Bi-level framing structure for improved efficiency DSL over...

Details Bi-level framing structure for improved efficiency DSL over... Bi-level framing structure for improved efficiency DSL over...

1999-02-17

2002-09-10

Hsu, Alpus H. (Department: 2665)

Multiplex communications

Communication techniques for information carried in plural...

Adaptive

C370S474000, C370S476000, C714S701000, C714S786000

Reexamination Certificate

active

06449288

ABSTRACT:

FIELD OF THE INVENTION
This invention relates to Digital-Subscriber Lines (DSL) systems, and more particularly to framing structures for lower line rates.
BACKGROUND OF THE INVENTION
Telephone systems are increasingly being used to carry data traffic as well as voice calls. While analog modems were sufficiently useful for lower data rates, graphics, audio, and video data transfers have increased data-rate requirements. Integrated Services Digital Network (ISDN) and more recently Digital-Subscriber Line (DSL) including asymmetric DSL (ADSL) have been developed to provide higher data rates.
DSL systems have been developed that carry data on many carriers at the same time. The carriers are modulated in phase and amplitude to carry the data signals. Since multiple carriers separated in frequency are used, this technique is known as discrete multi-tone (DMT).
The data to be transmitted over the phone line is first framed by adding sync bytes and error correction bytes and blocking into symbols which are generated at an average 4 KHz rate. Based on the line characteristics a mapper assigns a different number of bits to each tone used. A constellation encoder modulates the various carriers with the data bits to produce a frequency domain signal. This signal is then converted from the frequency domain to the time domain by an inverse fast-Fourier transform (IFFT). This time domain signal is then converted from digital to analog voltages that drive the physical phone line (copper twisted pair).
Various other encoding techniques such as trellis encoding can be inserted before the IFFT. The actual signal on the phone line bears little resemblance to the user data once the various transformations and encodings are performed. Nevertheless, the data is arranged into frames before the transform and encoding, and the received data is also arranged in frames once transforms and decodings are completed.
FIG. 1A
shows a high-rate DSL system. A user data stream is framed with sync and error-correction bytes to produce a 1.536 mega-bits-per-second (1.536 Mbits) stream. This information stream is divided into many frequency bins and input to IFFT
10
. IFFT
10
converts a set of frequency bins into a series of time points every 250 &mgr;sec in response to the 4 KHz system clock. Since a large number of frequency bins are used for the high user-data rate, the time-points output by IFFT
10
represents a symbol with many data bytes.
Symbol
12
represents 48 bytes of information transmitted over the phone line. A new symbol
12
is output by IFFT
10
for every period of the 4 KHz clock. Thus the data rate transmitted over the phone line, the line rate, is 48 bytes×4 KHz=192 Bytes/sec, or 1.536 Mbits.
Telephone systems have traditionally used 4-8 KHz system clocks, and occasionally 4 KHz framing clocks. The IFFT is also clocked at the 4 KHz rate, outputting symbols at the 4 KHz rate. Framing is defined based on this 4 KHz physical layer. Thus each frame contains 48 bytes for the high-rate DSL system. Such a DSL system is being proposed for an International Telecommunications Union (ITU) standard known as G.Lite.
FIG. 1B
shows a low-rate DSL system. Since the existing copper-pair telephone wires are used for DSL, the quality of the lines varies. Some customers may have poor-quality or longer lines that cannot support the high-rate DSL system of FIG.
1
A. The physical lines of
FIG. 1B
support a line rate of only 64 Kbit. When the 4 KHz system clock is used, and symbols are output by IFFT
10
at the 4 KHz rate, each symbol
12
represents only 2 bytes (16 bits).
FIG. 2A
shows a frame for a high-rate DSL system. Mux data frame
14
begins with one sync byte S, leaving 47 bytes for user data, the payload bytes P. The amount of the channel used for sync overhead is only 1/46.
FIG. 2B
shows a frame for a low-rate DSL system. Mux data frame
14
begins with one sync byte S. Since each symbol is only 3 bytes, only two bytes are available for user data, payload bytes P. One-third of the channel is used for sync overhead. Thus framing based on the 4 KHz physical layer is inefficient at low line rates.
Error correction is often employed in DSL systems. Reed-Solomon (RS) forward-error-correction (FEC) bytes can be appended to a series of mux data frames to allow for detection and correction of errors within the frames. The FEC bytes together with the mux data frames form a RS codeword.
FIG. 3A
shows a RS codeword using high-rate mux data frames. Four mux data frames
14
are provided with error correction by RS FEC bytes
16
. The number of bytes in FEC bytes
16
can be increased to improve error correction ability, but in this example one FEC byte is provided for each mux data frame
14
. Thus FEC bytes
16
includes 4 FEC bytes.
The channel overhead is relatively small. With 4 mux data frames, 4 sync bytes and 4 FEC bytes are used, for a total of 8 overhead bytes. The number of user payload bytes is 62×4, or 248 bytes.
FIG. 3B
shows a RS codeword using low-rate mux data frames. Four mux data frames
14
are provided with error correction by RS FEC bytes
16
. One FEC byte is still provided for each mux data frame
14
. Thus FEC bytes
16
includes 4 FEC bytes.
The channel overhead is quite high. With 4 mux data frames, 4 sync bytes and 4 FEC bytes are used, for a total of 8 overhead bytes. However, the number of user payload bytes is just 4 bytes. Thus ⅔'s of the channel is used for overhead.
FIG. 4
shows a framing structure for DSL. The G.Lite framing structure is based on the 4 KHz physical layer. User data and sync bytes are multiplexed into mux data frames
14
. Each mux data frame
14
has 1 sync byte and N
p
user payload bytes, for a total of K
i
bytes. Mux data frames
14
are arranged together into RS codewords. Each RS codeword
20
contains S mux data frames
14
. The RS codeword ends with R
i
FEC byte
16
.
The RS codewords
20
are then sent to the IFFT to be transformed into symbols for transmission over the phone line. The IFFT operates at a 4-KHz rate, continuously outputting one symbol or 4-KHz frame
22
every 250 &mgr;sec. The stream of 4-KHz frames
22
from the IFFT is converted to analog voltages to drive the phone line as output stream
24
.
The G.Lite standard requires that the number of 4-KHz frames
22
in a RS codeword is equal to the number of mux data frames
14
in the same RS codeword. Thus each 4-KHz frame
22
is slightly longer than each mux data frame
14
. The additional length is due to the FEC bytes
16
that must be allocated among the 4-KHz frames
22
. This number of frames, either mux data frames
14
or 4-KHz frames
22
, is known as parameter S. Each 4-KHz frame
22
is thus R
i
/S bytes longer than each mux data frame
14
.
The values of R
i
/S are further restricted to integer values. Integer values of R
i
/S ensures that the number of bytes per 4-KHz frame is also integer as the number of bytes per 4-KHz frame is equal to K
i
+R
i
/S . . . This simplifies data paths in the DSL system.
The restriction for integer values of R
i
/S ensures that at least as many FEC bytes as there are 4-KHz frames. Also, one sync byte is contained in each mux data frame
14
and thus there are as many sync bytes as 4-KHz frames. For high-rate systems, an overhead of 2 bytes per 4-KHz frame is small. However, for low-rate systems, this 2-byte-per frame overhead is great. When each 4-KHz frame has only 3 bytes, such as for 96 Kbits, over 66% of the channel is spent on overhead. Thus high-rate DSL systems do not scale well to lower rate systems. Bandwidth efficiency is especially poor for low line rates.
It is desirable to provide DSL at both high and low line rates. Then a single DSL board or chip set could be used for both high-rate and low-rate applications. A more efficient framing structure for low-line rates is desired. It is desired to continue to use the 4-KHz system clock for physical framing, but to increase the bandwidth available for user payload bytes at low line rates. It is desired to extend the framing structure for high

Affiliated with

Also associated with

Rating

Bi-level framing structure for improved efficiency DSL over... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Bi-level framing structure for improved efficiency DSL over..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Bi-level framing structure for improved efficiency DSL over... will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2860291