Pulse or digital communications – Systems using alternating or pulsating current – Plural channels for transmission of a single pulse train
Patent
1993-03-22
1995-09-05
Chin, Stephen
Pulse or digital communications
Systems using alternating or pulsating current
Plural channels for transmission of a single pulse train
371 371, 371 372, 371 377, 371 46, 375295, 375340, 375354, H04L 512, H04L 2302
Patent
active
054485924
DESCRIPTION:
BRIEF SUMMARY
RELATED APPLICATIONS
This application is related to commonly assigned U.S. patent application Ser. No. 08/122,525 filed Dec. 1, 1993 entitled FRAME SYNCHRONISATION FOR QAM and naming John D. Brownlie and Richard G. Willians as inventors.
FIELD OF THE INVENTION
The present invention relates to phase amplitude modulation particularly, though not exclusively, using block coding.
BACKGROUND OF THE INVENTION
In digital phase amplitude modulation, the modulated signal consists of a sequence of symbols in each of which a carrier has a selected phase and amplitude. Only certain phase/amplitude combinations are permitted; these combinations may be plotted on a diagram with in-phase and quadrature axes to form a pattern; the set of allowable points in this pattern is commonly referred to as a constellation. If for example, one has a 16 point constellation, it is a simple matter to modulate the signal with a 4-bit word to be transmitted by regarding each point as associated with a respective one of the 16 possible combinations of four bits.
It has been shown however that by using a larger constellation (e.g. 32 points) and a suitable coding of the 4 bits, the resultant inherent redundancy in the modulated symbol sequence can be exploited by a soft decision decoder to improve the reliability of decoding in the presence of noise to an extent which exceeds the degradation caused by the larger numbers of points and results in a net coding gain. Coding gain is defined as the difference (in dB) between the signal-to-noise ratio that a coded scheme needs to operate at a particular error rate and that needed by the equivalent uncoded system.
One method of achieving coding gain is by the use of convolutional coding; here however we are concerned primarily with block coding, though the synchronization arrangements to be described are not limited to such cases.
It is inherent in block coding that output symbols are generated on a block by block basis. Thus if a 16-point constellation is to be used for modulation at a rate of 3 bits/symbol, then the coding process needs to produce 4n bits (for selecting points in the constellation) for every 3n data bits received (where n is the length of a block of symbols).
Consider now the concept of set partitioning: the signal constellation is progressively partitioned into subsets having increasing minimum Euclidean distance between the points of each subset; .DELTA..sub.0 <.DELTA..sub.1 <.DELTA..sub.2 . . . as illustrated in FIG. 1 for 16 point quadrature amplitude modulation. The Euclidean distance is simply the linear distance on the phase diagram between adjacent points of a subset; thus, assuming the sixteen points are on a unit grid, ##EQU1##
This distance is significant in that it is a measure of the capability of a hard decision decoder to discriminate between points in the subset in the presence of noise. By labelling each partition with a binary digit as shown we form a partition tree. In FIG. 1 the label for each point is constructed from the bits labelling the partitions needed to reach it; it is convenient to write the bit corresponding to the first partition level as the rightmost bit of the point label and so on. We then see that points whose labels first differ (when read from right to left) in the ith position (where the rightmost bit is the first position) are a Euclidean distance of at least .DELTA..sub.i-1 apart.
The concept of set partitioning in this manner is discussed in G. Ungerboeck's paper "Channel coding with multilevel/phase signals", IEEE Trans IT-28, pp 55-67, January 1982.
The other primary consideration in the coding process is the Hamming distance of the code or codes employed. FIG. 2 illustrates the coding process, in the form of an array. Of a total of .SIGMA.k.sub.i input bits, k.sub.1 bits are coded by means of an (N, k.sub.1, d.sub.1) code to form a first row of bits a.sub.11 . . . a.sub.1N, k.sub.2 bits are coded by means of an (N, k.sub.2, d.sub.2) code to form the second row of bits a.sub.21 . . . a.sub.2N, and so on. The array has N c
REFERENCES:
patent: 3955141 (1976-05-01), Lyon et al.
patent: 4586182 (1986-04-01), Gallager
patent: 4597090 (1986-06-01), Forney, Jr.
patent: 4683578 (1987-07-01), Betts et al.
patent: 4899367 (1990-02-01), Sampei
Forney, Jr., "Multidimensional Constellations--Part I: Introduction, Figures of Merit, and Generalized Cross Constellations", IEEE Jornal on Selected Areas in Communications, vol. SAC-7, No. 6, Aug. 1989, New York, pp. 877-892.
CCITT Study Group Papers, "Constellation and Symbol Mapping", D133, pp. 1-7, Genova, 29 Oct.-6 Nov. 1991.
CCITT Study Group Papers, "Additional Details on AT&T's Candidate Modem for V.fast", D 157, pp. 1-6, 29 Oct.-6 Nov. 1991.
PCT International Search Report, European Patent Office, completed Oct. 24, 1991.
Brownlie John D.
Williams Richard G. C.
British Telecommunications public limited company
Chin Stephen
Le Amanda T.
LandOfFree
Coded QAM system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Coded QAM system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Coded QAM system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-479118