Telecommunications Industry Association
09_
(TIA)
Orange County, CA, 4-5 March 1997
TR30.1 Ad-hoc _Irvine
COMMITTEE CONTRIBUTION
Technical Committee TR30 Meetings
SOURCE:
U.S. Robotics
CONTRIBUTOR:
Dale Walsh
TITLE:
Multiple Modulus Conversion for Robbed Bit Signaling
Channels
PROJECT:
PN:3838
DISTRIBUTION:
Meeting attendees
__________________________________
Copyright Statement
Contributor's company grants an irrevocable royalty-free and compensation-free license to the
Telecommunications Industry Association (TIA) to use the text of this contribution in any TIA
publication.
Intellectual Property Statement
The individual preparing this contribution knows of patents, the use of which, may be essential to a
standard resulting in whole or part from this contribution.
Multiple Modulus Conversion for Robbed Bit Signaling
Channels
Introduction
This is a description of a Multiple Modulus Conversion (MMC) method for signaling on Robbed Bit Signaling
channels. First, motivation for the method is given as background, followed by a description of MMC as
applied to RBS channels.
Background
This method relates to data communications used on Digital Telephone Networks (DTN’s). In particular, a data
distribution system might consist of a digital source directly connected to the DTN, without any
Analog-to-Digital Converter (ADC) or Digital-to-Analog Converter (DAC). The telephone network routes digital
signals from this data source to the client’s local subscriber loop without any intermediary analog facilities, such
that the only analog portion of the link from server to client is the client’s local loop plus the associated analog
electronics at both ends of the loop, and the only Digital-to-Analog Converter in the transmission path is the one
at the telephone company’s end of the client’s subscriber loop. In such a system, digital data can be converted
into PCM codes, and fed to the DTN as 8-bit bytes (octets) at the networks clock rate of 8 kHz. The DTN’s DAC
converts each byte to one of 255 analog levels, referred to as points, which is sent over the client’s subscriber
loop and received by a device (i.e., a modem) at the client’s location. These analog levels, which we will call
points, are non-uniformly spaced and follow a logarithmic curve called  -law in North America and A-law in
Europe. There are 256 points defined by 8-bit coding. But due to the crowding of the smaller levels and noise
sources, not all can be used.
MCC is designed to improve data communication performance on certain network connections that utilize a
supervisory signaling technique called Robbed Bit Signaling (RBS). On RBS links, the least significant bit (lsb)
of the PCM code is usurped , or “robbed”, by the network periodically and used to convey control information.
The PCM codes are framed in groups of 12, and the lsb is robbed every 6th and 12th time slot. The two robbed
bits are used to form a 2 bit code for signaling control and status information between network equipment. The
customer whose data is carried by the network loses use of the lsb during those time slots. Ordinarily, these
channels are used for voice communications, and the bit robbing merely increases quantization noise of the
effected time slots. This effect is barely perceptible to the human ear. However, the above described data
transmission system will in general suffer an unacceptable level of data errors on RBS circuits unless some
coding method is devised to deal with the robbed bit problem.
The straightforward method for maintaining acceptable data error rates in the presence of RBS is to reduce the
number of levels encoded by 1/2, such that the ambiguity introduced during the 6th and 12th time slots’ lsb’s is
resolved at the receiver by interpreting two different received levels as the same transmitted code.
Unfortunately, this technique has the disadvantage of sacrificing one bit per symbol, or 8 kbits/second. A more
sophisticated and less costly technique is to encode the data such that the number of encoding levels is reduced
only during the 6th and 12 time slots within a frame. This means that the encoder needs to know the position of
the robbed bit time slots, and needs to have a means of mapping fewer points during those slots than during the
non-robbed slots. MCC is a technique to encode and map the data in a unified manner that minimizes the data
rate penalty due to RBS.
Description
The system described above can be viewed as a Multilevel Pulse Amplitude Modulation (MPAM) system; that
is, bits are sent via one of 256 points. These points are symmetric about zero; i.e., there are 128 positive and
128 negative levels, including two encodings of zero. MMC only defines symbol magnitudes. Sign-bit
assignments are treated separately.
Because there are 254 points (not including zero), the maximum number of bits that can be sent per signaling
interval (symbol) is just under 8 bits. To provide flexible data rates, it is necessary to send a fractional number
of bits per symbol. In addition, it is desired to mitigate the effects if RBS, such that the highest possible data
rate is achieved. Both these requirements can be meet with MMC.
An n symbol frame is defined where each symbol i within the frame is a set of
symbol position in the frame can use a different set of points.
Constellations consisting of
M i points. Note that each
M i points are selected to meet the following three criteria:
n
1.
2 K   M i , where K is the number of user bits to be transmitted in an n symbol frame.
i1
2.
Points in constellations used in RBS intervals, should be chosen such that the lsbs in their corresponding
 -law codes are least effected by RBS. (Note these constellations will always be smaller than the ones used
in non-RBS intervals, since less data is sent during this time.)
3.
The probability of symbol error is minimized and balanced. Minimizing the probability of symbol error is
accomplished by choosing points with maximal spacing, while constraining the points to have some specified
average power. In addition, the number of points with minimum spacing should be minimized (smallest
number of nearest neighbors). Balancing the symbol error probability, means that mutually, the M i should
have about the same symbol error rate, since then the overall error rate (frame or block error rate, for
example) will not be dominated by the symbol error rate of any one constellation.
The encoder collects K bits to send during an n symbol frame. MMC allows the K bits to be mapped to
n indices, representing K / n bits/symbol (not necessarily an integer).
The next step in the conversion process is to represent the K bits as an integer, R ,
R  bo  b1 2  b2 22 bK 1 2 K 1 ,
where bo is the lsb and
represented as
bK1 is the msb of K bits submitted to the K -bit frame. The number R may then be
R  r1  r2 M1  r3 M1 M2 rn M1 M2 Mn1.
The result (or output) of the MMC process are the values of
ri , i  1,2,..., n , where the ri are constrained to
0  ri  Mi . The value ri is the index of the symbol to be sent in symbol internal i. The
n values of ri correspond to the n analog levels that will be produced by the network’s D/A converter.
the interval
Other Publications
Modulus Converter for Fractional Rate Encoding, US Patent 5,103,227, William L. Betts, 4/7/92.
Data Communication System with Data Rate Throttling, US Patent 5,280,503, William L. Betts and Gordon
Bremer, 1/18/94.