
26
OLR – DSM (The Four Levels) Level 1
 Level 1: Each pair in a binder manages power so
as to avoid crosstalk with the other pairs in the
binder. This will lead to an increased total capacity
in the binder.
– Power Adaptive (PA) or Fixed Margin (FM) and Iterative
Water Filling (IWF) are modes of operation.
• PA – Power is minimized while maintaining a fixed rate and noise
margins that are specified in a given range.
• IWF – Very similar to PA except IMF does not adhere to a fixed
PSD, therefore ‘boosting’ is allowed. IWF can increase the power
in used tones by reallocating power from unused tones.
27
OLR – DSM (The Four Levels) Level 2
 Level 2: Similar to level 1; Here however, the
central DSM center considers the other pairs line
conditions as well.
– Optimal Spectrum Management (OSM) aka Optimal
Spectrum Balancing (OSB)
• The central DSM knows the cross-talk paths, the loop lengths,
and the service requirements of each pair in the binder. All the
used spectra is optimized by the central DSM by setting the
PSDMASK parameters for each pair based on the DSM
prediction of the complete binder performance. So for example, a
short line may be told to use the higher frequencies even though
the lower frequencies would have been used if only IWF was
applied.
28
OLR – DSM (The Four Levels) Level 3
 Level 3: The central DSM processes all of the signals from all the pairs
in a binder at once. All transmitters and/or receivers must be co-located.
– The central DSM will jointly process all of the signals in the binder rather
than processing each line individually.
– The binder is considered a whole entity aka (MIMO or vectoring). All the
signals are combined into a vectored signal and processed together. With
the joint processing, it is now possible to predict the induced crosstalk on
the other lines. That predicted crosstalk signal can be subtracted from the
actual received signal to reduce the crosstalk.
– This can be implemented in a point-to-point configuration or a point-tomultipoint configuration.
• Point-to-point – All processing is done at the receiver.
• Point-to-multipoint – One CO multiple CPE all processing is done at the CO.
29
OLR – Dynamic Spectrum Management (DSM)
30
Impulse Noise Protection

The basic idea with INP is to separate (in time) the data and the corresponding
error correction bytes for that data. This helps ensure that if an impulse occurs
at time t0 only the data will be corrupted; the RS correction bytes allow the data
to be fixed.
–
–
More memory is needed to store the data while waiting for the error correction data.
INP causes the data to be delayed.
Line 1
Frame #1
Error correction for
Frame #1
Error correction for
Frame #2
Frame #2
Line 2
Frame #1
Frame #3
Error correction for
Frame #1
Frame #2
Frame #3
Error correction for
Frame #3
X
Frame #4
Error correction for
Frame #2
X
Frame #4
Error correction for
Frame #4
X
X
Frame #5
Error correction for
Frame #3
X
X
Frame #5
Error correction for
Frame #5
Error correction for
Frame #6
Frame #6
Error correction for
Frame #4
Error correction for
Frame #5
Frame #6
Time
31
INP – ADSL2+
Down-stream
Significant Throughput Impact
32
INP – ADSL2+
Amendment 1 Down-stream
Significant Throughput Impact
33
INP – ADSL2+
Up-stream
Significant Throughput Impact
34
Impulse Noise Impairments
 VDSL is more susceptible to impulse noise events due to it’s use of a
wider frequency spectrum than ADSL. Noise sources are being
analyzed in several forms:
– REIN (Repetitive Electrical Impulse Noise)
• Less than 1 ms in duration
• No bit errors desired
• INP mitigation
– PEIN (Prolonged Electrical Impulse Noise)
• 1 to 10 ms in duration
• No bit errors desired
• INP mitigation
– SHINE (Single Isolated Impulse Noise Event)
• Duration greater than 10 ms
• Due to duration of events, bit errors will typically occur
• No loss of sync is desired
35
Transient – Long Term Interference Noise
Transient or longer term noise
sources make critical impacts on
DSL service performance:
•AM Radio
SW Station at
13.615 MHz
•Many operate, both base
band frequency of station and
difference signal between two
strong stations, in the ADSL
band, stronger at night
•Short Wave Radio
•Many short wave radio
stations operate in VDSL
bands from 3.2 MHz to 21.5
MHz
36
A tap acts like a filter
0
-10
Clean pair
-20
Insertion Loss (dB)
-30
-40
-50
44ft tap
-60
-70
-80
-90
0
1
2
3
4
5
6
7
8
9
10
Freq (MHz)
37
Longer taps = less impact
0
-10
Clean pair
-20
Insertion Loss (dB)
-30
-40
-50
-60
100ft tap
75ft tap
-70
50ft tap
-80
44ft tap
Short taps (under 200 ft) have more impact on VDSL
-90
0
1
2
3
4
5
6
7
8
9
10
Freq (MHz)
38