Note: Descriptions are shown in the official language in which they were submitted.
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
IMPROVED SCHEME FOR THE INITIALIZATION OF ADSL
MODEMS
CROSS-REFERENCES TO RELATED APPLICATIONS
[0l] NOT APPLICABLE
STATEMENT AS TO RIGHTS TO INVENTIONS MADE UNDER
FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT
[02J NOT APPLICABLE
REFERENCE TO A "SEQUENCE LISTING," A TABLE, OR A COMPUTER
PROGRAM LISTING APPENDIX SUBMITTED ON A COMPACT DISK.
[03] NOT APPLICABLE
BACKGROUND OF THE INVENTION
(04] The present invention relates generally to transfer of data using Digital
Subscriber Loop (DSL) technology, and specifically to an improved scheme for
initializing
the transfer.
[OS] Remote access and retrieval of data is becoming increasingly popular
in data communication. The proliferation of the Internet has provided a vast
network of
information that is available to the general public. As the Internet grows and
technology
advances, this information is becoming increasingly voluminous and the details
are becoming
increasingly intricate. What used to comprise mainly text information has
grown to include
still and moving images as well as sound. The increase in the volume of
information to be
transferred has presented a need for a high-speed Internet connection, since
traditional
telephone modems communicate at speeds too slow for efficient communication.
[06] One proposal for high-speed communication is the introduction of
Digital Subscriber Line (DSL) technology. One of the most attractive features
of DSL is that
it is implemented using an infrastructure that already exists. DSL shares
copper twisted pair
lines typically used for telephone communication. However, only a small
portion of the
available bandwidth of the twisted pair line (0 to 4kHz) is used for Plain Old
Telephone
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
Service (POTS). DSL takes advantage of the available frequency spectrum from
4kHz to
approximately 1.lMHz for transmitting data.
[07] Asymmetric DSL (ADSL) is currently the most practical form of DSL
technology, and therefore the most widely implemented. ADSL is asymmetric in
that its
downstream (to a subscriber) capacity is larger than its upstream (from the
subscriber)
capacity. Typically, a Discrete Multi-tone (DMT) scheme is used. The spectrum
from 4kHz
to 1.lMHz is divided into 256 sub-channels, or tones, each having a bandwidth
of
4.3125kHz. Each sub-channel uses Quadrature Amplitude Modulation (QAM) to
carry 2 to
bits/QAM symbol.
10 (08] According to the ADSL International Telecommunications Union
(ITU) 6.992.2 standard, several phases occur in order to initialize a
communication link.
These phases include handshaking, transceiver training, channel analysis and
exchange.
[09] Handshaking is used for determining the nature and capabilities of
communication endpoints (such as an ADSL modem) and for indicating which
protocol will
15 be used for the remainder of the initialization. The ADSL modem, or
termination unit, at a
central office is referred to as an ATU-C. Similarly, the ADSL termination
unit at the
subscriber, or remote location, is referred to as the ATU-R.
[10] The signaling method used for the handshake interchange is designed
to be robust. Biphase shift keying (BPSK) modulation is used to modulate
multiple single-
tone sub-carriers, aII carrying the same data. Typically, the ATU-C and ATU-R
exchange a
message containing information about the endpoint type, frequency range, and
number of
DMT sub-carriers supported.
(1l] During transceiver training, the transceivers at each end of the line
acquire a DMT symbol stream, adjust receiver gain, perform symbol-timing
recovery, and
train any equalizers. There is an optional echo cancellation training step
that can also be
performed during this phase.
[12] During channel analysis, the transceivers exchange capability
information and perform detailed channel characterization. Both the ATU-R and
ATU-C
attempt to measure specific channel characteristics such as unusable sub-
Garners, loop
attenuation on a per sub-carrier basis, signal-to-noise ratios (SNRs), and any
other channel
impairments that would affect the potential transmitted bit rates. Based on
the discovered
channel characteristics, the ATU-C makes the first offer of the overall bit
rates and coding
overhead that will be used for the connection.
2
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
[13] The exchange phase sets the final overall transmission rates in both the
upstream and downstream directions for the connection. These final rates are
determined
based on calculated channel parameters measured during the channel analysis
phase, and are
not necessarily the same as the preliminary rates offered during that phase.
[14] ' Furthermore, the exchange phase sets forward error correction (FEC)
and interleaves configurations. Generally, the configurations are close to the
optimum bit
rate for the channels. Four carriers are used to modulate the bits of the
messages, each carrier
being loaded with 2 bits using quadrature phase shift key (QPSK) modulation.
[15] Since the ATU-C controls data rates, if the ATU-R cannot support any
of the offered rates, both terminals will return to the beginning of the
initialization process.
Otherwise the ATU-R responds with the rate it can support.
j16] The information transferred during the exchange is important for
establishing the communication between the ATU-C and the ATU-R. Therefore, the
same
bits are also modulated into a set of back-up tones for improving robustness.
The tone sets
used by 6.992.1 Annex A and 6.992.2 standards are provided below in TABLE 1.
[17J
Primary Set (Index Backup Set (Index
No.) No.)
Upstream 43, 44, 45, 46 91, 92, 93, 94
Downstream 10, 11, 12, 13 20, 21, 22, 23
[18] TABLE 1
[19] Further details of the above-described process are described below
with reference to FIGS. 1 and 2.
[20J Referring to FIG. 1, a system for implementing ADSL service is
illustrated generally by numeral 100. The system 100 comprises a central
office transceiver
(ATU-C) 102, a splitter I04, a twisted pair loop 106, and a remote transceiver
(ATU-R) 108.
The splitter 104 includes a high pass filter 110 and a low pass filter 112.
The ATU-C 102 is
coupled between a broadband network 114, such as the Internet, and the high
pass filter 110
of the splitter 104. The low pass filter 112 of the splitter 104 is coupled to
a narrowband
network 116 such as a General Switched Telephone Network (GSTN) or Integrated
Services
Digital Network (ISDN). Output from the high pass 110 and low pass filters 1
I2 are
combined and coupled with the twisted pair loop 106.
3
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
[21] The twisted pair loop is, in turn, coupled with a customer-premises
wiring network 118. The customer-premises wiring network 118 is coupled via a
low pass
filter 112 with narrowband network devices 120, such as telephones, voiceband
modems, and
ISDN terminals. The customer-premises wiring network 118 is further coupled to
the
ATU-R 108 via a high pass filter 110. The ATU-R 108 is further coupled to a
plurality of
service modules 122 via a home network 124.
[22] The system 100 illustrated in FIG. 1 operates by transferring data
between the ATU-C 102 and the ATU-R 108 on a frequency spectrum above that
used for the
narrowband devices 120. Therefore, the system 100 provides the service modules
122 access
to a high-speed network connection across the twisted pair loop 106, which is
an existing
infrastructure.
[23] Often, the twisted pair loop 106 is long, resulting in an increase in the
bit error ratio (BER) for the transmission. This is particularly important
during the exchange,
since the transmission parameters are established at this point. As it is
known, the BER for
QPSK modulation is
[24] BER; = Q( SNR 1 ) (1)
[25] and the overall BER over the 4 carriers (i.e. the average BER for the
decoded message) is
4
[26] BER = 1 ~, BER; (2)
4 ;_,
[27] The Message Error Rate (MER) for a given message of L bits is then
[2g] MER =1- (1- BER)' (3)
[29] The initialization message includes cyclic redundancy check (CRC)
bytes; therefore, L is the number of bits of the message the CRC bytes are
computed from.
Because the MER increases with L, one should consider the max value of L
(L",~) for the
initialization messages when evaluating the reliability of the messaging
scheme.
[30] The following messages and corresponding message sizes are
transferred during the exchange.
[31] Downstream
[32] The first group of messages includes C-RATES-RA, C-CRC-RA1,
C-MSG-RA, and C-CRC-RA2. The messages comprise 960 bits for C-RATES-RA, 16
bits
4
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
for C-CRC-RAl, 48 bits for C-MSG-RA, and 16 bits for C-CRC-RA2, yielding a
total of
1,040 bits or 130 Discrete Multi-tone (DMT) symbols.
[33] The second group of messages includes C-MSG2, C-CRC3,
C-RATES2, and C-CRC4. The messages comprise 32 bits for C-MSG2, 16 bits for C-
CRC3,
8 bits for C-RATES2, and 16 bits for C-CRC4, yielding a total of 72 bits, or 9
DMT symbols.
[34] The third group of messages includes C-B&G and C-CRCS. The
messages comprise 496 bits for C-B&G and I6 bits for C-CRCS, yielding a total
of 512 bits,
or 64 DMT symbols.
[35] Upstream
[36] The first group of messages includes R-RATES-RA, R-CRC-RA2,
R-MSG-RA, and R-CRC-RA1. The messages comprise 8 bits for R-RATES-RA, 16 bits
for
R-CRC-RA2, 80 bits for R-MSG-RA, and 16 bits for R-CRC-RAl, yielding a total
of 120
bits, or I S DMT symbols.
[37] The second group of messages includes R-MSG2, R-CRC3, R-
RATES2, and R-CRC4. The messages comprise 32 bits for R-MSG2, 16 bits for R-
CRC3, 8
bits for R-RATES2, and 16 bits for R-CRC4, yielding a total of 72 bits, or 9
DMT symbols.
[38] The third group of messages includes R-B&G and R-CRCS. The
messages comprise 4080 bits for R-B&G and 16 bits for R-CRCS, yielding a total
of 4096
bits, or 512 DMT symbols.
[39] Therefore, it can be seen that the maximum bit length for a
downstream message is L",~ = 960 for C-RATES-RA. For upstream, the maximum bit
length
is L",~ = 4080 for R-B&G.
[40] In order to have the MER<10-2, substituting the values ofL",~ from
Equation (3) results in:
[41] Downstream (L"~~ = 960) BER < 10-5
[42] Upstream (L",~ = 4080) BER < 2.5 ~ 10~
[43] In terms of the required signal-to-noise ratio (SNR) in the Garners, this
means the upstream messages require only a fraction of a dB higher SNR to
compensate for
the longer message.
[44] Referring to FIG. 2, a timing diagram for the exchange in accordance
with the state of the art is illustrated generally by numeral 200. Generally,
the nomenclature
for message transmission uses an "R-" prefix for indicating that the message
originated from
5
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
the ATU-R, and a "C-" prefix for indicating that the message originated from
the ATU-C.
The sequence of messages on the left side represents messages sent from the
ATU-C to the
ATU-R and the sequence of message on the right side represents messages sent
from the
ATU-R. For both sides, the message sequence begins at the top of the page.
[45] After C-MEDLEY 202 the ATU-C enters C-REVERB4 204 where it
waits for messages 206 from the ATU-R. The messages 206 include R-R.A.TES-RA,
R-CRC-
RA2, R-MSG-RA, and R-CRC-RA1. If the expected messages 206 are not received
within
6,000 symbols, the ATU-C times out and the initialization fails. If the ATU-C
receives the
expected messages in the allotted time, it remains in C-REVERB4 204 for at
least another 80
symbols before it enters C-SEGUE2 208. After C-SEGUE2 208, the ATU-C transmits
a
series of messages 210 to the ATU-R. These messages 210 include C-RATES-RA, C-
CRC-
RA1, C-MSG-RA, and C-CRC-RA2.
[46] Once the ATU-R has sent its messages 206 it enters R-REVERB-RA
212, where it waits for the messages 210 from the ATU-C. If the ATU-R does
receive the
messages 210 within 4,000 symbols, it times out and the initialization fails.
The ATU-C and
ATU-R use predefined tone indices for transmitting the messages R-RATES-RA, R-
CRC
RA2, R-MSG-RA, R-CRC-RA1, C-RATES-RA, C-CRC-RA1, C-MSG-RA, and
C-CRC-RA2. An additional set of tone indices is used to transmit these
messages as a
backup.
[47] Optimally, the receiver combines the bits carried in the two sets of tone
for improving reliability of the transmission. However, the signal-to-noise
ratio (SNR) in the
frequency band of the backup tone is much lower than that in the frequency
band of the
primary tone. Therefore, on long loops, especially for the downstream tones,
the backup set
of tones is essentially ineffective. In these cases, the bit error ratio (BER)
is determined by
the SNR on the primary set. Within a set, the highest BER within the four
carriers
determines the overall bit error rate on the message.
[48] As a result, increasing the number of sets of Garners has limited
benefits, since it does not guarantee best performance and further complicates
the messaging
protocol. Furthermore, as is often the case, the tone assigned by the
designated indices may
have a poor SNR, causing the initialization to fail.
[49] Therefore, there is a need for a messaging protocol that improves the
reliability of the messages transferred during the initialization. It is an
object of the present
invention to obviate or mitigate at least some of the above-mentioned
disadvantages.
6
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
BRIEF ST.JM11~IARY OF THE INVENTION
[50] In accordance with an aspect of the present invention, there is provided
a method for initializing a communication link between a first transceiver and
a second
transceiver for transferring data therebetween. The method comprises the steps
of analyzing
channel properties of a plurality of sub-channels within the communication
link, identifying a
predefined number of sub-channels having an anticipated highest performance
for
communication, communicating the identified sub-channels between the first and
second
transceivers, and transmitting information for initializing the communication
link using the
identified sub-channels.
BRIEF DESCRIPTION OF THE DRAWINGS
[51] An embodiment of the invention will now be described by way of
example only with reference to the following drawings in which:
[52] FIG. 1 is block diagram illustrating a typical system for providing
ADSL service (prior art);
[53] FIG. 2 is a block diagram illustrating the flow of data during the
exchange (prior art);
[54] FIG. 3 is a block diagram illustrating the flow of data during the
exchange in accordance with an embodiment of invention;
[55] FIG. 4a is a graph illustrating the performance of the initialization
process over a varying loop length with 24 ADSL NEXT and FEXT;
[56] FIG. 4b is a graph illustrating the performance of the initialization
process over a varying loop length with 24 DSL NEXT; and
(57] FIG. 5 is a block diagram of an ATU-C and an ATU-R that implement
the present invention.
DETAILED DESCRIPTION OF THE INVENTION
[58] For convenience, like numerals in the description refer to like
structures in the drawings.
[59] Referring to FIG. 3, a timing diagram for improving the reliability of
the exchange is illustrated generally by numeral 300. Additional ATU-C
transmissions
C-REVERBx 302, C-SEGUEx 304, and C-MSGx/C-CRCx 306 are inserted between
C-MEDLEY 202 and C-REVERB4 204. Similarly, additional ATU-R transmissions
7
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
R-REVERBx 308, R-SEGUEx 310, and R-MSGx/R-CRCx 312 are inserted between
R-MEDLEY 314 and R-REVERB4 316.
[60] The content of the messages C-MSGx and R-MSGx includes the
indices of four tones with the best SNR available. C-MSGx includes the indices
for upstream
communication and R-MSGx includes the indices for downstream communication.
Therefore, rather than use fixed indexes to transfer the messages, the indices
of the four tones
are selected adaptively, in accordance with an estimated line SNR.
(61] The indices of the four tones are selected by the ATU-C and ATU-R to
correspond to tones with the best SNRs. The SNR estimate is available at the
exchange
because it takes place after both C-MEDLEY and R-MEDLEY (during channel
analysis).
During C-MEDLEY an estimate of the downstream SNR is determined at the ATU-R.
The
ATU-R determines the indices of the tones having the four highest SNRs for
downstream
communication and compiles them into R-MSGx. Similarly, during R-MEDLEY an
estimate
of the upstream SNR is determined at the ATU-C. The ATU-C determines the
indices of the
tones having the four highest SNRs for upstream communication and compiles
them into C-
MSGx. The sets of four indices, that is C-MSGx and R-MSGx, are exchanged
between the
ATU-R and the ATU-C using a more reliable 1-bit per symbol modulation.
(62] . The format of R-MSGx and C-MSGx is describes as follows. The
message comprises a prefix, a first Garner index, a second carrier index, a
third carrier index,
and a fourth carrier index. The prefix is four bytes and each of the carrier
indices is one byte
as illustrated in Table 2 below.
[63]
Prefix Cer Carner Carrier Carrier
index index #2 index #3 index #4
#1
Number 4 1 1 1 1
of
bytes
[64] TABLE 2
[65] The prefix is {01010101 01010101 01010101 01010101)2. The Garner
index fields contain the four carrier indexes with the best SNR in decreasing
order.
Therefore, the SNR of carrier index #1 is greater than or equal to the SNR of
Garner index #2,
which is greater than or equal to the SNR of carrier index #3, which is
greater than or equal to
8
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
the SNR of carrier index #4. The byte for each carrier index is the binary
representation of
the selected index.
[66] The message is followed by a 16-bit CRC that is transmitted using the
same 1-bit/symbol modulation format. Thus, 80 DMT symbols axe required for
transmitting
each of the 80-bit C-MSGx/C-CRCx message and 80-bit R-MSGxIR-CRCx message.
[67] Referring to FIG. 4a and FIG. 4b the performance of the messaging
scheme described herein is compared to that currently in use, with respect to
the MER of
C-RATES-RA. FIGS. 4a and 4b refer to two different cross talk scenarios. FIG.
4a has 24
ADSL near end cross tally (NEXT) and far end cross talk (FEXT). FIG. 4b has 24
DSL
NEXT. The vertical axis represents an increase in the MER. The horizontal axis
represents
an increase in loop length. The loop lengths are selected in order to allow
for a non-zero net
throughput in presence of a coding scheme. In particular, when Reed Solomon
(RS) FEC
only is used, a non-zero throughput is guaranteed for the l7kft and l8kft
loops in both FIGS.
4a and 4b. When Trellis and RS are used, reach can be extended to l9kft with
24 ADSL
NEXT and FEXT (FIG. 4a) and to 20kft with 24 ADSL NEXT (FIG. 4b).
[68] As illustrated in both FIGS. 4a and 4b, for these conditions the current
standard messaging scheme is inadequate, since the MER approaches 1 for these
loops.
Therefore, even though the channel allows a non-zero net data rate, the
unreliability of the
messages does not allow the link to activate. However, the messaging scheme
described in
the preferred embodiment is sufficiently reliable for all of these cases.
Furthermore, as a
result of the improved reliability of the selected set of carriers, only one
carrier set is
required.
[69] FIG. 5 shows an ATU-C 510 and an ATU-R 520 that implement the
present invention. The conventional features in the figure generally
correspond to FIG. 1 and
are not further detailed. The ATU-C 510 includes a processor 512, and the ATU-
R 520
includes a processor 522. In general, the processors 512, 522 control the ATU-
C S 10 and
ATU-R 520 to implement the above-described messaging scheme. The processors S
12, 522
may be implemented as specialized circuitry (e.g., an application-specific
integrated circuit),
a field-programmable gate array, as a general processor that is controlled by
software
(including microcode), or as a combination of two or more of these
implementations.
[70] In yet an alternate embodiment, each transceiver sends a stream of bits
as numerous as the number of the tones capable of being received. Each bit
corresponds to a
tone. If a bit is set to 1 then its associated tone is to be used during for
transmitting the
messages that help establish the communications link. For example, the ATU-C
transmits
9
CA 02436015 2003-07-23
WO 02/09330 PCT/USO1/23370
messages that include C-MSG-RA and C-RATES-RA. The ATU-R transmits messages
that
include R-MSG-RA and R-RATES-RA. If the bit is set to zero, its associated
tone is not
used for modulating the messages.
[71] In all of the embodiments described above, it is possible to use greater
or fewer than four tones for communicating the message as will be apparent to
a person
skilled in the art. Although the invention has been described with reference
to certain
specific embodiments, various modifications thereof will be apparent to those
skilled in the
art without departing from the spirit and scope of the invention as outlined
in the claims
appended hereto and their equivalents.
10
