WO1997016910A1

WO1997016910A1 - Method and apparatus for the provision and reception of symbols

Info

Publication number: WO1997016910A1
Application number: PCT/US1996/016826
Authority: WO
Inventors: Gregory C. White; Stephen P. Emeott; John G. Ronk; Patrick J. Doran
Original assignee: Motorola Inc.
Priority date: 1995-10-31
Filing date: 1996-10-18
Publication date: 1997-05-09
Also published as: GB9713516D0; BR9607557A; CN1166900A; AR004249A1; IL119467A0; FR2740634A1; MX9704951A; GB2312814A

Abstract

A symbol constellation is logically divided into a plurality of multi-symbol subsets (201-204). Additionally, a series of primary bits and a series of secondary bits are provided. At least a first selected primary bit is error control encoded (303) to provide an M bit pre-symbol (603) that uniquely corresponds to a multi-symbol subset (602). The M bit pre-symbol is modulated (304) by at least a first selected secondary bit to provide an N bit symbol (605) that uniquely corresponds to a symbol included in the multi-symbol subset. Recovered primary bits are determined by comparing a received symbol (609) with only one symbol in each of the multi-symbol subsets. Additionally, recovered secondary bits are determined by comparing the received symbol with predetermined decision boundaries (622). In this manner, varying degrees of error protection are provided allowing for a wider variety of coding rates.

Description

METHOD AND APPARATUS FOR THE PROVISION AND RECEPTION

OF SYMBOLS

Field of the Invention

The present invention relates generally to wireless digital communication devices and, in particular, to a method and apparatus for the provision and reception of symbols.

Background of the Invention

In the area of wireless digital communication devices, the use of error correction is known to combat transmission channel errors. In particular, trellis coded modulation (TCM) is a known method for performing error correction in which digital data (i.e., binary bits) is processed using known error correction techniques to produce symbols for transmission.

It is further known in the art to provide varying levels of error protection for different sets of voice coding parameters.

In particular, parameters that have a high degree of sensitivity to errors (i.e., they have a significant negative impact on recovered audio quality if received incorrectly) are given a high degree of error protection, whereas those parameters that have a low degree of sensitivity to errors (i.e., they have little or no impact on audio quality if received incorrectly) are given a lower degree of protection, if any.

Although TCM can be used advantageously in systems requiring varying levels of error protection, TCM is sometimes limited in the number of code rates available for use with a given modulation type. An error correction code rate is typically described as a ratio of k/n where k is the number of bits to be encoded and n is the number of bits comprising the resulting symbol. For example, a rate 1/2 code produces a symbol specified by two coded bits for every one bit input, a rate 2/3 code produces a symbol specified by three coded bits for every two bits input, etc. In general, as the code ratio decreases/increases, the correcting power of the code is increased/reduced. Table 1 illustrates the limited number of TCM code rates for various quadrature amplitude modulation (QAM) and phase shift keying (PSK) modulation schemes.

Modulation Code Rates

8PSK 2/3, (1/3 non-Ungerboeck)

16QAM 3/4, (2/4, 1/4 non-Ungerboeck)

64QAM 5/6, (4/6, 3/6, 2/6, 1 /6 non-Ungerboeck)

Table 1.

As shown in Table 1 , only one Ungerboeck code rate (i.e., having a ratio of N/N+1 ) is available for each modulation type. Although the non-Ungerboeck code rates are possible, they are difficult to design to achieve proper error protection performance and are typically not used in practice. Given the limited number of code rates available using TCM, designers are often faced with the choice protecting a few bits with high degrees of protection, or more bits with lesser degrees of protection. However, such alternatives do not always provide adequate error correction performance in systems requiring varying levels of protection, e.g., voice systems.

Therefore, it would be advantageous to provide an approach that allows a wider range of TCM coding rates to be chosen in systems that require varying levels of error protection. Brief Description of the Drawings

FIG. 1 is a schematic illustration of a prior art method for deriving symbols.

FIG. 2 is a schematic illustration of a first embodiment for deriving symbols in accordance with the present invention.

FIG. 3 is a flowchart of a method for providing a symbol based on multiple series of bits in accordance with the present invention.

FIG. 4 is a block diagram of a wireless digital communication device in accordance with the present invention.

FIG. 5 is a flowchart of a method for recovering primary and secondary bits based on a received symbol in accordance with the present invention.

FIG. 6 is a block diagram illustrating an exemplary encoding and decoding of a symbol in accordance with the present invention.

Description of a Preferred Embodiment

The present invention provides a method and apparatus for the provision of symbols based on TCM which allows multiple series of bits to be error protected with varying degrees of protection. The method and apparatus can be incorporated into a wireless digital communication device.

A symbol constellation is logically divided into a plurality of multi-symbol subsets. Additionally, a series of primary bits, requiring a first degree of protection, and a series of secondary bits, requiring a second degree of protection less than the first degree of protection, are provided. At least a first selected primary bit is error control encoded to provide, in general, an M bit pre-symbol, which M bit pre-symbol uniquely corresponds to a multi-symbol subset of the plurality of multi-symbol subsets. The M bit pre- symbol is modulated by at least a first selected secondary bit to provide an N bit symbol (where N > M) that uniquely corresponds to a symbol included in the multi-symbol subset. In the context of the present invention, modulation refers to the selection of a particular symbol within a region defined by the pre-symbol.

Upon reception, the N bit symbol is decoded by comparing it with only one symbol in each of the multi-symbol subsets to produce recovered primary bits. Additionally, the N bit symbol is compared against predetermined decision boundaries to produce recovered secondary bits. In this manner, multiple series of bits can be error protected with varying degrees of protection. Additionally, a wider variety of coding rates are provided.

The present invention can be more fully described with reference to FIGS. 1 -6. FIG. 1 is a schematic illustration of a prior art method for deriving symbols. Using this method, each bit of a series of bits receives approximately the same degree of protection from channel errors. In the example shown, a first input bit is coded with a convolutional code, and then mapped to one out of four initial 2-bit symbols where the Euclidean distances between the initial symbols are relatively small (left l-Q plane). A chosen symbol 101 is subsequently shifted, responsive to two other unencoded bits, to produce one of four possible 4-bit symbols 102-105 such that the Euclidean distances between the four possible symbols are greater than the Euclidean distances between the initial symbols. The method illustrated in FIG. 1 uses Euclidean distance between the symbols to provide error protection to the two unencoded bits, while using a convolutional code to provide error protection to the first input bit. This has a net effect of providing a similar level of protection to the both the first input bit and the two unencoded input bits. At a decoder, received symbols, generated in the manner described, are compared against all possible candidate symbols to determine the transmitted sequence of input bits. FIG. 2 is a schematic illustration of a first embodiment for deriving symbols in accordance with the present invention. In the example shown in FIG. 2, a primary bit is encoded, using a convolutional or trellis encoder, to produce a 2-bit pre- symbol that uniquely corresponds with a region (or multi- symbol subset) 201 -204. This process is described in greater detail below. (The present invention contemplates that a symbol constellation is partitioned into at least two multi- symbol subsets. For example, a 16QAM symbol constellation can be divided into two multi-symbol subsets comprising eight symbols each, or four multi-symbol subsets comprising four symbols each, or eight multi-symbol subsets comprising two symbols each. Alternatively, it is not necessary that every symbol in the constallation be a member of a multi-symbol subset. For example, eight symbols in a 16QAM symbol constallation could be divided into four multi-symbol subsets comprising two symbols each. The other eight symbols would not be used. Which symbols are included in which multi- symbol subset is a matter of design choice.)

In effect, the present invention assigns encoder state transitions to multi-symbol subsets, rather than individual symbols as in prior art methods. In a preferred embodiment, the multi-symbol subsets are chosen so that for each multi- symbol subset, the average Euclidean distance between any symbol in the multi-symbol subset to the nearest symbol in all other multi-symbol subsets is maximized. In this manner, coded bits not only benefit from the encoding, but also from a greater degree of Euclidean distance.

Once a pre-symbol corresponding to a multi-symbol subset is selected, two unencoded bits are used to select a particular 4-bit (i.e. N=M+2) symbol 205-208 from within the selected multi-symbol subset. The trade-off between the method of FIG. 2 and that of FIG. 1 is that the unencoded bits do not receive the benefit of greater Euclidean separation. Furthermore, the decoding procedure, as described later, is less complex.

FIG. 3 is a flowchart of a method for providing a symbol based on multiple series of bits in accordance with the present invention. At step 301 , series of primary and secondary bits are provided. Alternatively, more than one series of primary and/or secondary bits can be provided. Although these series of bits can generally comprise any type of information, in a preferred embodiment they comprise voice coding parameters having varying degrees of error sensitivity. In an alternate embodiment, the series of bits comprise image coding parameters having varying degrees of error sensitivity. In the context of the present invention, the one or more primary series of bits are assumed to have a high degree of error sensitivity and are therefore intended to receive strong protection; the one or more series of secondary bits are assumed to have a low degree of error sensitivity and are therefore intended to receive weak protection.

At step 302, at least one bit from the series of primary bits and at least one bit from the series of secondary bits are selected to produce at least a first selected primary bit and at least a first selected secondary bit. This selection can be done by taking the next available bit in each series of bits, or by selecting, out of sequential order, predetermined bits from each series of bits. Additionally, if there are more than one series of primary and/or secondary bits, the at least a first selected primary bit and the at least a first selected secondary bit can be selected from the multiple primary and secondary series of bits, respectively.

At step 303, the at least a first selected primary bit is encoded to provide an M bit pre-symbol that uniquely corresponds to a multi-symbol subset. In a preferred embodiment, the at least a first selected primary bit is encoded using a known trellis code to provide the M bit pre- symbol. In an alternate embodiment, a known convolutional code is used to provide M bits which are mapped to the M bit pre-symbol. As known in the art, both trellis and convolutional encoders map state transitions to candidate symbols. In this manner, the state transitions produced by a series of input bits produce an allowed sequence of symbols. In contrast, the present invention maps the state transitions produced by the input bits to multi-symbol subsets (or regions) that are identified by the M bit pre-symbols. If the multi-symbol subsets are chosen so as to maximize Euclidean distance between each multi-symbol subset, the input bits used to produce the M bit pre-symbols benefit not only from the redundancy provided by the encoder, but also from the

Euclidean distance provided by the multi-symbol subsets.

At step 304, the M bit pre-symbol is modulated by the at least a first selected secondary bit to provide an N bit symbol, wherein N > M and K=N-M. In practice, each N bit symbol is represented as an N bit symbol index. The "modulation" step is equivalent to choosing a specific symbol from within the multi-symbol subsets (or region) that uniquely corresponds to the M bit pre-symbol. As described above, each multi-symbol subset comprises a logically grouped collection of 2^K symbols, which symbols can be indexed 0 through 2^K-1 . (The term

"region" used to refer to a multi-symbol subset arises because the symbols included in each multi-symbol subset are typically chosen to be proximate to each other.) The selection (or modulation) process can be performed by directly mapping the K secondary bits comprising the at least a first selected secondary bit to the proper N bit symbol index in the given multi-symbol subset. In an alternate embodiment the selection is performed using a second encoder (trellis or convolutional). At step 305, the N-bit symbol can be optionally transmitted via a suitable communication resource, e.g., a radio frequency (RF) carrier. Regardless of step 305, it is determined at step 306 if all bits included in the series of primary and secondary bits have been encoded, modulated, and transmitted, then the process is complete. If there are bits remaining to be encoded, modulated, and optionally transmitted, the process is repeated at step 302. In a preferred embodiment, the series of primary and secondary bits are selected such that the bits in each series are exhausted simultaneously, e.g., all bits comprising a single coded voice frame.

FIG. 4 is a block diagram of a wireless digital communication device 400 in accordance with the present invention. The wireless digital communication device 400 comprises a voice coder 401 , an encoder 402, a pre-symbol modulator 403, and a wireless transmitter 404. Although not shown, in a preferred embodiment, one or more processing devices (e.g., a microprocessor and/or a digital signal processor) and memory devices (e.g., random-access and/or read-only memory) are included in the wireless digital communication device 400 and are used to implement, as software algorithms, the methods described by the present invention.

The voice coder 401 implements a digital speech compression algorithm that outputs at least two channels of digital voice coding parameters that characterize an input speech waveform (not shown). For example, the voice coder may implement the so-called vector-sum excited linear prediction coder (VSELP) or the so-called improved multi-band excitation coder (IMBE). A first channel 406 includes a series of voice coding parameters having a high degree of error sensitivity (referred to as the series of primary bits above) and a second channel 407 includes a series of voice coding parameters having a low degree of error sensitivity (referred to as the series of secondary bits above).

The first channel 406 is sent to the encoder 402. The encoder 402 performs the encoding process described in step 303 above, resulting in M bit pre-symbols 408. Additionally, the second channel 407 is sent to the pre-symbol modulator 403 which modulates the M bit pre-symbols 408 with the secondary bits included in the second channel, as described above in step 304, to produce N bit symbols 409. (The encoder 402 and pre-symbol modulator 403 together comprise a symbol coder that can be implemented as a table-lookup routine.) In a preferred embodiment, the N bit symbols 409 may comprise

PSK or QAM symbols. The resulting N bit symbols 409 are sent to the wireless transmitter 404, such as an RF transmitter, which transmits the symbols 409 via a wireless communication resource 410, e.g., an RF carrier. FIG. 5 is a flowchart of a method for recovering primary and secondary bits based on a received symbol in accordance with the present invention. The method described in FIG. 5 can be incorporated, for example, into a wireless digital communication device. At step 501 , a symbol is received via any suitable transmission medium, for example, an RF carrier.

The received symbol may differ from the transmitted symbol due to the effects of transmission noise and interference. It is assumed that the received symbol was generated using at least one primary bit and at least one secondary bit, as described above.

At step 502, the symbol constellation is divided into the multi-symbol subsets, as discussed above. Although shown in FIG. 5 as a separate step performed for each received symbol, in a preferred embodiment, the step of determining the multi- symbol subsets is performed at the time of system design. At step 503, a distance is determined from the received symbol to one, and only one, symbol in each of the multi- symbol subsets. Generally, the closest symbol in each multi- symbol subset is chosen for this determination, as described below with reference to FIG. 6. This allows a reduction in complexity over methods available in the prior art in which all candidate symbols are used.

At step 504, the above-determined distances are used to decode the at least one primary bit to produce recovered primary bits. In a preferred embodiment, steps 503 and 504 are performed using a Viterbi decoder in which the branch metrics correspond to Euclidean distance between the received symbol and multi-symbol subsets rather than Euclidean distance between the received and candidate symbols. In effect, steps 503 and 504 recover the at least one primary bit by determining the maximum-likelihood multi-symbol subset.

At step 505, the position of the received symbol with respect to at least one predetermined decision boundary is determined. In this manner, the at least one secondary bit can be determined to produce recovered secondary bits, an example of which is shown in FIG. 6. In one embodiment of the present invention, step 505 can be performed as part of the Viterbi algorithm, thereby taking advantage of the error correcting abilities of the trellis. Understanding of the present invention may be facilitated through the use of an example. FIG. 6 is a block diagram illustrating an exemplary encoding and decoding of a symbol in accordance with the present invention. In particular, FIG. 6 graphically illustrates the various phases of encoding and decoding of a symbol as described above.

An encoder 601 trellis (or convolutionally) encodes, in this example, one or two primary bits to produce an M bit pre- symbol 603 that corresponds to a multi-symbol subset 602. The pre-symbol 603 is passed to a pre-symbol modulator 604 which uses, in this example, one secondary bit to select a symbol 605 from the multi-symbol subset 602. The symbol 605 is then transmitted via a wireless communication resource 607.

A received symbol 609, which is most likely altered due to transmission noise and/or interference, is sent to a soft decision decoder 610, e.g., a Viterbi decoder. The received symbol 609 is compared with only one candidate symbol 612- 619 in each multi-symbol subset. As described above, the candidate symbols 612-619 are selected as those symbols from each subset having the smallest Euclidean distance from the received symbol 609. As discussed above, the branch metrics employed by the soft decision decoder 610 correspond to Euclidean distances between the plurality of multi-symbol subsets. In determining the multi-symbol subset 61 1 meeting the maximum-likelihood decision criteria imposed by the decoder soft decision 610, one or two recovered primary bits (corresponding to the one or two primary bits previously encoded) are determined.

Once determined, the maximum-likelihood multi-symbol subset 61 1 is sent to a hard decision decoder 621 . The hard decision decoder 621 compares the received symbol 609 with a decision boundary 622 corresponding to the multi-symbol subset 61 1 in order to determine a symbol 623 from the multi- symbol subset that was most likely transmitted. In so doing, a recovered secondary bit (corresponding to the secondary bit used to modulate the pre-symbol 603) is determined.

It is understood that although FIG. 6 illustrates a 16QAM constellation, the present invention is not, however, restricted to only this type or size of constellation. A device incorporating the present invention may be implemented with a

64QAM or 8PSK constellation, for example.

With the present invention, a method and apparatus is provided which allows multiple series of bits to be error protected with varying degrees of protection. A significant benefit of the present invention is the decreased decoder complexity resulting from the use of multi-symbol subsets. Additionally, by using multiple series of bits, each requiring different coding rates (or different degrees of error protection) , the present invention accommodates greater flexibility in code-rate selection. For example, a 16QAM 1/3 rate code is effectively achieved by implementing what appears to be a 2/4 rate coder, but which actually provides 1/3 rate coding to a primary input bit and uses a secondary bit, in unencoded form, to increase the effective code rate to 2/4.

Claims

1 . A method comprising steps of:

- providing a series of primary bits to be error protected with a first degree of protection;

- providing a series of secondary bits to be error protected with a second degree of protection, which second degree of protection is less than the first degree of protection; - error control encoding at least a first selected primary bit of the series of primary bits to provide a M bit pre-symbol corresponding to the at least the first selected primary bit, wherein a symbol constellation is divided into a plurality of multi-symbol subsets and the M bit pre-symbol uniquely corresponds to a multi-symbol subset of the plurality of multi-symbol subsets; and

- modulating the M bit pre-symbol with at least a first selected secondary bit of the series of secondary bits to provide an N bit symbol that uniquely corresponds to a symbol included in the multi-symbol subset, wherein M bits of the N bit symbol correspond to the at least the first selected primary bit and N-M bits of the N bit symbol corresponds to the at least the first selected secondary bit.

2. The method of claim 1 , the step of providing the series of primary bits further comprising the step of providing digital voice coding parameters having a high degree of error sensitivity to be error protected with the first degree of protection .

3. The method of claim 1 , the step of providing the series of secondary bits further comprising the step of providing digital voice coding parameters having a low degree of error sensitivity to be error protected using one-to-one bit to symbol-bit mapping without resultant redundant protection.

4. The method of claim 1 , the step of error control encoding the at least the first selected primary bit to provide the M bit pre-symbol further comprising the step of trellis encoding the at least the first selected primary bit to provide the M bit pre- symbol, wherein trellis state transitions have a one-to-one correspondence with the plurality of multi-symbol subsets.

5. The method of claim 1 , the step of error control encoding the at least the first selected primary bit to provide the M bit pre-symbol further comprising the step of convolutional encoding the at least the first selected primary bit to provide the M bit pre-symbol, wherein convolutional state transitions have a one-to-one correspondence with the plurality of multi- symbol subsets.

6. The method of claim 1 , the step of error control encoding the at least the first selected primary bit to provide the M bit pre-symbol further comprising the step of error control encoding the at least the first selected primary bit to provide a 3 bit pre-symbol.

7. The method of claim 1 , the step of modulating the M bit pre-symbol with the at least the first selected secondary bit to provide the N bit symbol further comprising the step of modulating the M bit pre-symbol with the at least the first selected secondary bit to provide a 2^N quadrature amplitude modulated symbol.

8. The method of claim 1 , further comprising a step of transmitting the N bit symbol via a wireless communication resource.

9. A method comprising the steps of:

- providing at least a first series of secondary bits to be error protected with a second degree of protection, which second degree of protection is less than the first degree of protection ;

- error control encoding at least a first selected primary bit of the series of primary bits to provide an M bit pre-symbol corresponding to the first selected primary bit, wherein a symbol constellation is divided into a plurality of multi- symbol subsets and the M bit pre-symbol uniquely corresponds to a multi-symbol subset of the plurality of multi-symbol subsets; and - modulating the M bit pre-symbol with a first selected secondary bit of the at least the first series of secondary bits to produce an N bit symbol that uniquely corresponds to a symbol in the multi-symbol subset, wherein M is not less than two and N is greater than M, and wherein M bits of the N bit symbol correspond to the at least the first selected primary bit and N-M bits of the N bit symbol corresponds to the first selected secondary bit.

10. The method of claim 9, further comprising steps of:

- providing at least a second series of secondary bits to be error protected with the second degree of protection;

- error control encoding at least the first selected primary bit of the series of primary bits to provide an N-2 bit pre-symbol corresponding to the first selected primary bit, wherein the N-2 bit pre-symbol uniquely corresponds to the multi-symbol subset; and

- modulating the N-2 bit pre-symbol with the first selected secondary bit and a second selected secondary bit of the at least the second series of secondary bits to produce an N bit symbol that uniquely corresponds to a symbol in the multi-symbol subset, wherein N is greater than 3, and wherein at least two bits of the N bit symbol correspond to the at least the first selected primary bit, 1 bit of the N bit symbol corresponds to the first selected secondary bit, and 1 bit of the N bit symbol corresponds to the second selected secondary bit.

1 1 . The method of claim 9, the step of error control encoding the first selected primary bit to provide an N-1 bit pre-symbol further comprising the step of trellis encoding the first selected primary bit to provide the N-1 bit pre-symbol, wherein trellis state transitions have a one-to-one correspondence with the plurality of multi-symbol subsets.

12. The method of claim 9, the step of error control encoding the first selected primary bit to provide an N-1 bit pre-symbol further comprising the step of convolutional encoding the first selected primary bit to provide the N-1 bit pre-symbol, wherein convolutional state transitions have a one-to-one correspondence with the plurality of multi-symbol subsets.

13. An apparatus for a symbol coder, comprising:

- an encoder that receives a series of primary bits to be error protected with a first degree of protection and that provides at least one M bit pre-symbol corresponding to at least one primary bit of the series of primary bits, wherein a symbol constellation is divided into a plurality of multi- symbol subsets and the at least one M bit pre-symbol uniquely corresponds to a multi-symbol subset of the plurality of multi-symbol subsets; and - a pre-symbol modulator, coupled to the encoder, that receives the at least one M bit pre-symbol and a series of secondary bits to be error protected with a second degree of protection, which second degree of protection is less than the first degree of protection, and that modulates the M bit pre- symbol with at least one secondary bit of the series of secondary bits to provide at least one N bit symbol that uniquely corresponds to a symbol included in the multi-symbol subset, wherein M bits of the at least one N bit symbol correspond to the at least one pπmary bit and N-M bits of the N bit symbol corresponds to the at least one secondary bit.

14. The symbol coder of claim 13, the encoder further comprising a trellis encoder, wherein trellis state transitions of the trellis encoder have a one-to-one correspondence with the plurality of multi-symbol subsets.

15. The symbol coder of claim 13, the encoder further comprising a convolutional encoder, wherein convolutional state transitions of the convolutional encoder have a one-to- one correspondence with the plurality of multi-symbol subsets.

16. An apparatus for a wireless digital communication device, comprising:

- an encoder that receives a series of primary bits to be error protected with a first degree of protection and that provides at least one M bit pre-symbol corresponding to at least one primary bit of the series of primary bits, wherein a symbol constellation is divided into a plurality of multi- symbol subsets and the at least one M bit pre-symbol uniquely corresponds to a multi-symbol subset of the plurality of multi-symbol subsets;

- a pre-symbol modulator, coupled to the encoder, that receives the at least one M bit pre-symbol and a series of secondary bits to be error protected with a second degree of protection, which second degree of protection is less than the first degree of protection, and that modulates the M bit pre- symbol with at least one secondary bit of the series of secondary bits to provide at least one N bit symbol that uniquely corresponds to a symbol included in the multi-symbol subset, wherein M bits of the at least one N bit symbol correspond to the at least one primary bit and N-M bits of the

N bit symbol correspond to the at least one secondary bit; and

- a wireless transmitter, coupled to the pre-symbol modulator, that receives the at least one N bit symbol and that transmits the at least one N bit symbol via a wireless communication resource.

17. The wireless digital communication device of claim 16, the encoder further comprising a trellis encoder, wherein trellis state transitions of the trellis encoder have a one-to- one correspondence with the plurality of multi-symbol subsets.

18. The wireless digital communication device of claim 16, the encoder further comprising a convolutional encoder, wherein convolutional state transitions of the convolutional encoder have a one-to-one correspondence with the plurality of multi-symbol subsets.

19. The wireless digital communication device of claim 16, the wireless transmitter further comprising a radio frequency transmitter.

20. The wireless digital communication device of claim 16, further comprising :

- a voice coder, coupled to the encoder and the pre- symbol modulator, that provides digital voice coding parameters having a high degree of error sensitivity as the series of primary bits and that provides digital voice coding parameters having a low degree of error sensitivity as the series of secondary bits.

21 . A method, comprising the steps of:

- receiving an N bit symbol, wherein M bits of the N bit symbol correspond to at least one primary bit and N-M bits of the N bit symbol corresponds to at least one secondary bit; - dividing a symbol constellation into a plurality of multi-symbol subsets; and

- decoding the N bit symbol by determining distances between the N bit symbol and only one symbol of each multi- symbol subset.

22. The method of claim 21 , the step of decoding the N bit symbol further comprising Viterbi decoding the N bit symbol to produce at least one recovered primary bit.

23. The method of claim 22, the step of decoding the N bit symbol further comprising steps of:

- determining a position of the N bit symbol with respect to at least one predetermined decision boundary; and

- determining at least one recovered secondary bit as a function of the position of the N bit symbol with respect to the at least one predetermined decision boundary.