WO2002056480A1 - Decodeur de viterbi - Google Patents
Decodeur de viterbi Download PDFInfo
- Publication number
- WO2002056480A1 WO2002056480A1 PCT/JP2001/010624 JP0110624W WO02056480A1 WO 2002056480 A1 WO2002056480 A1 WO 2002056480A1 JP 0110624 W JP0110624 W JP 0110624W WO 02056480 A1 WO02056480 A1 WO 02056480A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- register
- stored
- memory
- comparison
- value
- Prior art date
Links
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/65—Purpose and implementation aspects
- H03M13/6561—Parallelized implementations
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/37—Decoding methods or techniques, not specific to the particular type of coding provided for in groups H03M13/03 - H03M13/35
- H03M13/39—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes
- H03M13/41—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes using the Viterbi algorithm or Viterbi processors
- H03M13/4107—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes using the Viterbi algorithm or Viterbi processors implementing add, compare, select [ACS] operations
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/37—Decoding methods or techniques, not specific to the particular type of coding provided for in groups H03M13/03 - H03M13/35
- H03M13/39—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes
- H03M13/41—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes using the Viterbi algorithm or Viterbi processors
- H03M13/4161—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes using the Viterbi algorithm or Viterbi processors implementing path management
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/65—Purpose and implementation aspects
- H03M13/6563—Implementations using multi-port memories
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/65—Purpose and implementation aspects
- H03M13/6566—Implementations concerning memory access contentions
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/65—Purpose and implementation aspects
- H03M13/6569—Implementation on processors, e.g. DSPs, or software implementations
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/37—Decoding methods or techniques, not specific to the particular type of coding provided for in groups H03M13/03 - H03M13/35
- H03M13/39—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes
- H03M13/41—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes using the Viterbi algorithm or Viterbi processors
- H03M13/4161—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes using the Viterbi algorithm or Viterbi processors implementing path management
- H03M13/4169—Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes using the Viterbi algorithm or Viterbi processors implementing path management using traceback
Definitions
- the present invention relates to a Viterbi decoding (Viterbi decoding) processing apparatus, and more particularly to a Viterbi decoding processing apparatus that performs a Viterbi decoding process for realizing high-quality digital communication.
- Viterbi decoding In digital wireless communication, a combination of convolutional coding and Viterbi decoding is widely used to suppress the effects of bit errors occurring on the transmission path and achieve high-quality communication.
- This Viterbi decoding process is realized by a digital signal processor (DSP: Digital Signal Processor) incorporated in communication equipment. Therefore, in recent years, DSP is required to have an architecture that can efficiently implement Viterbi decoding.
- Viterbi decoding uses the iterative structure of convolutional codes to efficiently execute maximum likelihood decoding by soft-decision. A series of processes called addition, comparison, and selection (ACS: Add Compare Select) is performed. This is realized by repetition and a traceback operation that finally decodes the data.
- DSP Digital Signal Processor
- FIG. 4 is an explanatory diagram showing the addition / comparison / selection processing.
- states at each time point and a pattern of branches showing transitions of these states are represented by simple repetition.
- pm (c) MAX (pm (a) + bm, pm (b-bm) (1)
- pm '(c + 1) MAX (pm (a)-bm, pm (b) + bm) (2)
- pm (a) and pm (b) are two states S a (t) that can be connected to states S c (t + 1) and S c + 1 (t + 1).
- the path metric value of S b (t), and bm and bm are the path and the received data sequence associated with the branch that connects each state, as shown in Fig. 4.
- a function for selecting the larger one of the values X and y is expressed as MAX (, y), and any one of the values x and y is used in the present invention.
- the function that selects the smaller value is denoted as MIN (x, y)
- MIN may be used instead of MAX depending on the branch metric representation method.
- FIG. 5 shows a block diagram of a conventional Viterbi decoding processing device according to Document 1 described above.
- the operands (address information) input to the adders / subtractors 52 and 53 and the value of the operation result are temporarily stored in one of a plurality of registers provided in the register file 51.
- the adders / subtracters 52 and 53 take two registers R i and Rj as input operands respectively, and the upper side R i (H) and R j (H) and the lower side R i (L) and Rj (L ) Is performed in parallel. The result of each addition and subtraction is stored in the upper Rk (H) and lower Rk (L) of the register Rk, respectively.
- the comparators 54 and 56 each take one register R i as an operand and perform a magnitude comparison operation on the upper side R i (H) and the lower side R i (L). Regis evening Store in Rk. At the same time, one bit of the result of the comparison operation is stored in each of the history registers 55 and 57.
- the addition, comparison, and operation processing for two states is executed in the following procedure.
- the branch metric value bm and the value of one bm are read from the memory 59 and stored in the upper and lower sides of the register R0, and the path metric values pm (a) and pm (b) are stored in the memory. It is read from 59 and stored in the upper and lower sides of register R1, respectively.
- the adder / subtractor 52 the register R0 and the register R1 are added in parallel, the value of pm (a) + bm is stored in the upper side of the register R2, and pm (b) is stored in the lower side. —The value of bm is stored.
- the adder / subtractor 53 performs parallel subtraction of the register R0 and the register R1, and stores the value of pm (a) _bm in the upper part of the register R3 and pm in the lower part.
- Reference 2 proposes an architecture that directly exchanges the operands and operation results of an arithmetic unit with the memory, which is different from an architecture that uses a register file as described above. However, they are essentially the same in that they have two sets of adder / subtractor and comparator and perform addition, comparison, and selection processing in parallel.
- the present invention is intended to solve such a problem, and has a high operating speed.
- a Viterbi decoding processing device capable of efficiently performing addition, comparison, and selection processing without requiring complicated processing. It is intended to provide.
- a Viterbi decoding device includes a memory for storing a path metric and a branch metric appearing in a Viterbi decoding process in a one-word width, and an upper side and a lower side respectively.
- a register file having a plurality of registers of at least two words wide for temporarily storing data of one word width, and a predetermined value using a path metric and a branch metric read from the memory via a bus to an arbitrary register of the register file; And a calculating means for calculating a new path metric based on the arithmetic expression of the following formula, wherein the calculating means takes predetermined two registers of each register as input operands, Compare the upper and lower values of the first register of the One of the values is stored in the upper part of the third register of the register, the upper and lower values of the other second register are compared, and the value is selected according to the comparison result.
- a comparison arithmetic unit that stores either value on the lower side of the third register is used.
- a 2-word data bus is used as the bus, and the upper and lower sides of the third register are stored in a register file. It is designed to simultaneously read 2-bit data stored in the lower side and store them at adjacent addresses in the memory via the data bus.
- a history register for simultaneously storing 2-bit values respectively representing the comparison results of the two comparison operations may be provided. At the time of the comparison operation by the comparison arithmetic unit, the larger value or the smaller value of the two values is selected according to a predetermined instruction.
- the arithmetic means takes two predetermined registers of the register as input operands, and sets the upper side of the fourth register and the upper side of the other fifth register.
- the first addition using the value of the first register and the second addition using the lower value of the fourth register and the lower value of the fifth register are performed in parallel, and the first addition is performed.
- the first adder / subtracter stores the result of the second addition in the upper side of the first register and stores the result of the second addition in the upper side of the first register, and inputs the fourth and fifth registers.
- the first subtraction using the upper values of the fourth and fifth registers as operands, and the lower values of the fourth and fifth registers and the lower values of the fifth register Is performed in parallel with the second subtraction using, and the result of the first subtraction is stored in the second register.
- a second second may be further provided a subtracter to store the results in the upper side of the second record Soo evening subtraction.
- the one-word-wide branch metric for the two branches that transit from the two states at the immediately preceding point to the one state at that point on the trellis diagram is Store the two branch metrics stored in memory in the register file at the same time via the data bus in the register file and load them to the upper and lower sides of the fifth register. It may be.
- FIG. 1 is a block diagram showing a video decoding apparatus according to an embodiment of the present invention.
- Figure 2 shows a configuration example of a register.
- FIG. 3 is an explanatory diagram illustrating the operation of the comparison arithmetic unit.
- FIG. 4 is an explanatory diagram showing the addition / comparison / selection processing.
- FIG. 5 is a block diagram showing a conventional Viterbi decoding processing device. Detailed description of the embodiment
- FIG. 1 is a block diagram showing a video decoding processing device according to one embodiment of the present invention.
- the video processing device includes arithmetic means 10, a register file 11, a history register 15, A memory 16, data buses 17 and 18, and a program control unit 19 are provided. Further, the arithmetic means 10 is provided with two adder / subtracters 12 and 13 and a comparator 14.
- the register file 11 is provided with a plurality of registers for temporarily storing input operands (a location for storing input data) to the arithmetic unit and operation results from the arithmetic means 10.
- Figure 2 shows an example of the register configuration.
- one register 20 has a 40-bit width, and the upper 8 bits (2 1) are headroom when used as an accumulator of a multiply-accumulate unit, and are denoted by R i (E). .
- the other 32 bits can be used for storing input operands and operation results to the operation means 10, and the upper 16 bits R i (H) corresponding to the upper side and the lower bits corresponding to the lower side It can also be used as a 16-bit register (2 2, 2 3) by separating it into 16-bit R i (L).
- the register has 40 bits
- the header has 8 bits
- the upper and lower bits are 16 bits.
- the number of bits does not necessarily have to conform to this.
- the adders / subtractors 1 2 and 1 3 take the two registers R i and R j as operands, and the upper side R i (H), R j (H) and the lower side R i (L), R j ( It is possible to perform parallel addition and subtraction operations on L). The results of addition and subtraction are These are stored in the upper Rk (H) and lower Rk (L) of the display Rk, respectively.
- the comparison operation unit 14 takes two registers R i and R j as input operands and performs two comparison operations in parallel. At the same time, the two bits representing the results of the two comparison operations are stored together in the history register 15.
- FIG. 3 shows the functions of the comparator 14 and the history register 15 in more detail.
- the upper side R i (H) and R i (L) of the register R i (31) are subjected to a magnitude comparison operation by the comparator 33, and the larger or smaller value is determined by the instruction into the register Rk (35). Stored in upper Rk (H).
- the upper side Rj (H) and Rj (L) of the second operand, the register register Rj (32), are compared for magnitude by the comparator 34, and the larger or smaller one is determined by the instruction.
- Each of the comparators 33 and 34 outputs one bit representing the operation result.
- the history register 15 consists of a fixed number of bits of the register 36, the contents of which are shifted two bits to the right, and the two bits output from the comparators 33 and 34 are stored in the register 36 in the history register 15. Stored in the upper two bits at the same time.
- the memory 16 stores a path metric and a branch metric used in the Viterbi decoding process in a one-word width, and is connected to the register file 11 via buses 17 and 18 each having a two-word width.
- the memory 16 has a multi-bank memory or a dual-port memory configuration, so that load / store can be performed in parallel via two buses 17 and 18.
- the buses 17 and 18 can simultaneously transfer 2-word data at adjacent addresses on the memory 16 by utilizing the 2-word width of each, and can load the upper and lower sides of the register Ri. Conversely, two words of data stored on the upper and lower sides of the register Ri can be simultaneously transferred and stored at adjacent addresses on the memory 16.
- the program control unit 19 executes the addition / comparison-arithmetic processing by sequentially outputting the instruction 1 OA based on a predetermined program 19A to control each of the above-described circuit units, and performs the Viterbi decoding processing. .
- the comparison operation unit 14 switches the comparison condition (MAXZMIN) based on the instruction 1 OA.
- the Viterbi decoder of Fig. 1 the information Each time a received data sequence for one report bit is obtained, the addition, comparison, and arithmetic processing for each state is performed in the following procedure.
- the program control unit 19 calculates a branch metric for each branch transitioning from the state immediately before to the state at the time in the trellis diagram using a predetermined circuit unit (not shown). Are stored in the memory 16 in advance. At this time, the memory 16 stores 1-word-wide branch metrics relating to the two branches that transit from the two states at the immediately preceding point on the trellis diagram to the same state at that point in adjacent addresses. deep. Then, the instruction 1 OA is output to each circuit section in FIG. 1 and the surviving path is selected by sequentially performing addition, comparison, and arithmetic processing for each state at the time.
- This embodiment focuses on the two states at the time at adjacent positions on the trellis diagram, and makes efficient addition and comparison by taking advantage of the commonality in addition, comparison, and arithmetic processing for these two states.
- ⁇ Arithmetic processing is performed.
- the addition and comparison of two adjacent states S c (t + 1) and S c + 1 (t + 1) on the trellis diagram at the time point t + 1 ⁇ Calculation processing will be described.
- the branch metrics bm, — bm relating to the two branches from the state S a (t), S b (t) transitioning to the state S c (t + 1) are adjacent to the memory 16 in advance as described above. Is stored at the specified address.
- the values of these branch metrics bm and one bm are read out simultaneously from the memory 16, and the register R0 (the fourth register) of the register file 11 is read via the bus 17 or the bus 18 having a width of 2 words.
- the adder / subtractor 12 performs parallel addition of the register R0 and the register R1, and stores the value of pm (a) + bm on the upper side of the register R2 (the first register). The value of pm (b) — bm is stored on the side.
- the adder / subtractor 13 performs parallel subtraction of the register R0 and the register R1, and the upper side of the register R3 (second register) has a value of pm (a) -one bm, and the lower side has pm (b). ) Stores the value of + bm.
- the comparator 14 compares the upper side and the lower side of the register R2 and the upper side and the lower side of the register R3, and the larger value is the higher side of the register R4 (third register). And 2 bits indicating the comparison result are stored in the history register 15. At this time, the value stored in the upper part of register R4 is equal to pm '(c) in the above equation (1), and the value stored in the lower part of register R4 is pm' (c + 1) in the equation (2). ) be equivalent to. The two bits stored in the history register 15 are used for traceback processing for final decoding using the last surviving path.
- new path metrics pm ′ (c), m of states S c (t + 1) and S c + 1 (t + 1) adjacent on the trellis diagram '(c + 1) is read and stored simultaneously to adjacent addresses in memory 16.
- the two registers R2 (first register) and register R3 (second register) are taken as input operands, and the upper and lower sides of register R2 are compared and calculated, and according to the comparison result,
- the selected value is stored in the upper side of another register R 4 (third register), the upper and lower sides of the register R 3 are compared, and the value selected in accordance with the comparison result is calculated.
- Data is stored in the lower side of register R4, and the upper and lower side data of register R4 are simultaneously read out and stored in adjacent addresses on memory 16 via bus 17 or bus 18 having a 2-word width.
- the store operation to the memory is performed. Only needs to be added once, It becomes possible to perform efficiently compare and selection process. Therefore, addition / comparison / selection processing can be performed efficiently without increasing the operating speed or requiring complicated processing.
- one-word-wide branch metrics relating to the two branches that transition from an arbitrary state at the immediately preceding point on the trellis diagram to two new states at that point are stored in adjacent addresses, respectively.
- these two branch metrics are read out simultaneously via the bus 17 or bus 18 having a width of 2 ⁇ and the upper part of the register R 1 (the fifth register) is read. Since the load is made on the side and the lower side, the load operation from the memory only needs to be performed once compared with the case where the two branch metrics are separately loaded from the memory as in the past, and the addition, comparison, and selection Processing can be performed efficiently.
- addition, comparison, and selection processing for two states can be performed with a total of three load operations and one store operation in addition to the path operation of the path metric.
- the cycle required for the operation can be reduced to about 2 Z 3 even when compared with the case where the operation is performed twice and the store operation is performed twice.
- two buses 17 and 18 having a width of two words are provided, and the load operation described above is performed in parallel between the memory 16 and the register file 11 via these two buses. Since the store operation can be executed, pipeline processing becomes possible, and on average, two cycles of addition, comparison, and selection processing for two states can be executed. It can be reduced to about 1 Z3. .
- the 2-bit values representing the comparison results of the two comparison operations executed by the comparison operation unit 14 are stored at the same time.
- the number of history registers can be reduced and the circuit configuration can be simplified, as compared with the case where the comparison result is stored in different history registers one bit at a time.
- the comparison arithmetic unit 14 selects the larger or smaller one of the two values to be compared according to the instruction 1 OA from the program control unit 19. Therefore, it is possible to easily switch the selection criteria as needed.
- adder / subtracters 52 and 53 are provided.
- the adder / subtractor 52 takes two register bits R 0 (fourth register bit) and a register R 1 (fifth register) as input operands, and A first addition using the upper value of one register R0 and the upper value of the other register R1, and a second addition using the lower value of the register R0 and the lower value of the register R1. Addition and execution are performed in parallel, the first addition result is stored in the upper part of register R2 (first register), and the second addition result is stored in the lower part of register R2.
- a first subtraction using the upper value of the register R0 and the upper value of the register R1 and a second subtraction using the lower value of the register R0 and the lower value of the register R1 are performed.
- the first subtraction result is stored in the upper part of the register R 3 (the second register evening), and the second subtraction is performed. Since the result was such that store the lower side of the register R 3, a subtraction processing for two states in parallel can be efficiently processed.
- the present invention takes two predetermined registers as input operands, compares the upper and lower values of one of the first registers, and selects a value according to the comparison result.
- the stored value is stored in the upper part of the third register of the register, and the upper and lower values of the other second register are compared and calculated, and according to the comparison result,
- a comparison arithmetic unit that stores one of the selected values at the lower side of the third register and a 2-word data bus are provided, and the upper and lower sides of the third register are provided by the comparison arithmetic unit.
- the two word width data stored in the memory are simultaneously read out and stored at adjacent addresses on the memory via the data bus.
Landscapes
- Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Error Detection And Correction (AREA)
- Detection And Correction Of Errors (AREA)
- Detection And Prevention Of Errors In Transmission (AREA)
Description
Claims
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US10/466,302 US20040054958A1 (en) | 2001-01-15 | 2001-12-05 | Viterbi decoder |
EP01273082A EP1355431B1 (en) | 2001-01-15 | 2001-12-05 | Viterbi decoding processor |
DE60117705T DE60117705T2 (de) | 2001-01-15 | 2001-12-05 | Viterbi-Decodierungsprozessor |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2001-006066 | 2001-01-15 | ||
JP2001006066A JP3984790B2 (ja) | 2001-01-15 | 2001-01-15 | ビタビ復号処理装置 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2002056480A1 true WO2002056480A1 (fr) | 2002-07-18 |
Family
ID=18874020
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2001/010624 WO2002056480A1 (fr) | 2001-01-15 | 2001-12-05 | Decodeur de viterbi |
Country Status (5)
Country | Link |
---|---|
US (1) | US20040054958A1 (ja) |
EP (1) | EP1355431B1 (ja) |
JP (1) | JP3984790B2 (ja) |
DE (1) | DE60117705T2 (ja) |
WO (1) | WO2002056480A1 (ja) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4408783B2 (ja) * | 2004-09-29 | 2010-02-03 | Necエレクトロニクス株式会社 | 復号装置及び復号方法 |
JP5502739B2 (ja) * | 2007-10-26 | 2014-05-28 | クゥアルコム・インコーポレイテッド | 最適化ビタビ復号器およびgnss受信機 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0690178A (ja) * | 1992-07-23 | 1994-03-29 | Matsushita Electric Ind Co Ltd | 演算装置 |
JPH10171778A (ja) * | 1996-12-13 | 1998-06-26 | Nec Corp | ブロックiirプロセッサ |
JPH11261426A (ja) * | 1997-11-14 | 1999-09-24 | Lucent Technol Inc | 自動ビタビトレースバックビット格納機能を有する並列算術論理プロセッサ |
JP2000222174A (ja) * | 1999-02-03 | 2000-08-11 | Nec Corp | 信号処理プロセッサ及びそれに用いる丸め機能付き積和演算器 |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5440504A (en) * | 1993-02-19 | 1995-08-08 | Matsushita Electric Industrial Co., Ltd. | Arithmetic apparatus for digital signal processor |
JP3338374B2 (ja) * | 1997-06-30 | 2002-10-28 | 松下電器産業株式会社 | 演算処理方法および装置 |
US6601215B1 (en) * | 2000-02-01 | 2003-07-29 | Agere Systems Inc. | Traceback buffer management for VLSI Viterbi decoders |
-
2001
- 2001-01-15 JP JP2001006066A patent/JP3984790B2/ja not_active Expired - Lifetime
- 2001-12-05 WO PCT/JP2001/010624 patent/WO2002056480A1/ja active IP Right Grant
- 2001-12-05 EP EP01273082A patent/EP1355431B1/en not_active Expired - Lifetime
- 2001-12-05 DE DE60117705T patent/DE60117705T2/de not_active Expired - Lifetime
- 2001-12-05 US US10/466,302 patent/US20040054958A1/en not_active Abandoned
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0690178A (ja) * | 1992-07-23 | 1994-03-29 | Matsushita Electric Ind Co Ltd | 演算装置 |
JPH10171778A (ja) * | 1996-12-13 | 1998-06-26 | Nec Corp | ブロックiirプロセッサ |
JPH11261426A (ja) * | 1997-11-14 | 1999-09-24 | Lucent Technol Inc | 自動ビタビトレースバックビット格納機能を有する並列算術論理プロセッサ |
JP2000222174A (ja) * | 1999-02-03 | 2000-08-11 | Nec Corp | 信号処理プロセッサ及びそれに用いる丸め機能付き積和演算器 |
Non-Patent Citations (2)
Title |
---|
See also references of EP1355431A4 * |
TAKAHIRO KUMURA ET AL.: "Viterbi senyou meirei wo sonaeta hanyou DSP ni yoru kousoku viterbi fukugou", DENSHI JOHO TSUUSHIN GAKKAI, 2001-NEN SOUGOU TAIKAI KOUEN RONBUNSHUU (KISO, KYOUKAI) (A-4-40), 9 March 2001 (2001-03-09), pages 140, XP002908804 * |
Also Published As
Publication number | Publication date |
---|---|
JP2002217747A (ja) | 2002-08-02 |
DE60117705D1 (de) | 2006-05-04 |
EP1355431A1 (en) | 2003-10-22 |
EP1355431B1 (en) | 2006-03-08 |
EP1355431A4 (en) | 2005-06-22 |
JP3984790B2 (ja) | 2007-10-03 |
US20040054958A1 (en) | 2004-03-18 |
DE60117705T2 (de) | 2006-08-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US5946361A (en) | Viterbi decoding method and circuit with accelerated back-tracing and efficient path metric calculation | |
JP3338374B2 (ja) | 演算処理方法および装置 | |
KR100439211B1 (ko) | 연산처리장치 | |
JPH10107651A (ja) | ビタビ復号装置 | |
US5742621A (en) | Method for implementing an add-compare-select butterfly operation in a data processing system and instruction therefor | |
WO2002056480A1 (fr) | Decodeur de viterbi | |
JP3607116B2 (ja) | 演算処理装置 | |
EP1322041A1 (en) | Viterbi decoder using restructured trellis | |
JP3191442B2 (ja) | ビタビ復号用演算装置 | |
EP1058392A1 (en) | Method for implementing a plurality of add-compare-select butterfly operations in parallel, in a data processing system | |
JP3996858B2 (ja) | 演算処理装置 | |
JP2002521907A (ja) | ビタビ復号器のための高速メトリック計算 | |
JP3634333B2 (ja) | ディジタル信号処理プロセッサ | |
KR100414152B1 (ko) | 프로그래머블 프로세서에서의 비터비 디코딩 연산방법 및그 연산방법을 실행하기 위한 연산회로 | |
JPH04352518A (ja) | 演算装置 | |
JP3193830B2 (ja) | 演算回路 | |
JP2001024526A (ja) | ビタビ復号装置 | |
JPH0722969A (ja) | 演算装置 | |
JPH0766735A (ja) | ビタビ復号装置およびステートメトリック正規化方法 | |
JP2001186021A (ja) | 誤り検出装置及びそれを内蔵した演算処理装置 | |
JP2001222408A (ja) | プロセッサ | |
JPH08234983A (ja) | ディジタル・シグナル・プロセッサ | |
JPH0746145A (ja) | 演算装置 | |
JPH10107652A (ja) | 演算装置 | |
JP2003051750A (ja) | 演算処理装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): KR US |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE TR |
|
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2001273082 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 10466302 Country of ref document: US |
|
WWP | Wipo information: published in national office |
Ref document number: 2001273082 Country of ref document: EP |
|
WWG | Wipo information: grant in national office |
Ref document number: 2001273082 Country of ref document: EP |