WO2018014738A1 - 一种适用于OvXDM***的快速译码方法、装置及OvXDM*** - Google Patents

一种适用于OvXDM***的快速译码方法、装置及OvXDM*** Download PDF

Info

Publication number
WO2018014738A1
WO2018014738A1 PCT/CN2017/092067 CN2017092067W WO2018014738A1 WO 2018014738 A1 WO2018014738 A1 WO 2018014738A1 CN 2017092067 W CN2017092067 W CN 2017092067W WO 2018014738 A1 WO2018014738 A1 WO 2018014738A1
Authority
WO
WIPO (PCT)
Prior art keywords
measure
path
measures
paths
module
Prior art date
Application number
PCT/CN2017/092067
Other languages
English (en)
French (fr)
Inventor
刘若鹏
季春霖
徐兴安
刘子红
张莎莎
Original Assignee
深圳超级数据链技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from CN201610587517.4A external-priority patent/CN107645363B/zh
Priority claimed from CN201610584674.XA external-priority patent/CN107645359B/zh
Application filed by 深圳超级数据链技术有限公司 filed Critical 深圳超级数据链技术有限公司
Priority to JP2019503461A priority Critical patent/JP6744982B2/ja
Priority to KR1020197005393A priority patent/KR102203029B1/ko
Priority to EP17830372.3A priority patent/EP3490177A4/en
Publication of WO2018014738A1 publication Critical patent/WO2018014738A1/zh
Priority to US16/254,548 priority patent/US20190229838A1/en

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L1/00—Arrangements for detecting or preventing errors in the information received
    • H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
    • H04L1/0045—Arrangements at the receiver end
    • H04L1/0054—Maximum-likelihood or sequential decoding, e.g. Viterbi, Fano, ZJ algorithms
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L1/00—Arrangements for detecting or preventing errors in the information received
    • H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
    • H04L1/0056—Systems characterized by the type of code used
    • H04L1/0059—Convolutional codes
    • H04L1/006—Trellis-coded modulation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04J—MULTIPLEX COMMUNICATION
    • H04J13/00—Code division multiplex systems
    • H04J13/10—Code generation
    • H04J13/102—Combining codes
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L1/00—Arrangements for detecting or preventing errors in the information received
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L1/00—Arrangements for detecting or preventing errors in the information received
    • H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
    • H04L1/0045—Arrangements at the receiver end
    • H04L1/0047—Decoding adapted to other signal detection operation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L1/00—Arrangements for detecting or preventing errors in the information received
    • H04L1/004—Arrangements for detecting or preventing errors in the information received by using forward error control
    • H04L1/0045—Arrangements at the receiver end
    • H04L1/0052—Realisations of complexity reduction techniques, e.g. pipelining or use of look-up tables
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L5/00—Arrangements affording multiple use of the transmission path
    • H04L5/02—Channels characterised by the type of signal
    • H04L5/023—Multiplexing of multicarrier modulation signals
    • H04L5/026—Multiplexing of multicarrier modulation signals using code division

Definitions

  • the present invention relates to the field of signal processing, and in particular, to a fast decoding method, apparatus, and OvXDM system suitable for an OvXDM system.
  • overlapping multiplex systems whether it is Overlapped Time Division Multiplexing (OvTDM), Overlapped Frequency Division Multiplexing (OvFDM), or Overlapped Code Division Multiplexing (OvCDM) System
  • OFDM Overlapped Time Division Multiplexing
  • OFDM Overlapped Frequency Division Multiplexing
  • OFCDM Overlapped Code Division Multiplexing
  • overlapping space division multiplexing (OvSDM Overlapped Space Division Multiplexing) system
  • OFvHDM Overlapped Hybrid Division Multiplexing
  • Trellis trellis
  • the number of nodes determines the complexity of decoding, and for a system with an overlap number of K and a modulation dimension of M ( M is an integer greater than or equal to 2), and the number of nodes in the stable state in the corresponding trellis diagram is MK-1, so the decoding complexity increases exponentially with the number of overlaps K.
  • the spectral efficiency of the system is 2K/symbol, so the larger the number of overlaps K, the higher the spectral efficiency.
  • the requirement of improving the spectral efficiency is such that the larger the number of overlaps K is, the better, and on the other hand, the smaller the number of times of overlap K is, the better, in order to reduce the decoding complexity, in particular, when the number of overlaps K is increased to a certain extent.
  • the value for example, if K is greater than 8, the decoding complexity increases sharply.
  • the existing decoding method is difficult to meet the requirements of real-time decoding, and the spectral efficiency, decoding complexity and decoding efficiency form a contradiction requirement.
  • the number of nodes in the corresponding trellis diagram is M K'-1 , so the decoding complexity will follow the number K of coding paths . 'And the index increased.
  • the number of coding branches K' is required to be as large as possible, so that the spectral efficiency is higher, but at the same time, the decoding complexity increases sharply with the increase of K', so the spectrum efficiency and decoding complexity, The decoding efficiency creates a pair of contradictory requirements.
  • the present application provides a fast decoding method, apparatus and OvXDM system suitable for an OvXDM system, and the decoding complexity does not increase sharply with the increase of K/K' as in the conventional decoding scheme, and the frequency is solved.
  • the present application provides a fast decoding method suitable for an OvXDM system, comprising the following steps:
  • Step 1 respectively calculating a measure between all potential paths of the first r symbols and the first r received symbols in the received symbol sequence
  • Step 2 Sort the calculated measures, and store the smaller R n measures and their corresponding paths;
  • Step 3 performing M-dimensional expansion on the last node of each path currently stored, calculating an instantaneous measure between the extended path and the corresponding received symbol in the received symbol sequence, and corresponding each instantaneous measure to the previous moment.
  • the accumulated measures are added to obtain an accumulated measure of each path after the current time is added;
  • Step 4 Sort the accumulated measures of the added paths, and store the smaller Rn measures and their corresponding paths;
  • Step 5 When step 3 is extended to the node corresponding to the last symbol in the received symbol sequence, step 4 stops correspondingly when the smaller R n measures corresponding to the entire received symbol sequence and their corresponding paths are stored, otherwise the repetition is repeated. Step three and step four;
  • Step 6 selecting a path with the smallest measure as the decoding path to perform the decision output
  • R n is a positive integer, which is preset according to requirements and is smaller than the number of nodes of the trellis corresponding to the OvXDM system.
  • the present application provides a fast decoding apparatus suitable for an OvXDM system, including:
  • a first calculating module configured to separately calculate a measure between all the potential paths of the first r symbols and the first r received symbols
  • a first sorting module configured to sort the calculated measures
  • R n distance memories and corresponding R n path memories are respectively used for storing smaller R n measures obtained in the first sorting module and their respective corresponding paths;
  • An extension module configured to perform M-dimensional expansion on a last node of each path currently stored
  • the second calculating module calculates an instantaneous measure between the extended received path and the corresponding received symbol in the received symbol sequence, and adds each instantaneous measure to the accumulated measure corresponding to the previous time, to obtain each of the current time added Cumulative measure of the path;
  • a second sorting module configured to sort the accumulated measures of the added paths obtained in the second calculating module, wherein the smaller R n measures and their corresponding paths are used to update the R n
  • the distance memory and the corresponding R n path memories respectively stop when the smaller R n measures corresponding to the entire received symbol sequence and their respective paths are stopped;
  • the decision output module selects the stored path memory stored in the path memory corresponding to the smallest distance
  • the path is used as a decoding path for decision output;
  • R n is a positive integer, which is preset according to requirements and is smaller than the number of nodes of the trellis corresponding to the OvXDM system.
  • a method for quickly decoding an OvXDM system includes the following steps:
  • Step 1 respectively calculating a measure between all potential paths of the first r symbols and the first r received symbols in the received symbol sequence
  • Step 2 Sort the calculated measures, and store the smaller R n measures and their corresponding paths;
  • Step 3 Extend the path corresponding to the currently stored minimum measure, calculate the instantaneous measure between the extended path and the corresponding received symbol, and add each instantaneous measure to the accumulated measure corresponding to the previous time to obtain the current The cumulative measure of each extended path after the time is added;
  • Step 4 Sort the accumulated measure of each extended path with the remaining un-expanded R n -1 measures, and store the smaller R n measures and their corresponding paths;
  • Step 5 When the path corresponding to the minimum measure currently stored in step 3 is extended to reach the depth of the received symbol sequence, the instantaneous measure between the extended path and the corresponding received symbol is calculated, and each instantaneous measure is compared, and The path corresponding to the smallest instantaneous measure is used as the decoding path; otherwise, steps 3 and 4 are repeated;
  • R n is a positive integer and is less than the number of nodes of the trellis diagram corresponding to the high-weight OvXDM system.
  • a fast decoding apparatus for an OvXDM system includes:
  • a first calculating module configured to separately calculate a measure between all potential paths of the first r symbols and the first r received symbols in the received symbol sequence
  • a first sorting module configured to sort the calculated measures
  • R n distance memories and corresponding R n path memories are respectively used for storing smaller R n measures obtained in the first sorting module and their respective corresponding paths;
  • An expansion module configured to expand a path corresponding to the currently stored minimum measure
  • a second calculating module configured to calculate an instantaneous measure between the extended path and the corresponding received symbol, and add each instantaneous measure to the accumulated measure corresponding to the previous time, to obtain each of the current time added The cumulative measure of the extended path;
  • a second sorting module configured to sort the accumulated measure of the extended paths calculated in the second calculating module and the remaining un-expanded R n -1 measures, wherein the smaller R n measures and The respective corresponding paths are used to update the values of the R n distance memories and the corresponding R n path memories;
  • the second calculation module calculates the instantaneous measure between the extended path and the corresponding received symbol, and compares The output module compares each instantaneous measure and uses the path corresponding to the smallest instantaneous measure as the decoding path; otherwise, the expansion module, the second calculation module, and the second sequencing module repeatedly work;
  • R n is a positive integer and is less than the number of nodes of the trellis corresponding to the OvXDM system.
  • FIG. 1 is a schematic flowchart of a fast decoding method applicable to an OvXDM system according to an embodiment of the present application
  • FIG. 2 is a schematic structural diagram of a fast decoding apparatus applicable to an OvXDM system according to an embodiment of the present application
  • FIG. 3 is a schematic structural diagram of a transmitting end of an OvFDM system according to a first embodiment of the present application
  • 4(a) and 4(b) are schematic diagrams showing the structure of a receiving end of an OvFDM system in the first embodiment of the present application;
  • FIG. 5 is a code tree diagram of an OvFDM system in a first embodiment of the present application.
  • FIG. 6 is a trellis diagram of decoding of an OvFDM system in a first embodiment of the present application
  • FIG. 7 is a schematic diagram of a trellis diagram extension of an OvFDM system in a first embodiment of the present application
  • FIG. 8 is a schematic diagram of decoding of a fast decoding method applicable to an OvFDM system according to a first embodiment of the present application.
  • FIG. 9 is a performance comparison diagram of a fast decoding method and a conventional decoding method applicable to an OvFDM system according to a first embodiment of the present application.
  • FIG. 10 is a fast decoding method and a traditional translation applicable to an OvFDM system according to a first embodiment of the present application. Comparison time of decoding method of code method;
  • FIG. 11 is a schematic structural diagram of a transmitting end of an OvTDM system according to a second embodiment of the present application.
  • FIG. 12(a) is a schematic diagram of a preprocessing unit of an OvTDM system in a second embodiment of the present application
  • FIG. 12(b) is a schematic diagram of a sequence detecting unit of an OvTDM system in a second embodiment of the present application.
  • FIG. 13 is a code tree diagram of an OvTDM system in a second embodiment of the present application.
  • FIG. 15 is a schematic diagram of a trellis diagram extension of an OvTDM system in a second embodiment of the present application.
  • 16 is a schematic diagram of decoding of a fast decoding method applicable to an OvTDM system in a second embodiment of the present application;
  • 17 is a performance comparison diagram of a fast decoding method and a conventional decoding method applicable to an OvTDM system in a second embodiment of the present application;
  • 18 is a comparison diagram of decoding times of a fast decoding method applicable to an OvTDM system and a conventional decoding method in a second embodiment of the present application;
  • FIG. 19 is a schematic structural diagram of an OvCDM system according to a third embodiment of the present application.
  • FIG. 20 is a schematic structural diagram of an encoder of an OvCDM system according to a third embodiment of the present application.
  • 21 is a schematic diagram of an encoding matrix of an OvCDM system in a third embodiment of the present application.
  • FIG. 22 is a schematic structural diagram of a decoder of an OvCDM system according to a third embodiment of the present application.
  • FIG. 23 is a cell diagram corresponding to an OvCDM system in a third embodiment of the present application.
  • 24 is a schematic diagram of decoding of a fast decoding method applicable to an OvCDM system according to a third embodiment of the present application.
  • FIG. 25 is a schematic flow chart of another fast decoding method suitable for an OvXDM system.
  • the traditional decoding scheme generally uses the Viterbi decoding scheme, which is based on the full extension of all nodes in the corresponding lattice graph of the system. For the measurement of each path, the path with the smallest measure is selected as the decoding path. As can be seen from the principle of the Viterbi decoding scheme, the complexity of decoding will overlap. The number of times/coded branches increases and the index increases.
  • the state does not need to traverse all state nodes and their extended paths, and only needs to select some state nodes and paths to expand by measuring ordering, thereby greatly reducing the decoding complexity and improving the decoding efficiency. The details are described below.
  • the present application discloses a fast decoding method suitable for the OvXDM system, as shown in FIG. 1 and FIG. 25, which includes steps S01-S19.
  • Step S01 respectively calculating a measure between all the potential paths of the first r symbols and the first r received symbols in the received symbol sequence.
  • M is an integer equal to or greater than 2.
  • the Euclidean distance is the real distance between the two signals. It can truly reflect the distance between the actual signal and the ideal signal. In this patent, the Euclidean distance is defined as
  • step S03 the respective measures calculated in step S01 are sorted.
  • Step S05 Store the smaller R n measures and their corresponding paths obtained by sorting in step S03.
  • Step S07 expands the last node of each path currently stored.
  • the M-dimensional extension is performed on the last node of each path currently stored.
  • Step S09 Calculate an instantaneous measure between the extended path and the corresponding received symbol in the received symbol sequence.
  • each instantaneous measure calculated in step S09 is added to the accumulated measure corresponding to the previous time to obtain an accumulated measure of each path after the current time is added.
  • the accumulated time of the previous time is multiplied by the weighting factor and then added to the instantaneous measure. This is to make As the depth of the path increases, the reference of the node measure farther away from the current node is gradually weakened, so that the decoding accuracy is higher.
  • the weighting factor has a value greater than 0 and less than or equal to 1.
  • Step S13 Sort the accumulated measures of the added paths in step S11. Or sorting along with the remaining unscaled Rn -1 measures stored.
  • Step S15 storing the smaller R n measures and their respective corresponding paths.
  • Step S17 when step S07 is extended to the node corresponding to the last symbol in the received symbol sequence, step S15 stops correspondingly storing the smaller R n measures corresponding to the entire received symbol sequence and their respective corresponding paths, otherwise it is repeated. Steps S07 to S15.
  • step S19 is performed, and the instantaneous measure between the extended path and the corresponding received symbol is calculated, and Comparing each instantaneous measure, the path corresponding to the smallest measure is taken as the decoding path; otherwise, steps S07 to S15 are repeated.
  • Step S19 Select a path with the smallest measure as the decoding path to perform the decision output.
  • R n is a positive integer, which is preset according to requirements and smaller than the number of nodes of the trellis corresponding to the OvXDM system.
  • R n is smaller than M K-1 .
  • r is the value of log M R n rounded down.
  • X may represent any domain, including time domain T, frequency domain F, space S, code domain C, or hybrid H.
  • the OvXDM system may be an OvTDM system, an OvFDM system, an OvCDM system, or an OvSDM system. Or OvHDM system, etc.
  • the present application does not need to traverse all state nodes and their extended paths in the decoding process, and only needs to select partial state nodes and paths for expansion by measurement ordering, thereby greatly reducing decoding complexity and improving decoding efficiency. .
  • the number of paths discarded each time is M K -R n , so the smaller the R n
  • the decoding complexity is lower, but R n cannot be infinitely small, because the smaller the R n is , the larger the decoding performance loss is, and the higher the signal-to-noise ratio is required under the same bit error rate condition. Therefore, the selection of R n is also very important. It is necessary to ensure that the decoding performance loss is small while reducing the complexity.
  • R n is generally smaller than M K-1 and the size is equal to M K-4 , where K
  • M represents an M-dimensional system (M is an integer greater than or equal to 2), at which time the decoding complexity is greatly reduced while the decoding performance is guaranteed.
  • M represents an M-dimensional system (M is an integer greater than or equal to 2), at which time the decoding complexity is greatly reduced while the decoding performance is guaranteed.
  • M represents an M-dimensional system (M is an integer greater than or equal to 2), at which time the decoding complexity is greatly reduced while the decoding performance is guaranteed.
  • M is an integer greater than or equal to 2
  • the number of paths discarded per extension is M LK' -R n
  • R n is less than M L(K'-4) .
  • the present application also proposes a fast decoding device suitable for the OvXDM system, as shown in FIG. 2, which includes a first computing module 01, a first sorting module 03, R n distance memories 05, R n Path memory 07, expansion module 09, second calculation module 11, second sequencing module 13, and decision output module 15.
  • the first calculation module 01 is configured to separately calculate a measure between all the potential paths of the first r symbols and the first r received symbols.
  • the first sorting module 03 is configured to sort the respective measures calculated by the first calculating module 01.
  • the R n distance memories 05 are used to store the smaller R n measures obtained in the first sorting module, respectively, and the corresponding R n path memories 07 are used to store the paths corresponding to the R n measures respectively.
  • R n is less than M K-1 , where M represents the dimension of the system and the value is an integer greater than or equal to 2.
  • the extension module 09 is configured to perform M-dimensional expansion on the last node of each path currently stored. Alternatively, in some embodiments, only the path corresponding to the smallest of the currently stored measures is expanded.
  • the second calculating module 11 is configured to calculate an instantaneous measure between the extended path and the corresponding received symbol in the received symbol sequence, and add each instantaneous measure to the accumulated measure corresponding to the previous time to obtain the current time.
  • the weighting factor module 17 is configured to: when the second calculating module 11 adds each instantaneous measure to the accumulated measure corresponding to the previous time, first multiply the accumulated measure of the previous time by the weighting factor, thereby The accumulated measurement at the previous time is first multiplied by the weighting factor and then added to the instantaneous measurement.
  • the weighting factor module 17 is introduced to make the reference of the node measure farther away from the current node gradually weakened as the path depth increases, so that the decoding accuracy is higher.
  • the weighting factor has a value greater than 0 and less than or equal to 1.
  • the second sorting module 13 is configured to sort the accumulated measures of the added paths obtained in the second calculating module 11, wherein the smaller R n measures and their respective paths are used to update the R n distance memory 05 and corresponding R n path memory 07 values.
  • the expansion module 09, the second calculation module 11 and the second sequencing module 13 repeat the operation until the extension module 09 expands to the node corresponding to the last symbol in the received symbol sequence such that R n distance memories 05 and corresponding R n paths
  • the memory 07 is stopped when storing the smaller R n measures corresponding to the entire received symbol sequence and their respective corresponding paths.
  • the second calculation module 11 calculates the instantaneous measure between the extended path and the corresponding received symbol, and compares the output.
  • the module 15 compares the instantaneous measures and uses the path corresponding to the smallest instantaneous measure as the decoding path; otherwise, the expansion module 09, the second calculation module 11, and the second sequencing module 13 repeatedly operate.
  • the decision output module 15 selects the path stored in the path memory 07 corresponding to the stored distance memory 05 with the smallest measure as the decoding path for decision output.
  • R n is a positive integer, which is preset according to requirements and smaller than the number of nodes of the trellis corresponding to the OvXDM system.
  • r is the value of log M R n rounded down, where M represents the dimension of the system and is an integer greater than or equal to two.
  • R n is smaller than M K-1 and the size is equal to M K-4 , where K is the number of overlaps of received symbols, where M represents the dimension of the system, It is an integer greater than or equal to 2; when the OvXDM system is an OvCDM system, the Rn is less than M L(K'-4) and the size is equal to M L(K'-2) , where K' is a received symbol
  • L is the encoding constraint length of the received symbols
  • M is the dimension of the system, which is an integer greater than or equal to 2.
  • the present application also discloses an OvXDM system including the above-described fast decoding device suitable for the OvXDM system.
  • the OvXDM system may be an OvTDM system, an OvFDM system, an OvCDM, an OvSDM or an OvHDM system.
  • This embodiment may be described by taking an OvFDM system as an example.
  • FIG. 3 it is the transmitting end of the OvFDM system, which first encodes the frequency domain signal according to a certain rule, and then converts the frequency domain signal into a time domain signal, that is, inverse Fourier transform, and then sends the signal out.
  • an initial envelope waveform is first generated according to the design parameter; then the initial envelope waveform is shifted in a frequency domain according to a predetermined spectral interval according to the number of overlapping multiplexing, to obtain an envelope waveform of each subcarrier; The data sequence is multiplied by the corresponding subcarrier envelope waveform to obtain a modulation envelope waveform of each subcarrier; and the modulation envelope waveform of each subcarrier is superposed on the frequency domain to obtain a complex modulation envelope in the frequency domain.
  • the receiving end of the OvFDM system receives a signal in the time domain through the antenna. If the receiving signal is to be decoded, the time domain signal needs to be converted into a frequency domain signal, that is, after the Fourier transform. Can be processed.
  • symbol synchronization is first formed on the received signal in the time domain; then the received signal of each symbol time interval is sampled and quantized to become a received digital signal sequence; the time domain signal is converted into a frequency domain signal, and then The frequency domain signal is segmented by the spectral interval ⁇ B to form an actual received signal segmentation spectrum; and a one-to-one correspondence between the received signal spectrum and the transmitted data symbol sequence is formed, and finally, the data is detected according to the one-to-one correspondence relationship.
  • Symbol sequence Both the inverse Fourier transform and the Fourier transform in the OvFDM system involve the setting of the number of sampling points. The number of sampling points should be consistent, and the value is 2 n , and n is a positive integer.
  • the distance memory 05 is selected as R n
  • the path memory 07 is also R n , wherein R n is less than 2 K-1 , thereby reducing the complexity of decoding.
  • R n is less than 2 K-1 , thereby reducing the complexity of decoding.
  • a sequence of symbols of length r is expressed as: For 2D modulation, there are 2 r possible combinations of information in combination. Thus a matrix of 2 r * r dimensions, denoted as msg(2 r *r), where each row represents a sequence of symbols of length r.
  • the instantaneous measure of each set of symbol sequences and the first r received symbols is calculated separately, defined as: Where Vr is the first r received symbols, and x r,k is the window function of OvFDM. Is the rkth symbol in the i-th row of msg(2 r *r).
  • the decoding reaches r nodes, and the corresponding instantaneous measure is d r,i , and the corresponding 2r decoding path is the i-th row symbol of msg(2 r *r), and the instantaneous measure and the path are respectively stored in the distance memory. 05 and path memory 07.
  • each node can perform 2D expansion at the same time, expand upward when the input is +1, and expand downward when the input is -1, as shown in Figure 7, calculate the node to which the extension arrives.
  • Instantaneous measure The instantaneous measure is added to the accumulated measure at the previous moment of the path to obtain a measure of 2 r+1 paths.
  • the fast decoding method proposed by the present application only needs to extend R n nodes at a time, and the traditional method needs to expand 2 K-1 nodes.
  • R n when K is compared
  • R n can take a relatively small value and satisfies R n ⁇ 2 K-1 , so the decoding complexity is greatly reduced for a decoding sequence of length N.
  • the present embodiment is convenient for description, and the fast decoding method of the present application is described by a two-dimensional modulation system. In fact, the method is applicable to a system of M-dimensional modulation, where M can be greater than or equal to 2. Integer.
  • the fast decoding method of the present application is as follows:
  • d prev represents the cumulative measure corresponding to the node before the expansion
  • d prev represents the cumulative measure corresponding to the node before the expansion
  • the path memories U 1 , U 2 , U 3 , U 4 and the distance memories d 1 , d 2 , d 3 , d 4 respectively store four paths and their corresponding measures.
  • the path corresponding to the minimum distance is the final decoding result.
  • FIG. 9 and FIG. 10 respectively show performance comparison and operation time comparison between the fast decoding method of the present example and the current Viterbi decoding method. It can be clearly seen from the figure that the performance of the fast decoding method of this example is not lost. 1dB, but the time is greatly compressed, which can greatly reduce the decoding complexity and improve the decoding efficiency while ensuring the decoding performance.
  • This embodiment may be described by taking an OvTDM system as an example.
  • the transmitting end of the OvTDM system first generates an initial envelope waveform in a time domain according to design parameters; and then shifts the initial envelope waveform in a time domain according to a predetermined time interval according to the number of overlapping multiplexing times.
  • FIG. 12 it is a receiving end of the OvTDM system, which forms a received digital signal sequence for the received signals in each frame, and performs detection on the formed received digital signal sequence to obtain modulation in the frame length in all symbols. The decision of the modulated data on.
  • a two-dimensional modulation OvTDM system with K times of overlap the number of nodes of the trellis diagram is 2 K-1 , the distance memory required in the traditional decoding algorithm is 2 K-1 , and the path memory is also 2 K-1
  • the number of nodes of the corresponding trellis diagram is also 4.
  • the distance memory 05 is selected as Rn
  • the path memory 07 is also R n , wherein R n is less than 2 K-1 , thereby reducing the complexity of decoding.
  • R n is less than 2 K-1 , thereby reducing the complexity of decoding.
  • a sequence of symbols of length r is expressed as: For 2D modulation, there are 2 r possible combinations of information in combination. Thus a matrix of 2 r * r dimensions, denoted as msg(2 r *r), where each row represents a sequence of symbols of length r.
  • the instantaneous measure of each set of symbol sequences and the first r received symbols is calculated separately, defined as: Where v n is the first r received symbols, and x r,k is the window function of OvTDM. Is the rkth symbol in the i-th row of msg(2 r *r).
  • the decoding reaches r nodes, and the corresponding instantaneous measure is d r,i , and the corresponding 2 r decoding path is the i-th row symbol of msg(2 r *r), and the instantaneous measure and the path are respectively stored in the distance.
  • each node can perform 2D expansion at the same time, expand upward when the input is +1, and expand downward when the input is -1, as shown in Figure 15, calculate the node of the extended arrival.
  • Instantaneous measure The instantaneous measure is added to the accumulated measure at the previous moment of the path to obtain a measure of 2 r+1 paths.
  • the fast decoding method proposed by the present application only needs to extend R n nodes at a time, and the traditional method needs to expand 2 K-1 nodes.
  • R n when K is compared
  • R n can take a relatively small value and satisfies R n ⁇ 2 K-1 , so the decoding complexity is greatly reduced for a decoding sequence of length N.
  • the present embodiment is convenient for description, and the fast decoding method of the present application is described by a two-dimensional modulation system. In fact, the method is applicable to a system of M-dimensional modulation, where M can be greater than or equal to 2. Integer.
  • the fast decoding method of the present application is as follows:
  • d 1 , d 2 , d 3 , d 4 are stored in the distance memory 05, and the corresponding paths U 1 , U 2 , U 3 , U 4 are stored in the path memory 07, in order to facilitate the text, the following d 1 , d 2 , d 3 , and d 4 refer to the above four distance memories 05, and U 1 , U 2 , U 3 , and U 4 refer to the above four path memories 07.
  • d prev represents the cumulative measure corresponding to the node before the expansion
  • d prev represents the cumulative measure corresponding to the node before the expansion
  • the path memories U 1 , U 2 , U 3 , U 4 and the distance memories d 1 , d 2 , d 3 , d 4 respectively store four paths and their corresponding measures.
  • the path corresponding to the minimum distance is the final decoding result.
  • FIG. 17 and FIG. 18 respectively show the performance comparison and operation time comparison between the fast decoding method of the present example and the current Viterbi decoding method, which is a result of 10000 Monte Carlo simulation of the computer, and it can be clearly seen from the figure.
  • the performance of the fast decoding method of this example loses less than 1 dB, but the time of the localization is greatly compressed, that is, the decoding complexity can be greatly reduced while the decoding complexity is improved, and the decoding performance is guaranteed.
  • This embodiment may be described by taking an OvCDM system as an example.
  • the core of the overlapping code division multiplexing of the OvCDM system is overlap and multiplexing, with the aim of improving the spectral efficiency of the communication system.
  • the OvCDM system generalizes the convolutional coding coefficients to the generalized convolutional coding model of the complex domain, and generates the constraint relationship by symbol overlap.
  • the main parameters include the number of coding branches K' and the length of the coding constraint L.
  • the system structure diagram is shown in Fig. 19.
  • the corresponding encoder structure is as shown in FIG.
  • the key of the OvCDM system is the coding matrix, that is, the convolutional expansion coefficient, which is required to satisfy the linear relationship.
  • the input sequence corresponds to the output sequence one by one, so theoretically, there is no error decoding, and generally all the larger-measurement matrix is searched by computer.
  • the coding matrix the coding matrix is arranged as shown in FIG.
  • u i u i, 0 u i, 1 u i, 2 ...
  • OvCDM code rate Where n is the length of the substream. When n is long, the bit rate loss caused by the tailing of the shift register is negligible, so there is r OVCDM ⁇ k.
  • the traditional binary domain convolutional coding model has a code rate generally less than 1, which leads to loss of spectral efficiency.
  • the convolutional code rate of the complex domain of OvCDM is equal to 1, and the one-way convolutional coding extension does not cause spectral efficiency loss, and additional coding gain is added.
  • the receiving end After receiving the signal, the receiving end synchronizes the signal, estimates the channel, digitizes the data, and then quickly decodes the processed data.
  • the core of the decoding algorithm is to calculate the received signal and the ideal state, and use the path memory and the measure to determine the best decoding path to obtain the final detection sequence.
  • the sequence detection process block diagram is shown in Figure 22.
  • This embodiment is applicable to the fast decoding method of the OvCDM system, and the specific steps are as follows:
  • Each node includes a total of S states, and the first node is measured.
  • the method is to calculate the ideal signal waveform and the received signal sequence of all m strips from the previous state to the state. Between the measures d s,m (l,l+1), the expression is
  • the measures d s,m (l,l+1) of the respective states S of the current node are added to the measures d s', l-1 of their respective departure states S' to form an accumulated measure of the new m paths.
  • the path measures are sorted, the R n paths with smaller path measures are selected, and the corresponding paths are stored in the path memory, and the measures are stored in the distance memory. The remaining paths are discarded and the next phase is extended from the reserved path.
  • Steps (2)-(4) are repeated until the end of decoding. At this time, R n paths and their corresponding path measures are reserved in the memory, and the path with the smallest measurement is the decoding result.
  • the decoding process of the OvCDM system is illustrated.
  • the corresponding trellis diagram corresponding to the OvCDM system is shown in Figure 23.
  • u 1 ⁇ +1,-1,-1,-1,+1,+1,-1,-1 ⁇
  • the encoding implementation process of the OvCDM system can also adopt other methods.
  • the invention of the present application lies in the decoding process and not in the encoding process.
  • the first symbol received is 1-j, which is measured separately from the ideal four states (1,1), (1,-1), (-1,1), (-1,-1).
  • the paths corresponding to the first three smaller measures are (1, -1), (1, 1), (-1, -1) in descending order.
  • the corresponding first three smaller measures and their corresponding paths are stored in the distance memory 05 and the path memory 07, respectively.
  • the second symbol-2 is processed, and the currently reserved paths are respectively (1, -1), (1, 1), (-1, -1), and each state is expanded, and each state can be Perform a four-dimensional expansion, so get 12 paths, calculate the measure of each path and the current received symbol, and sort the measures of the 12 paths.
  • the first three smaller measures are retained, the memory is updated, and the measure and the corresponding path are stored in the distance memory 05 and the path memory 07, respectively.
  • the following symbols use the same method to obtain 12 measures by accumulating and retaining the three paths with the smallest measure and their corresponding measures. After the last symbol is calculated, three paths and their corresponding path measures are obtained. The sequence corresponding to the path with the smallest measure is the final decoded output sequence, which is (1, -1, -1, -1, -1). 1,1,-1,1,1,-1,1,-1,-1,-1,-1,-1,1), the decoding ends.
  • a sequence of symbols of length r is represented as For 2D modulation, there are 2 r possible combinations of information in combination.
  • a matrix of 2 r * r dimensions can be obtained, denoted as msg(2 r *r), where each row represents a sequence of symbols of length r.
  • the instantaneous measure of each set of symbol sequences and the first r received symbols is calculated separately, defined as: Where v r is the first r received symbols, and x r,j is the window function of the OvFDM system. Is the (rj)th symbol in the i-th row of msg(2 r *r).
  • the decoding reaches the rth node, and includes 2 r decoding paths U and 2 r instantaneous measures d, and the node depth of each decoding path is r.
  • the instantaneous measure and the corresponding decoding path are stored in the distance memory 05 and the path memory 07, respectively.
  • the 2 r instantaneous measures obtained in (1) are sequentially sorted, and the indexes idx min and idx max corresponding to the minimum and maximum measures are respectively found, and the instantaneous measure corresponding to the minimum index of the measure is The depth of the index node corresponding to the minimum of the measure is increased by 1, and the node depth becomes r+1.
  • the decoding path corresponding to the minimum index of the measurement is extended, and can be extended to two branches of 1 and -1, that is, the decoding path becomes with
  • the extended decoding path has 2 r +1, but we only need to keep 2 r paths, so the decoding path corresponding to the largest measure index found in (2) is deleted, and the decoding branch will be expanded to 1.
  • index idx max can also be replaced with -1 branch
  • -1 branch Continue to be stored in memory with index idx min . At this time, only the depth of the node corresponding to the idx min and idx max indexes is r+1 in the decoding path, and the node depth of the remaining indexes is still r.
  • the extended two paths are respectively calculated with the received symbols and the calculation formula is the same as the formula for calculating the measure in (2), and the measures of the two paths are obtained as d' r+1 and d" r+1 respectively .
  • the instantaneous measures corresponding to the r nodes are respectively added to obtain an accumulated measure.
  • a weighting factor alpha can be introduced, and the value is greater than 0 and less than or equal to 1, the specific value depends on system requirements,
  • the purpose is to gradually weaken the node measurement reference far away from the current node as the depth of the node increases, and the addition process can be expressed as:
  • the obtained accumulated measures are stored in the corresponding idx min and idx max memories in (3), respectively.
  • the remaining symbol sequences are also subjected to the above methods (2) to (4), and the filtering is extended until the node depth reaches the symbol sequence depth N, and the instantaneous measure after the last symbol expansion is compared, and the smaller one is the final output translation. Code path.
  • This embodiment uses the OvTDM system as an example for description.
  • OvTDM For a two-dimensionally modulated OvTDM system, starting from the origin of the trellis diagram, it has only two arriving nodes, no matter how large the overlap weight K is.
  • the instantaneous measures arriving at the branches of the two nodes are respectively calculated, and the smallest ones are selected from the instantaneous path measures of the two arriving nodes to expand, and the instantaneous measures of the extended two branches are respectively calculated, and the two arriving nodes are recorded.
  • R n arrival nodes and their path instantaneous measures are retained at a time, each time extending from the node with the smallest instantaneous path measure, adding one arrival node per expansion, the path with larger instantaneous path measure and its measurement abandon. This continues until the end of the data frame of the trellis, whose path to the node with the smallest instantaneous path measure is the decision output.
  • the decoding step will be described in detail below.
  • R n 2 r may be used .
  • a sequence of symbols of length r is represented as For 2D modulation, there are 2 r possible combinations of information in combination.
  • a matrix of 2 r * r dimensions can be obtained, denoted as msg(2 r *r), where each row represents a sequence of symbols of length r.
  • the instantaneous measure of each set of symbol sequences and the first r received symbols is calculated separately, defined as: Where v r is the first r received symbols, and x r,j is the window function of the OvTDM system. Is the (rj)th symbol in the i-th row of msg(2 r *r).
  • the decoding reaches the rth node, and includes 2 r decoding paths U and 2 r instantaneous measures d, and the node depth of each decoding path is r.
  • the instantaneous measure and the corresponding decoding path are stored in the distance memory 05 and the path memory 07, respectively.
  • the 2 r instantaneous measures obtained in (1) are sequentially sorted, and the indexes idx min and idx max corresponding to the minimum and maximum measures are respectively found, and the instantaneous measure corresponding to the minimum index of the measure is The depth of the index node corresponding to the minimum of the measure is increased by 1, and the node depth becomes r+1.
  • the decoding path corresponding to the minimum index of the measurement is extended, and can be extended to two branches of 1 and -1, that is, the decoding path becomes with
  • the extended decoding path has 2 r +1, but we only need to keep 2 r paths, so the decoding path corresponding to the largest measure index found in (2) is deleted, and the decoding branch will be expanded to 1.
  • index idx max can also be replaced with -1 branch
  • -1 branch Continue to be stored in memory with index idx min . At this time, only the depth of the node corresponding to the idx min and idx max indexes is r+1 in the decoding path, and the node depth of the remaining indexes is still r.
  • the extended two paths are respectively calculated with the received symbols and the calculation formula is the same as the formula for calculating the measure in (2), and the measures of the two paths are obtained as d' r+1 and d" r+1 respectively .
  • the instantaneous measures corresponding to the r nodes are added to obtain an accumulated measure.
  • a weighting factor alpha can be introduced, and the value is 0 to 1. The specific value depends on the system requirements, and the purpose is to follow As the depth of the node increases, the node measurement reference that is farther away from the current node is gradually weakened, and the addition process can be expressed as:
  • the obtained accumulated measures are stored in the corresponding idx min and idx max memories in (3), respectively.
  • the remaining symbol sequences are also subjected to the above methods (2) to (4), and the filtering is extended until the node depth reaches the symbol sequence depth N, and the instantaneous measure after the last symbol expansion is compared, and the smaller one is the final output translation. Code path.
  • the number of coding branches of the OvCDM system is K' and the coding constraint length L, and the dimension of the coded output vector is N, which uses two-dimensional modulation, so the symbol is + 1, -1. Since the number of coding branches is K' and the length of the coding constraint is L, the number of nodes that can reach each state is 2 K' , which includes 2 K' states S, so the total number of nodes reachable in all states is 2 2K' .
  • the distance memory 05 is selected as R n
  • the path memory 07 is also R n , wherein R n is less than 2 K'-1 , thereby reducing the complexity of decoding. In an embodiment, let r be The value of log 2 R n is rounded down.
  • a sequence of symbols of length r is represented as For 2D modulation, there are 2 r possible combinations of information in combination.
  • a matrix of 2 r * r dimensions can be obtained, denoted as msg(2 r *r), where each row represents a sequence of symbols of length r.
  • the instantaneous measure of each set of symbol sequences and the first r received symbols is calculated separately, defined as: Where v r is the first r received symbols, and x r, j is the encoding matrix of OvCDM, Is the (rj)th symbol in the i-th row of msg(2 r *r).
  • the decoding reaches the rth node, and includes 2 r decoding paths U and 2 r instantaneous measures d, and the node depth of each decoding path is r.
  • the instantaneous measure and the corresponding decoding path are stored in the distance memory 05 and the path memory 07, respectively.
  • Each node includes a total of S states, and the first node is measured.
  • the method is to calculate the ideal signal symbols and received signal sequences of all m pieces from the previous state to the state. Between the measures d s,m (l,l+1), the expression is
  • the weighting factor alpha can be introduced, and the value is 0 to 1. The specific value depends on the system requirements. The purpose of this is to gradually weaken the node measurement farther from the current node as the depth of the node increases. reference.
  • Steps (3)-(5) are repeated until the end of decoding.
  • the R n paths and their corresponding paths Euclidean distance are reserved in the memory, and the path with the smallest Euclidean distance is the decoding result.
  • the reserved path number R n determines the information retained by the decoding process.
  • the number of discarded paths is 2 2K' -R n , so the smaller the R n is, the smaller the decoding is.
  • R n cannot be infinitely small, and the smaller the R n is , the greater the loss of decoding performance is, and the higher signal-to-noise ratio is required under the same bit error rate condition. Therefore, it is necessary to select an appropriate R n according to the actual system and channel, and to reduce the decoding complexity while ensuring a small performance loss of decoding.
  • the value of R n is selected to be greater than or equal to 2 L(K'-4) and less than or equal to 2 L(K'-2) , thereby ensuring decoding performance while greatly reducing decoding complexity.
  • the fast decoding method and device proposed by the present application can be widely applied to an actual mobile communication system, such as TD-LTE, TD-SCDMA, etc., in the OvXDM system, and can also be widely applied to satellite communication and microwave viewing.
  • any wireless communication system such as communication, scatter communication, atmospheric optical communication, infrared communication and aquatic communication.
  • the fast decoding method and apparatus of the present application can be applied to both large-capacity wireless transmissions and small-capacity lightweight radio systems.
  • the fast decoding method proposed in this patent filters the nodes to be extended during the path expansion of the decoding. Since the correct path must be one of the better paths, only the nodes with better paths need to be selected. Expansion, discarding poor nodes, and no longer expanding them later, thereby reducing the complexity of decoding and improving the decoding efficiency of the system.

Landscapes

  • Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Artificial Intelligence (AREA)
  • Error Detection And Correction (AREA)

Abstract

本申请公开了一种适用于OvXDM***的快速译码方法、装置及OvXDM***。所述方法在译码过程中无需遍历所有状态节点及其扩展路径，只需通过测度排序选取部分状态节点和路径进行扩展，因此可以大幅度降低译码复杂度，并提高译码效率，其译码复杂度并不会像传统译码方案一样随着重叠复用次数K的增加而急剧增加，解决了频谱效率与译码复杂度、译码效率这一对矛盾需求。

Description

一种适用于OvXDM***的快速译码方法、装置及OvXDM*** 技术领域
本发明涉及涉及信号处理领域，尤其涉及一种适用于OvXDM***的快速译码方法、装置及OvXDM***。
背景技术
对于重叠复用***——不管是重叠时分复用(OvTDM，Overlapped Time Division Multiplexing)***、重叠频分复用(OvFDM，Overlapped Frequency Division Multiplexing)***还是重叠码分复用(OvCDM，Overlapped Code Division Multiplexing)***、重叠空分复用(OvSDM，Overlapped Space Division Multiplexing)***、重叠混合复用(OvHDM，Overlapped Hybrid Division Multiplexing)***其传统的译码中，都需要不断访问格状图(Trellis)中的节点，并为每一个节点设置两个存储器，一个用于存储到达该节点的相对最佳路径，一个用于存储到达该节点的相对最佳路径对应的测度。
对于OvTDM***和OvFDM***，由于译码过程中，需要对格状图中每个节点进行扩展，因此节点数决定了译码的复杂度，而对于重叠次数为K和调制维度为M的***(M是大于等于2的整数)，其对应的格状图中稳定状态的节点数为MK-1，因此译码复杂度会随着重叠次数K而指数增加。而在OvTDM***和OvFDM***中，***的频谱效率为2K/符号，因此重叠次数K越大频谱效率越高。因此，一方面出于提高频谱效率的要求使得重叠次数K越大越好，另一方面出于降低译码复杂度的要求使得重叠次数K越小越好，特别地，当重叠次数K增加到一定值，例如K大于8后，译码复杂度急剧增加，现有的译码方法难以满足实时译码的要求，频谱效率与译码复杂度、译码效率形成了一对矛盾需求。
类似地，对于一个编码支路数为K’的M维调制OvCDM***，其对应的格状图中稳定状态的节点数为MK’-1，因此译码复杂度会随着编码去路数K’而指数增加。而在OvCDM***中，需要尽可能大的编码支路数K’，以使得频谱效率越高，但是同时译码复杂度会随K’的增加而急剧增加，因此频谱效率与译码复杂度、译码效率形成了一对矛盾需求。
发明内容
本申请提供一种适用于OvXDM***的快速译码方法、装置及OvXDM***，其译码复杂度并不会像传统译码方案一样随着K/K’的增加而急剧增加，解决了频 谱效率与译码复杂度、译码效率这一对矛盾需求。
根据本申请的第一方面，本申请提供一种适用于OvXDM***的快速译码方法，包括以下步骤：
步骤一、分别计算前r个符号所潜在的全部路径与接收符号序列中前r个接收符号之间的测度；
步骤二、对计算得到各测度进行排序，并存储其中较小的Rn个测度及其各自对应的路径；
步骤三、对当前存储的每条路径的最后一个节点进行M维扩展，对扩展出的路径计算其与接收符号序列中对应接收符号之间的瞬时测度，并将各瞬时测度与其前一时刻对应的累加测度相加，得到当前时刻相加后的各路径的累加测度；
步骤四、对所述相加后的各路径的累加测度进行排序，并存储其中较小的Rn个测度及其各自对应的路径；
步骤五、当步骤三中扩展到接收符号序列中最后一个符号对应的节点，步骤四相应地存储有对应整个接收符号序列的较小的Rn个测度及其各自对应的路径时停止，否则重复步骤三和步骤四；
步骤六、选取测度最小的一条路径作为译码路径，以进行判决输出；
Rn为一正整数，其根据需求进行预设，且小于OvXDM***对应的格状图的节点数。
根据本申请的第二方面，本申请提供一种适用于OvXDM***的快速译码装置，包括：
第一计算模块，用于分别计算前r个符号所潜在的全部条路径与前r个接收符号之间的测度；
第一排序模块，用于对计算得到各测度进行排序；
Rn个距离存储器及对应的Rn个路径存储器，分别用于存储第一排序模块中得到的较小的Rn个测度及其各自对应的路径；
扩展模块，用于对当前存储的每条路径的最后一个节点进行M维扩展；
第二计算模块，对扩展出的路径计算其与接收符号序列中对应接收符号之间的瞬时测度，并将各瞬时测度与其前一时刻对应的累加测度相加，得到当前时刻相加后的各路径的累加测度；
第二排序模块，用于对第二计算模块中得到的所述相加后的各路径的累加测度进行排序，其中较小的Rn个测度及其各自对应的路径用于更新所述Rn个距离存储器及对应的Rn个路径存储器中的值；扩展模块、第二计算模块和第二排序模块重复进行工作，直到扩展模块扩展到接收符号序列中最后一个符号对应的节点以使得Rn个距离存储器及对应的Rn个路径存储器分别存储有对应整个接收符号序列的较小的Rn个测度及其各自对应的路径时停止；
判决输出模块，选取存储的测度最小的距离存储器对应的路径存储器中存储 的路径作为译码路径，以进行判决输出；
Rn为一正整数，其根据需求进行预设，且小于OvXDM***对应的格状图的节点数。
根据本申请的第三个方面，一种OvXDM***的快速译码方法，包括以下步骤：
步骤一、分别计算前r个符号所潜在的全部路径与接收符号序列中前r个接收符号之间的测度；
步骤二、对计算得到的各测度进行排序，并存储其中较小的Rn个测度及其各自对应的路径；
步骤三、对当前存储的最小测度对应的路径进行扩展，对扩展出的路径计算其与对应接收符号之间的瞬时测度，并将各瞬时测度与前一时刻对应的累加测度相加，得到当前时刻相加后的各扩展路径的累加测度；
步骤四、对所述各扩展路径的累加测度与存储的其余未进行扩展的Rn-1个测度进行排序，并存储其中较小的Rn个测度及其各自对应的路径；
步骤五、当步骤三中当前存储的最小测度对应的路径扩展后到达接收符号序列的深度时，则对扩展出的路径计算其与对应接收符号之间的瞬时测度，并比较各瞬时测度，将最小的瞬时测度对应的路径作为译码路径；否则重复步骤三和步骤四；
其中Rn为一正整数，且小于高重OvXDM***对应的格状图的节点数。
本发明的第四个方面，一种OvXDM***的快速译码装置，包括：
第一计算模块，用于分别计算前r个符号所潜在的全部路径与接收符号序列中前r个接收符号之间的测度；
第一排序模块，用于对计算得到的各测度进行排序；
Rn个距离存储器及对应的Rn个路径存储器，分别用于存储第一排序模块中得到的较小的Rn个测度及其各自对应的路径；
扩展模块，用于对当前存储的最小测度对应的路径进行扩展；
第二计算模块，用于对扩展模块扩展出的路径计算其与对应接收符号之间的瞬时测度，并将各瞬时测度与前一时刻对应的累加测度相加，得到当前时刻相加后的各扩展路径的累加测度；
第二排序模块，用于对第二计算模块中计算得到的所述各扩展路径的累加测度与存储的其余未进行扩展的Rn-1个测度进行排序，其中较小的Rn个测度及其各自对应的路径用于更新所述Rn个距离存储器及对应的Rn个路径存储器的值；
比较输出模块；当扩展模块对当前存储的最小测度对应的路径进行扩展达到接收符号序列的深度时，第二计算模块对扩展模块扩展出的路径计算其与对应接收符号之间的瞬时测度，比较输出模块比较各瞬时测度，并将最小的瞬时测度对应的路径作为译码路径；否则扩展模块、第二计算模块和第二排序模块重复进行工作；
其中Rn为一正整数，且小于OvXDM***对应的格状图的节点数。
本申请的有益效果是：
依上述实施的适用于OvXDM***的快速译码方法、装置及OvXDM***，由于在译码过程中无需遍历所有状态节点及其扩展路径，只需通过测度排序选取部分状态节点和路径进行扩展，因此可以大幅度降低译码复杂度，并提高译码效率。
附图说明
图1为本申请一实施例中的适用于OvXDM***的快速译码方法的流程示意图；
图2为本申请一实施例中的适用于OvXDM***的快速译码装置的结构示意图；
图3为本申请第一种实施例中OvFDM***的发射端的结构示意图；
图4(a)和(b)本申请第一种实施例中OvFDM***的接收端的结构示意图；
图5为本申请第一种实施例中OvFDM***的码树图；
图6为本申请第一种实施例中OvFDM***的译码的格状图；
图7为本申请第一种实施例中OvFDM***的格状图扩展的示意图；
图8为本申请第一种实施例中适用于OvFDM***的快速译码方法的译码示意图；
图9为本申请第一种实施例中适用于OvFDM***的快速译码方法与传统译码方法的性能对比图；
图10为本申请第一种实施例中适用于OvFDM***的快速译码方法与传统译 码方法的译码时间对比图；
图11为本申请第二种实施例中OvTDM***的发射端的结构示意图；
图12(a)为本申请第二种实施例中OvTDM***的预处理单元示意图；
图12(b)为本申请第二种实施例中OvTDM***的序列检测单元示意图；
图13为本申请第二种实施例中OvTDM***的码树图；
图14为本申请第二种实施例中OvTDM***的译码的格状图；
图15为本申请第二种实施例中OvTDM***的格状图扩展的示意图；
图16为本申请第二种实施例中适用于OvTDM***的快速译码方法的译码示意图；
图17为本申请第二种实施例中适用于OvTDM***的快速译码方法与传统译码方法的性能对比图；
图18为本申请第二种实施例中适用于OvTDM***的快速译码方法与传统译码方法的译码时间对比图；
图19为本申请第三种实施例中OvCDM***的结构示意图；
图20为本申请第三种实施例中OvCDM***的编码器的结构示意图；
图21为本申请第三种实施例中OvCDM***的编码矩阵的示意图；
图22为本申请第三种实施例中OvCDM***的译码器的结构示意图；
图23为本申请第三种实施例中OvCDM***对应的格状图；
图24为本申请第三种实施例中适用于OvCDM***的快速译码方法的译码示意图；
图25为另一适用于OvXDM***的快速译码方法的流程示意图。
具体实施方式
下面通过具体实施方式结合附图对本申请作进一步详细说明。
对于OvXDM***，例如OvTDM***、OvFDM***和OvCDM***，传统的译码方案一般采用维特比(Viterbi)译码方案，其原理是对***的对应的格状图中全部节点都进行充分扩展，计算每条路径的测度，最后选取测度最小的路径作为译码路径。从维特比译码方案的原理可以看出，其译码的复杂度会随着重叠 次数/编码支路数的增加而指数增加。
本申请在译码过程中无需遍历所有状态节点及其扩展路径，只需通过测度排序选取部分状态节点和路径进行扩展，因此可以大幅度降低译码复杂度，并提高译码效率。下面具体说明。
本申请公开了一种适用于OvXDM***的快速译码方法，如图1和图25所示，其包括步骤S01～S19。
步骤S01、分别计算前r个符号所潜在的全部条路径与接收符号序列中前r个接收符号之间的测度。对于M维调制的OvTDM***和OvFDM***等，前r个符号所潜在的全部条路径数量为Mr，其中M为大于或等于2的整数。本申请中的测度表示两个信号之间的距离，定义为：
Figure PCTCN2017092067-appb-000001
当p＝2时即为欧式距离，欧式距离是两个信号之间的真实距离，能够真实的反应实际信号和理想信号之间的距离，本专利中欧氏距离定义为
Figure PCTCN2017092067-appb-000002
步骤S03、对步骤S01中计算得到各测度进行排序。
步骤S05、存储步骤S03中排序得到的其中较小的Rn个测度及其各自对应的路径。
步骤S07、如图1所示，对当前存储的每条路径的最后一个节点进行扩展。对于M维调制的OvTDM***和OvFDM***等，是对当前存储的每条路径的最后一个节点进行M维扩展。
或者，如图25所示，在该步骤中，仅对当前存储的最小测度对应的路径进行扩展。
步骤S09、对扩展出的路径计算其与接收符号序列中对应接收符号之间的瞬时测度。
步骤S11、在一个实施例中，如图1所示，将步骤S09计算得到的各瞬时测度与其前一时刻对应的累加测度相加，得到当前时刻相加后的各路径的累加测度。在另一较优的实施例中，如图25所示，将各瞬时测度与前一时刻对应的累加测度相加时，前一时刻累加测度先与权重因子相乘后再与瞬时测度相加，这是为了使 得随着路径深度的增加，逐渐弱化距离当前节点较远的节点测度的参考，使得译码准确度更高。在一较优实施例中，权重因子的取值为大于0且小于等于1。
步骤S13、对步骤S11中所述相加后的各路径的累加测度进行排序。或者在进行排序时连同存储的其余未进行扩展的Rn-1个测度进行排序。
步骤S15、存储其中较小的Rn个测度及其各自对应的路径。
步骤S17、当步骤S07中扩展到接收符号序列中最后一个符号对应的节点，步骤S15相应地存储有对应整个接收符号序列的较小的Rn个测度及其各自对应的路径时停止，否则重复步骤S07～S15。
又或者，当步骤S07中对当前存储的最小测度对应的路径进行扩展后达到接收符号序列的深度时，则进行步骤S19、对扩展出的路径计算其与对应接收符号之间的瞬时测度，并比较各瞬时测度，将最小的测度对应的路径作为译码路径；否则重复步骤S07～S15。
步骤S19、选取测度最小的一条路径作为译码路径，以进行判决输出。
在上述译码方法中，Rn为一正整数，其根据需求进行预设，且小于OvXDM***对应的格状图的节点数。对于重叠次数为K的M维调制的OvTDM***和OvFDM***等，Rn小于MK-1。在一较优实施例中，r为logMRn的数值向下取整。
在一实施例中，X可代表任何域，包括时域T，频域F，空间S，码域C或混合H等，对应的，OvXDM***可为OvTDM***、OvFDM***、OvCDM***、OvSDM***或OvHDM***等。
如上所述，本申请在译码过程中无需遍历所有状态节点及其扩展路径，只需通过测度排序选取部分状态节点和路径进行扩展，因此可以大幅度降低译码复杂度，并提高译码效率。上述译码方法过程中，选取的是每次对Rn个节点进行扩展。因此，在译码过程中，保留路径数Rn决定了译码过程中保留的信息，对于OvTDM***和OvFDM***，每次扩展舍弃的路径数量为MK-Rn，因此Rn越小那相应地译码复杂度就越低，但是Rn不可能无限小，因为Rn越小，译码性能损失越大，相同误码率条件下需要更高的信噪比。因此Rn选取也十分关键，要保证在降低复杂度的同时，使得译码性能损失较小，经试验表明，一般选取Rn小于MK-1，且大小或等于MK-4，其中K为接收符号的重叠次数，M表示M维***(M是大于等 于2的整数)，此时大幅降低译码复杂度的同时还保证了译码性能。类似地，对于编码支路数为K’和编码约束长度为L的OvCDM***，每次扩展舍弃的路径数量为MLK’-Rn，一般选取Rn小于ML(K’-4)，且大小或等于ML(K’-2)时，能大幅降低译码复杂度的同时还保证了译码性能。
相应地，本申请还提出了一种适用于OvXDM***的快速译码装置，如图2所示，其包括第一计算模块01、第一排序模块03、Rn个距离存储器05、Rn个路径存储器07、扩展模块09、第二计算模块11、第二排序模块13和判决输出模块15。
第一计算模块01用于分别计算前r个符号所潜在的全部条路径与前r个接收符号之间的测度。
第一排序模块03用于对第一计算模块01计算得到各测度进行排序。
Rn个距离存储器05用于分别存储第一排序模块中得到的较小的Rn个测度，对应的Rn个路径存储器07用于分别存储上述Rn个测度对应的路径。在一较优实施例中，Rn小于MK-1，其中M表示***的维度，取值是大于等于2的整数。
扩展模块09用于对当前存储的每条路径的最后一个节点进行M维扩展。或者，在一些实施例中，仅对当前存储的测度中最小值对应的路径进行扩展。
第二计算模块11用于对扩展出的路径计算其与接收符号序列中对应接收符号之间的瞬时测度，并将各瞬时测度与其前一时刻对应的累加测度相加，得到当前时刻相加后的各路径的累加测度。在一较优实施例中，权重因子模块17用于在第二计算模块11将各瞬时测度与前一时刻对应的累加测度相加时，先对前一时刻累加测度乘以权重因子，从而使得前一时刻累加测度先与权重因子相乘后再与瞬时测度相加。引入权重因子模块17是为了使得随着路径深度的增加，逐渐弱化距离当前节点较远的节点测度的参考，使得译码准确度更高。在一较优实施例中，权重因子的取值为大于0且小于等于1。
第二排序模块13用于对第二计算模块11中得到的所述相加后的各路径的累加测度进行排序，其中较小的Rn个测度及其各自对应的路径用于更新所述Rn个距离存储器05及对应的Rn个路径存储器07中的值。扩展模块09、第二计算模块11和第二排序模块13重复进行工作，直到扩展模块09扩展到接收符号序列中 最后一个符号对应的节点以使得Rn个距离存储器05及对应的Rn个路径存储器07分别存储有对应整个接收符号序列的较小的Rn个测度及其各自对应的路径时停止。当然，对于扩展模块09仅对当前存储的测度中最小值对应的路径进行扩展的情况，第二计算模块11对扩展模块09扩展出的路径计算其与对应接收符号之间的瞬时测度，比较输出模块15比较各瞬时测度，并将最小的瞬时测度对应的路径作为译码路径；否则扩展模块09、第二计算模块11和第二排序模块13重复进行工作。
判决输出模块15选取存储的测度最小的距离存储器05对应的路径存储器07中存储的路径作为译码路径，以进行判决输出。
在上述快速译码装置中，Rn为一正整数，其根据需求进行预设，且小于OvXDM***对应的格状图的节点数。在一实施例中，r为logMRn的数值向下取整，其中M表示***的维度，为大于或等于2的整数。
在一实施例中，当OvXDM***为OvTDM***、OvFDM***时，Rn小于MK-1，且大小或等于MK-4，其中K为接收符号的重叠次数，其中M表示***的维度，其是大于或等于2的整数；当OvXDM***为OvCDM***时，所述Rn小于ML(K’-4)，且大小或等于ML(K’-2)，其中K’为接收符号的编码支路数，L为接收符号的编码约束长度，同样M表示***的维度，其是大于或等于2的整数。
本申请还公开了一种OvXDM***，其包括上述的适用于OvXDM***的快速译码装置，在一实施例中，OvXDM***可以为OvTDM***、OvFDM***、OvCDM、OvSDM或OvHDM***。
下面再通过若干个实施例对本申请作进一步的说明。
实施例一
本实施例不妨以OvFDM***为例进行说明。
如图3所示，为OvFDM***发送端，其首先将频域信号按照一定的规律进行编码，然后将频域信号转换为时域信号即进行傅氏反变换，之后才将信号发送出去。具体地，先根据设计参数生成一个初始包络波形；然后根据重叠复用次数将上述初始包络波形在频域上按预定的频谱间隔进行移位，得到各子载波包络波 形；再将输入数据序列与各自对应的子载波包络波形相乘，得到各子载波的调制包络波形；再将各子载波的调制包络波形在频域上进行叠加，得到频域上的复调制包络波形，最后将上述频域上的复调制包络波形变换为时域上的复调制包络波形以发送，其中频谱间隔为子载波频谱间隔△B，其中子载波频谱间隔△B＝B/K，B为所述初始包络波形的带宽，K为重叠复用次数。如图4所示为OvFDM***接收端，其通过天线收到的信号是时域的信号，如果要对接收信号译码，首先需要将时域信号转换为频域信号，即进行傅氏变换之后才能处理。具体地，先对接收信号在时间域形成符号同步；然后对各个符号时间区间的接收信号进行取样、量化，使它变成接收数字信号序列；将时域信号转换为频域信号，再对该频率域信号以频谱间隔△B分段，形成实际接收信号分段频谱；再形成接收信号频谱与发送的数据符号序列之间的一一对应关系，最后根据此种一一对应的关系，检测数据符号序列。OvFDM***中的傅氏反变换和傅氏变换都涉及采样点数的设置，两者的采样点数应保持一致，且取值为2n，n为正整数。
一个重叠次数为K的2维调制OvFDM***，即M＝2，其格状图的节点数为2K-1，传统译码算法中需要的距离存储器为2K-1个，路径存储器也为2K-1个，如图5所示，其为一个接收信号长度N＝4，重叠次数K＝3时***对应的码树图，从图中可以看出，码树图的状态数为2K-1＝4，相应地，如图6所示，对应的格状图的节点数也为4，当译码状态充分展开时，每个译码步骤共有8条路径。本实施例中，选取距离存储器05为Rn个，路径存储器07也为Rn个，其中Rn小于2K-1，从而降低译码的复杂度，在一实施例中，令r为log2Rn的数值向下取整。下面详述译码步骤。
(1)确定初始前r个符号路径和对应测度。
长度为r的符号序列表示为：
Figure PCTCN2017092067-appb-000003
对于2维调制，共有2r种可能组合信息，组合形式为
Figure PCTCN2017092067-appb-000004
因此可到2r*r维的矩阵，表示为msg(2r*r)，其中每一行表示一组长度为r的符号序列。分别计算每组符号序列与前r个接收符号的瞬时测度，定义为：
Figure PCTCN2017092067-appb-000005
其中Vr为前r个接收 符号，xr,k为OvFDM的窗函数，
Figure PCTCN2017092067-appb-000006
为msg(2r*r)中第i行中第r-k个符号。此时译码到达r个节点，对应的瞬时测度为dr,i，对应的2r条译码路径为msg(2r*r)的第i行符号，将瞬时测度和路径分别存入距离存储器05和路径存储器07。
(2)对2r个节点进行扩展，每个节点可以同时进行2维扩展，当输入为+1时向上扩展，输入为－1时向下扩展，如图7所示，计算扩展到达的节点的瞬时测度：
Figure PCTCN2017092067-appb-000007
将瞬时测度与该路径前一时刻的累加测度相加，得到2r+1条路径的测度。
(3)对2r+1个路径测度进行排序，保留路径测度较小的前Rn个节点及其路径，将后面几个测度较大的节点舍弃；。
(4)继续对Rn个节点进行二维扩展，并计算扩展到达节点的瞬时测度，将瞬时测度与该节点前一时刻的累加测度相加，不管格状图是否完全展开，从第n步开始只保留前Rn个到达节点及其路径测度，具有较大测度的路径及其距离全部抛弃，每次对剩余的Rn个节点进行扩展，如此继续进行直至格状图的数据帧结束。
从以上步骤可以看出，本申请提出的快速译码方法每次仅需对Rn个节点进行扩展，而传统方法需要对2K-1个节点进行扩展，在一实施例中，当K较大时，Rn可以取值相对较小，并且满足Rn＜2K-1，因此对于长度为N的译码序列，译码复杂度大幅降低。需要说明的是，本实施例为描述方便，是以2维调制***进行说明本申请的快速译码方法，实际上，本方法适用于M维调制的***，其中M可以为大于或等于2的整数。
下面再以一个实际的例子说明本实施例。
以重叠次数K＝5的二维调制的OvFDM***为例，其采用矩形复用窗H＝{1 1 1 1 1}进行调制。需要说明的是，本申请的快速译码文法，适用于各种复用窗函数。因此，当格状图完全展开后共有2K-1＝16个节点，本例子选取Rn＝4，因此r为log2Rn的数值向下取整，即为2，需要说明的是，本申请的快速译码方法中，Rn的取值只要小于2K-1即可降低译码复杂度。
不妨令发送码序列为xi＝{+1 +1 -1 +1 -1 +1 +1 +1 -1 +1}，经过OvFDM*** 调制后，得到的接收序列为yi＝{+1 +2 +1 +2 +1 +1 +1 +3 +1 +3}。
如图6所示，其采用本申请的快速译码方法如下：
(1)由于r＝2，因此首先计算两个符号能够组合得到的序列组合分别为：U1＝{+1 +1}，U2＝{+1 -1}，U3＝{-1 +1}，U4＝{-1 -1}，将接收序列的前两个符号{yi(1) yi(2)}分别与U1、U2、U3、U4计算测度，得到d1、d2、d3、d4，将其存入距离存储器05，同时将与之对应的路径U1、U2、U3、U4存入路径存储器07，为方便行文，下面以d1、d2、d3、d4指代上述的4个距离存储器05，以U1、U2、U3、U4指代上述的4个路径存储器07。
(2)对当前的四个节点进行扩展，每个节点对应一条路径，对其进行二维扩展，因此可以得到8条潜在路径，分别为S1＝{U1 +1}、S2＝{U1 -1}、S3＝{U2 +1}、S4＝{U2 -1}、S5＝{U3 +1}、S6＝{U3 -1}、S7＝{U4 +1}、S8＝{U4 -1}。接着计算潜在路径对应测度：
Figure PCTCN2017092067-appb-000008
其中dprev表示扩展前的节点对应的累加测度；
(3)对di′进行排序，保留较小的4个路径瞬时距离，将其更新到距离存储器d1、d2、d3、d4中，并将其对应的路径Si更新到路径存储器U1、U2、U3、U4中，此时格状图中仍只到达4个节点，同时对应4条路径。
(4)重复步骤(2)和(3)，对当前的四个节点进行扩展，每个节点对应一条路径，对其进行二维扩展，因此可以得到8条潜在路径，分别为S1＝{U1 +1}、S2＝{U1 -1}、S3＝{U2 +1}、S4＝{U2 -1}、S5＝{U3 +1}、S6＝{U3 -1}、S7＝{U4 +1}、S8＝{U4 -1}，接着计算潜在路径对应测度：
Figure PCTCN2017092067-appb-000009
其中dprev表示扩展前的节点对应的累加测度；
(5)对di′进行排序，保留较小的4个路径的测度，将其更新到距离存储器d1、d2、d3、d4中，并将其对应的路径Si更新到路径存储器U1、U2、U3、U4中，此时格状图中仍只到达4个节点，同时对应4条路径。
(6)直到最后一个符号计算完毕，此时路径存储器U1、U2、U3、U4和距离存储器d1、d2、d3、d4分别存储了四条路径及其对应的测度，最小距离对应的路径即为最终的译码结果。
图9和图10分别给出了本例子的快速译码方法与目前的维特比译码方法的性能对比和运算时间对比，从图中可以明显看出本例子的快速译码方法性能损失不到1dB，但是时间却大幅压缩，即在能大幅降低译码复杂度、提高译码效率的同时还保证了译码性能。
实施例二
本实施例不妨以OvTDM***为例进行说明。
如图11所示，为OvTDM***发送端，先根据设计参数生成一个时域内的初始包络波形；然后根据重叠复用次数将上述初始包络波形在时域上按预定的时间间隔进行移位，得到各个时刻的偏移包络波形；再将输入数据序列与各个时刻的偏移包络波形相乘，得到各个时刻的调制包络波形；再将各个时刻的调制包络波形在时域上进行叠加，得到时域上的复调制包络波形以发送，其中时间间隔为△t，△t＝T/K，T为所述初始包络波形的时域宽度，K为重叠复用次数。如图12所示，为OvTDM***接收端，其对每一帧内的接收信号形成接收数字信号序列，再对形成的接收数字信号序列实施检测，以得到所述帧长内的调制在全部符号上的调制数据的判决。
一个重叠次数为K的2维调制OvTDM***，其格状图的节点数为2K-1，传统译码算法中需要的距离存储器为2K-1个，路径存储器也为2K-1个，如图13所示，其为一个接收信号长度N＝4，重叠次数K＝3时***对应的码树图，从图中可以看出，码树图的状态数为2K-1＝4，相应地，如图14所示，对应的格状图的节点数也为4，当译码状态充分展开时，每个译码步骤共有8条路径。本实施例中，选取距离存储器05为Rn个，路径存储器07也为Rn个，其中Rn小于2K-1，从而降低译码的复杂度，在一实施例中，令r为log2Rn的数值向下取整。下面详述译码步骤。
(1)长度为r的符号序列表示为：
Figure PCTCN2017092067-appb-000010
对于2维调制，共有2r种可能组合信息，组合形式为
Figure PCTCN2017092067-appb-000011
因此可到2r*r维的矩阵，表示为msg(2r*r)，其中每一行表示一组长度为r的符号序列。分别计算每组符号序列与前r个接收符号的瞬时测度，定义为：
Figure PCTCN2017092067-appb-000012
其中vn为前r个接收符号，xr,k为OvTDM的窗函数，
Figure PCTCN2017092067-appb-000013
为msg(2r*r)中第i行中第r-k个符号。此时译码到达r个节点，对应的瞬时测度为dr,i，对应的2r条译码路径为msg(2r*r)的第i行符号，将瞬时测度和路径分别存入距离存储器05和路径存储器07。
(2)对2r个节点进行扩展，每个节点可以同时进行2维扩展，当输入为+1时向上扩展，输入为－1时向下扩展，如图15所示，计算扩展到达的节点的瞬时测度：
Figure PCTCN2017092067-appb-000014
将瞬时测度与该路径前一时刻的累加测度相加，得到2r+1条路径的测度。
(3)对2r+1个路径测度进行排序，保留路径测度较小的前Rn个节点及其路径，将后面几个测度较大的节点舍弃；。
(4)继续对Rn个节点进行二维扩展，并计算扩展到达节点的瞬时测度，将瞬时测度与该节点前一时刻的累加测度相加，不管格状图是否完全展开，从第n步开始只保留前Rn个到达节点及其路径测度，具有较大测度的路径及其距离全部抛弃，每次对剩余的Rn个节点进行扩展，如此继续进行直至格状图的数据帧结束。
从以上步骤可以看出，本申请提出的快速译码方法每次仅需对Rn个节点进行扩展，而传统方法需要对2K-1个节点进行扩展，在一实施例中，当K较大时，Rn可以取值相对较小，并且满足Rn＜2K-1，因此对于长度为N的译码序列，译码复杂度大幅降低。需要说明的是，本实施例为描述方便，是以2维调制***进行说明本申请的快速译码方法，实际上，本方法适用于M维调制的***，其中M可以为大于或等于2的整数。
下面再以一个实际的例子说明本实施例。
以重叠次数K＝5的二维调制的OvTDM***为例，其采用矩形复用窗H＝{1 1 1 1 1}进行调制。需要说明的是，本申请的快速译码文法，适用于各种复用窗函数。因此，当格状图完全展开后共有2K-1＝16个节点，本例子选取Rn＝4，因此r为log2Rn的数值向下取整，即为2，需要说明的是，本申请的快速译码方法中，Rn的取值只要小于2K-1即可降低译码复杂度。
不妨令发送码序列为xi＝{+1 +1 -1 +1 -1 +1 +1 +1 -1 +1}，经过OvTDM***调制后，得到的接收序列为yi＝{+1 +2 +1 +2 +1 +1 +1 +3 +1 +3}。
如图16所示，其采用本申请的快速译码方法如下：
(1)由于r＝2，因此首先计算两个符号能够组合得到的序列组合分别为：U1＝{+1 +1}，U2＝{+1 -1}，U3＝{-1 +1}，U4＝{-1 -1}，将接收序列的前两个符号{yi(1) yi(2)}分别与U1、U2、U3、U4计算测度，得到d1、d2、d3、d4，将其存入距离存储器05，同时将与之对应的路径U1、U2、U3、U4存入路径存储器07，为方便行文，下面以d1、d2、d3、d4指代上述的4个距离存储器05，以U1、U2、U3、U4指代上述的4个路径存储器07。
(2)对当前的四个节点进行扩展，每个节点对应一条路径，对其进行二维扩展，因此可以得到8条潜在路径，分别为S1＝{U1 +1}、S2＝{U1 -1}、S3＝{U2 +1}、S4＝{U2 -1}、S5＝{U3 +1}、S6＝{U3 -1}、S7＝{U4 +1}、S8＝{U4 -1}。接着计算潜在路径对应测度：
Figure PCTCN2017092067-appb-000015
其中dprev表示扩展前的节点对应的累加测度；
(3)对di′进行排序，保留较小的4个路径瞬时距离，将其更新到距离存储器d1、d2、d3、d4中，并将其对应的路径Si更新到路径存储器U1、U2、U3、U4中，此时格状图中仍只到达4个节点，同时对应4条路径。
(4)重复步骤(2)和(3)，对当前的四个节点进行扩展，每个节点对应一条路径，对其进行二维扩展，因此可以得到8条潜在路径，分别为S1＝{U1 +1}、S2＝{U1 -1}、S3＝{U2 +1}、S4＝{U2 -1}、S5＝{U3 +1}、S6＝{U3 -1}、S7＝{U4 +1}、S8＝{U4 -1}，接着计算潜在路径对应测度：
Figure PCTCN2017092067-appb-000016
其中dprev表示扩展前的节点对应的累加测度；
(5)对di′进行排序，保留较小的4个路径的测度，将其更新到距离存储器d1、d2、d3、d4中，并将其对应的路径Si更新到路径存储器U1、U2、U3、U4中，此时格状图中仍只到达4个节点，同时对应4条路径。
(6)直到最后一个符号计算完毕，此时路径存储器U1、U2、U3、U4和距离存储器d1、d2、d3、d4分别存储了四条路径及其对应的测度，最小距离对应的路径即为最终的译码结果。
图17和图18分别给出了本例子的快速译码方法与目前的维特比译码方法的性能对比和运算时间对比，其为计算机10000次蒙特卡罗仿真结果，从图中可以明显看出本例子的快速译码方法性能损失不到1dB，但是去处时间却大幅压缩，即在能大幅降低译码复杂度、提高译码效率的同时还保证了译码性能
实施例三
本实施例不妨以OvCDM***为例进行说明。
OvCDM***的重叠码分复用的核心是重叠和复用，目的是提高通信***的频谱效率。OvCDM***将卷积编码系数推广到复数域的广义卷积编码模型，通过符号重叠产生约束关系，主要参数包括编码支路数K’路和编码约束长度L，其***结构图如图19所示，对应的编码器结构如附图20所示。OvCDM***的关键是编码矩阵，即卷积扩展系数，要求其满足线性关系，此时输入序列与输出序列一一对应，因此理论上可以无误码的解码，一般通过计算机搜索所有测度较大的矩阵作为编码矩阵，其编码矩阵排列如附图21所示。
为了说明本实施例的快速译码方法，先给出OvCDM***的编码过程。
(1)将待发送数据经过串并转换成为K’路子数据流，第i路上的数据流记为ui＝ui,0ui,1ui,2…。比如K’＝2时，u0＝u0,0u0,1u0,2…，u1＝u1,0u1,1u1,2…，
(2)将每一路数据送入一个移位寄存器进行加权叠加，第i路的加权系数为bi＝bi,0bi,1bi,2…，其为一复向量。
(3)把各路信号相加输出，得到最终OvCDM编码器的输出为c＝c0c1c2…，
Figure PCTCN2017092067-appb-000017
OvCDM的码率为
Figure PCTCN2017092067-appb-000018
其中n为子数据流长度。当n很长时，由移位寄存器拖尾所带来的码率损失可以忽略不计，于是有rOVCDM≈k。
传统的二元域卷积编码模型码率一般小于1，会导致频谱效率损失。而OvCDM的复数域的卷积编码码率等于1，单路的卷积编码扩展不会导致频谱效率损失，还会增加额外的编码增益。
接收端收到信号后，先对信号进行同步、信道估计、数字化处理，再对处理后的数据进行快速译码。译码算法的核心是通过计算接收信号与理想状态的测度，采用路径存储器和测度判决出最佳的译码路径，得到最终的检测序列，序列检测过程框图如附图22所示。
本实施例适用于OvCDM***的快速译码方法，具体步骤如下。
不妨令OvCDM***的编码支路数为K’路和编码约束长度L，编码输出矢量的维数为N，采用二维调制，因此码元为+1、－1。
(1)初始状态：
令初始节点状态即l＝0的路径测度为d0,0＝0；
(2)计算节点测度：
每个节点共计包括S个状态，对第l个节点求其测度，方法为计算全部m条从前一状态转移到此状态的理想信号波形与接收信号序列
Figure PCTCN2017092067-appb-000019
之间的测度ds,m(l,l+1)，其表达式为
Figure PCTCN2017092067-appb-000020
(3)计算累加测度：
将当前节点各个状态S的测度ds,m(l,l+1)与它们各自出发状态S'的测度ds',l-1相加，形成新的m个路径的累加测度。
(4)路径筛选：
对路径测度进行排序，选择路径测度较小的Rn条路径，将其对应的路径存入 路径存储器，测度存入距离存储器。其余路径被舍弃，下一阶段从保留路径进行扩展。
(5)最终路径确定：
重复步骤(2)-(4)直至译码结束，此时存储器中共保留了Rn条路径及其对应的路径测度，测度最小的路径即为译码结果。
下面再以一个实例进行说明。
本案例以输入数据流为u＝{+1,-1,-1,-1,-1,+1,-1,+1,+1,-1,+1,-1,-1,-1,-1,+1}，K’＝2，L＝2，编码矩阵
Figure PCTCN2017092067-appb-000021
为例说明OvCDM***译码过程，该参数设计下对应OvCDM***对应的格状图如图23所示。
(1)编码过程
首先对于K’＝2，将输入数据流u转换为两路，对应为：
u1＝{+1,-1,-1,-1,+1,+1,-1,-1}
u2＝{-1,-1,+1,+1,-1,-1,-1,+1}
每一路的卷积系数表示为：b0＝[1 j],b1＝[j 1]，参照附图20编码结构和附图23的格状图，其编码输出为c＝{1-j,-2,-2,0,2-2j,0,-2,-2}。OvCDM***的编码实现过程也可以采用其他方式，本申请的发明点在于译码过程而不在于编码过程。
(2)译码过程，如图24所示：
经过信号同步、信道估计、数字化处理后，接收端得到信号序列，为方便说明，假定为理想状态，此时接收信号序列为c＝{1-j,-2,-2,0,2-2j,0,-2,-2}，利用接收信号进行译码：
接收到的第一个符号为1-j,将其与理想的四个状态(1,1)，(1，-1)，(-1,1)，(-1，-1)分别求测度，此时，前3个较小的测度对应的路径从小到达依次为(1,-1)、(1,1)、(-1，-1)。将对应的前3个较小的测度及其对应的路径分别存入距离存储器05和路径存储器07中。
接着处理第2个符号-2，当前保留的路径对应的状态分别为(1,-1)、(1,1)、(-1，-1)，对每个状态进行扩展，每个状态可以进行四维扩展，因此得到12条路径，计算每条路径与当前接收符号的测度，并对这12条路径的测度进行排序， 保留前3个较小的测度，更新存储器，将测度与对应的路径分别存入距离存储器05和路径存储器07中。
后面的符号采用相同的方法，通过求累加求得12条测度，并保留测度最小的三条路径及其对应的测度。直至最后一个符号计算完毕，此时得到3条路径及其对应的路径测度，测度最小的路径对应的序列即为最终译码输出序列，为(1，-1，-1，-1，-1，1，-1，1，1，-1，1，-1，-1，-1，-1，1)，译码结束。
在以上的实施例一到实施例三中，在对存储的路径进行扩展的时候，均是进行对存储的每条路径的最后一个节点进行扩展的方案，在以下的实施例中则仅对当前存储的最小测度对应的路径进行扩展。
实施例四
以OvFDM***为例，重叠次数为K的2维调制OvFDM***，即M＝2，其格状图的节点数为2K-1，传统译码算法中需要的距离存储器为2K-1个，路径存储器也为2K-1个，本实施例中使用K＝3的低重叠进行说明。
(1)确定初始前r个符号路径和对应测度。
长度为r的符号序列表示为
Figure PCTCN2017092067-appb-000022
对于2维调制，共有2r种可能组合信息，组合形式为
Figure PCTCN2017092067-appb-000023
因此可得到2r*r维的矩阵，表示为msg(2r*r)，其中每一行表示一组长度为r的符号序列。分别计算每组符号序列与前r个接收符号的瞬时测度，定义为：
Figure PCTCN2017092067-appb-000024
其中vr为前r个接收符号，xr,j为OvFDM***的窗函数，
Figure PCTCN2017092067-appb-000025
为msg(2r*r)中第i行中第(r-j)个符号。此时译码到达第r个节点，包含2r条译码路径U和2r个瞬时测度d，每条译码路径的节点深度均为r。将瞬时测度和对应的译码路径分别存入距离存储器05和路径存储器07。
(2)寻找测度最小值和最大值。
将(1)中得到的2r个瞬时测度依次排序，分别找出测度最小和最大对应的索引idxmin、idxmax，测度最小索引对应的瞬时测度为
Figure PCTCN2017092067-appb-000026
将测度最小对应的索引节点深度加1，节点深度变为r+1。
(3)扩展节点
对测度最小索引对应的译码路径进行扩展，可扩展为1和-1两条支路，即译码路径变为
Figure PCTCN2017092067-appb-000027
和
Figure PCTCN2017092067-appb-000028
扩展后的译码路径共有2r+1,但是我们仅需要保留2r条路径，因此将(2)中找到的最大测度索引对应的译码路径删除，将扩展为1的译码支路
Figure PCTCN2017092067-appb-000029
存储在索引为idxmax的存储器中(也可以用-1支路替代)，-1支路
Figure PCTCN2017092067-appb-000030
继续存储在索引为idxmin的存储器中。此时译码路径中只有idxmin、idxmax索引对应的节点深度为r+1，其余索引的节点深度仍为r。
(4)计算累加测度
将扩展后的两条路径分别与接收到的符号求瞬时测度，计算公式同(2)中计算测度的公式，得到两条路径的测度分别为d'r+1、d”r+1，将其分别与r节点对应的瞬时测度相加，得到累加测度。在相加过程，可以引入权重因子alpha,取值为大于0且小于等于1的数，具体数值视***需求而定，这样做的目的是随着节点深度的增加，逐渐弱化距离当前节点较远的节点测度参考，相加过程可表示为：
Figure PCTCN2017092067-appb-000031
得到的累加测度分别存储在(3)中对应的idxmin、idxmax存储器中。
对剩余的符号序列同样采用上述(2)～(4)的方式，进行筛选扩展，直到节点深度到达符号序列深度N，比较最后一次符号扩展后的瞬时测度，较小者即为最终输出的译码路径。
从以上步骤中可以看出，本专利中提出的快速译码方法每次仅需对Rn个节点进行扩展，而经典方法中需要对2K-1次个节点进行扩展，当K较大时，Rn可以取值相对较小，并且满足Rn<2K-1，因此对于长度为N的译码序列，译码复杂度大幅降低。
实施例五
本实施例以OvTDM***为例进行说明。对于一个二维调制的OvTDM***，从格状图的原点开始，不管重叠重数K多大，它只有两个到达节点。分别计算到达这两个节点支路的瞬时测度，从两个到达节点的瞬时路径测度中选取最小者进 行扩展，分别计算扩展两支路的瞬时测度，并记下这两个到达节点。每个时刻最多只保留Rn条到达节点及其路径瞬时测度，每次均从具有最小瞬时路径测度的节点扩展，每扩展一次增加一个到达节点，具有较大瞬时路径测度的路径及其测度全部抛弃。如此继续进行直至格状图的数据帧结束，其具有最小瞬时路径测度的到达节点的路径就是判决输出。下面详述译码步骤，本实施例中不妨令Rn＝2r。
(1)确定初始前r个符号路径和对应测度。
长度为r的符号序列表示为
Figure PCTCN2017092067-appb-000032
对于2维调制，共有2r种可能组合信息，组合形式为
Figure PCTCN2017092067-appb-000033
因此可得到2r*r维的矩阵，表示为msg(2r*r)，其中每一行表示一组长度为r的符号序列。分别计算每组符号序列与前r个接收符号的瞬时测度，定义为：
Figure PCTCN2017092067-appb-000034
其中vr为前r个接收符号，xr,j为OvTDM***的窗函数，
Figure PCTCN2017092067-appb-000035
为msg(2r*r)中第i行中第(r-j)个符号。此时译码到达第r个节点，包含2r条译码路径U和2r个瞬时测度d，每条译码路径的节点深度均为r。将瞬时测度和对应的译码路径分别存入距离存储器05和路径存储器07。
(2)寻找测度最小值和最大值。
将(1)中得到的2r个瞬时测度依次排序，分别找出测度最小和最大对应的索引idxmin、idxmax，测度最小索引对应的瞬时测度为
Figure PCTCN2017092067-appb-000036
将测度最小对应的索引节点深度加1，节点深度变为r+1。
(3)扩展节点
对测度最小索引对应的译码路径进行扩展，可扩展为1和-1两条支路，即译码路径变为
Figure PCTCN2017092067-appb-000037
和
Figure PCTCN2017092067-appb-000038
扩展后的译码路径共有2r+1,但是我们仅需要保留2r条路径，因此将(2)中找到的最大测度索引对应的译码路径删除，将扩展为1的译码支路
Figure PCTCN2017092067-appb-000039
存储在索引为idxmax的存储器中(也可以用-1支路替代)，-1支路
Figure PCTCN2017092067-appb-000040
继续存储在索引为idxmin的存储器中。此时译码路径中只有idxmin、idxmax索引对应的节点深度为r+1，其余索引的节点深度仍为r。
(4)计算累加测度
将扩展后的两条路径分别与接收到的符号求瞬时测度，计算公式同(2)中计算测度的公式，得到两条路径的测度分别为d'r+1、d”r+1，将其分别与r节点对应的瞬时测度相加，得到累加测度。在相加过程，可以引入权重因子alpha,取值为0～1的数，具体数值视***需求而定，这样做的目的是随着节点深度的增加，逐渐弱化距离当前节点较远的节点测度参考，相加过程可表示为：
Figure PCTCN2017092067-appb-000041
得到的累加测度分别存储在(3)中对应的idxmin、idxmax存储器中。
对剩余的符号序列同样采用上述(2)～(4)的方式，进行筛选扩展，直到节点深度到达符号序列深度N，比较最后一次符号扩展后的瞬时测度，较小者即为最终输出的译码路径。
从以上步骤中可以看出，本专利中提出的快速译码方法每次仅需对Rn个节点进行扩展，而经典方法中需要对2K-1次个节点进行扩展，当K较大时，Rn可以取值相对较小，并且满足Rn<2K-1，因此对于长度为N的译码序列，译码复杂度大幅降低。
实施例六
OvCDM***的结构再次不在赘述，本实施例中不妨令OvCDM***的编码支路数为K’路和编码约束长度L，编码输出矢量的维数为N，采用二维调制，因此码元为+1、－1。由于编码支路数为K’路和编码约束长度为L，则每个状态可到达的节点数m为2K’，共计包含2K’个状态S，因此所有状态可到达的节点数共为22K’。本实施例中，选取距离存储器05为Rn个，路径存储器07也为Rn个，其中Rn小于2K’-1，从而降低译码的复杂度，在一实施例中，令r为log2Rn的数值向下取整。
(1)初始状态：
令初始节点状态(l＝0)的路径测度为d0,0＝0。
(2)确定初始前r个符号路径和对应测度
长度为r的符号序列表示为
Figure PCTCN2017092067-appb-000042
对于2维调制，共有2r种可能组合信息，组合形式为
Figure PCTCN2017092067-appb-000043
因此可得到2r*r维的矩阵，表示为msg(2r*r)， 其中每一行表示一组长度为r的符号序列。分别计算每组符号序列与前r个接收符号的瞬时测度，定义为：
Figure PCTCN2017092067-appb-000044
其中vr为前r个接收符号，xr,j为OvCDM的编码矩阵，
Figure PCTCN2017092067-appb-000045
为msg(2r*r)中第i行中第(r-j)个符号。此时译码到达第r个节点，包含2r条译码路径U和2r个瞬时测度d，每条译码路径的节点深度均为r。将瞬时测度和对应的译码路径分别存入距离存储器05和路径存储器07。
(3)计算节点测度：
每个节点共计包括S个状态，对第l个节点求其测度，方法为计算全部m条从前一状态转移到此状态的理想信号符号与接收信号序列
Figure PCTCN2017092067-appb-000046
之间的测度ds,m(l,l+1)，其表达式为
Figure PCTCN2017092067-appb-000047
(4)计算累加测度：
将当前节点各个状态S的欧氏距离ds,m(l,l+1)与它们各自出发状态S'的测度ds',l-1相加，形成新的m个路径的累加欧氏距离。在相加过程，可以引入权重因子alpha,取值为0～1的数，具体数值视***需求而定，这样做的目的是随着节点深度的增加，逐渐弱化距离当前节点较远的节点测度参考。
(5)路径筛选：
对累加的欧氏距离进行排序，选择累加欧氏距离最小的路径对其进行扩展，可扩展出m条支路，对这m条支路依次求当前节点测度，保留测度最小的路径，同时舍弃其余(m-1)条路径，此时存储器中共包含Rn条路径及其对应的测度。
(6)最终路径确定：
重复步骤(3)-(5)直至译码结束，此时存储器中共保留了Rn条路径及其对应的路径欧氏距离，欧氏距离最小的路径即为译码结果。
本实施例中，保留路径数Rn决定了译码过程保留的信息，对于编码支路数为K’的OvCDM***，舍弃的路径数量为22K’-Rn，因此Rn越小译码的复杂度越低。但Rn不可能无限小，Rn越小，译码性能损失越大，相同误码率条件下需要更高的 信噪比。因此需要根据实际的***及信道，选择合适的Rn，在降低译码复杂度的同时，保证译码的性能损失较小。一般选择Rn的值为大于等于2L(K’-4)且小于等于2L(K’-2)，此时能够保证译码性能同时大幅降低译码复杂度。
本申请提出的快速译码方法及装置，除了应用在OvXDM***中，也可广泛应用于实际移动通信***中，如TD-LTE、TD-SCDMA等***，还可以广泛应用于卫星通信、微波视距通信、散射通信、大气层光通信、红外通信与水生通信等任何无线通信***中。本申请的快速译码方法及装置既可以应用于大容量无线传输，也可以应用于小容量的轻型无线电***。
本专利中提出的快速译码方法，在译码的路径扩展期间，对待扩展的节点进行筛选，由于正确路径必定为较优的几个路径之一，因此只需选出路径较优的节点进行扩展，将较差的节点舍弃，后期不再对其进行扩展，从而降低译码的复杂度，提高了***的译码效率
以上内容是结合具体的实施方式对本申请所作的进一步详细说明，不能认定本申请的具体实施只局限于这些说明。对于本申请所属技术领域的普通技术人员来说，在不脱离本申请发明构思的前提下，还可以做出若干简单推演或替换。

Claims (18)

  1. 一种适用于OvXDM***的快速译码方法，其特征在于，包括以下步骤：
    步骤一、分别计算前r个符号所潜在的全部路径与接收符号序列中前r个接收符号之间的测度；所述r小于所有符号的个数；
    步骤二、对计算得到各测度进行排序，并存储其中较小的Rn个测度及其各自对应的路径；
    步骤三、对当前存储的每条路径的最后一个节点进行扩展，对扩展出的路径计算其与接收符号序列中对应接收符号之间的瞬时测度，并将各瞬时测度与其前一时刻对应的累加测度相加，得到当前时刻相加后的各路径的累加测度；所述瞬时测度为：
    Figure PCTCN2017092067-appb-100001
    其中Vr为前r个接收符号，
    Figure PCTCN2017092067-appb-100002
    为窗函数，
    Figure PCTCN2017092067-appb-100003
    为msg(2r*r)中第i行中第r-k个符号；
    步骤四、对所述相加后的各路径的累加测度进行排序，
    并存储其中较小的Rn个测度及其各自对应的路径；
    步骤五、当步骤三中扩展到接收符号序列中最后一个符号对应的节点，步骤四相应地存储有对应整个接收符号序列的较小的Rn个测度及其各自对应的路径时停止，否则重复步骤三和步骤四；
    步骤六、选取测度最小的一条路径作为译码路径，以进行判决输出；
    其中，Rn为一正整数，且小于OvXDM***对应的格状图的节点数。
  2. 如权利要求1所述的适用于OvXDM***的快速译码方法，其特征在于，所述OvXDM***为OvTDM***、OvFDM***、OvCDM***、OvSDM***或OvHDM***。
  3. 如权利要求2所述的适用于OvXDM***的快速译码方法，其特征在于， 当所述OvXDM***为OvTDM***或OvFDM***时，所述Rn小于MK-1，且大小或等于MK-4，其中K为接收符号的重叠次数，M表示维度，M的取值是大于等于2的整数；当所述OvXDM***为OvCDM***时，所述Rn小于ML(K’-4)，且大小或等于ML(K’-2)，其中K’为接收符号的编码支路数，L为接收符号的编码约束长度。
  4. 如权利要求1至3中任一项所述的适用于OvXDM***的快速译码方法，其特征在于，r为logMRn的数值向下取整，其中M表示维度，取值是大于等于2的整数。
  5. 一种适用于OvXDM***的快速译码装置，其特征在于，包括：
    第一计算模块，用于分别计算前r个符号所潜在的全部路径与前r个接收符号之间的测度；
    第一排序模块，用于对计算得到各测度进行排序；
    Rn个距离存储器及对应的Rn个路径存储器，分别用于存储第一排序模块中得到的较小的Rn个测度及其各自对应的路径；
    扩展模块，用于对当前存储的每条路径的最后一个节点进行M维扩展；
    第二计算模块，对扩展出的路径计算其与接收符号序列中对应接收符号之间的瞬时测度，并将各瞬时测度与其前一时刻对应的累加测度相加，得到当前时刻相加后的各路径的累加测度；所述瞬时测度为：
    Figure PCTCN2017092067-appb-100004
    其中Vr为前r个接收符号，
    Figure PCTCN2017092067-appb-100005
    为窗函数，
    Figure PCTCN2017092067-appb-100006
    为msg(2r*r)中第i行中第r-k个符号；
    第二排序模块，用于对第二计算模块中得到的所述相加后的各路径的累加测度进行排序，其中较小的Rn个测度及其各自对应的路径用于更新所述Rn个距离存储器及对应的Rn个路径存储器中的值；扩展模块、第二计算模块和第二排序模 块重复进行工作，直到扩展模块扩展到接收符号序列中最后一个符号对应的节点以使得Rn个距离存储器及对应的Rn个路径存储器分别存储有对应整个接收符号序列的较小的Rn个测度及其各自对应的路径时停止；
    判决输出模块，选取存储的测度最小的距离存储器对应的路径存储器中存储的路径作为译码路径，以进行判决输出；
    其中，Rn为一正整数，其根据需求进行预设，且小于OvXDM***对应的格状图的节点数。
  6. 如权利要求5所述的适用于OvXDM***的快速译码装置，其特征在于，当所述OvXDM***为OvTDM***、OvFDM***时，所述Rn小于MK-1，且大小或等于MK-4，其中K为接收符号的重叠次数，M表示维度，取值是大于等于2的整数。
  7. 如权利要求5所述的适用于OvXDM***的快速译码装置，其特征在于，当所述OvXDM***为OvCDM***时，所述Rn小于ML(K’-4)，且大小或等于ML(K’-2)，其中K’为接收符号的编码支路数，L为接收符号的编码约束长度。
  8. 如权利要求5-7任一项所述的适用于OvXDM***的快速译码装置，其特征在于，r为logMRn的数值向下取整，其中M表示维度，取值是大于等于2的整数。
  9. 一种OvXDM***，其特征在于，包括如权利要求5至8中任一项所述的适用于OvXDM***的快速译码装置。
  10. 如权利要求9所述的OvXDM***，其特征在于，所述OvXDM***为OvTDM***、OvFDM***、OvCDM***、OvSDM***或OvHDM***。
  11. 一种OvXDM***的快速译码方法，其特征在于，包括以下步骤：
    步骤一、分别计算前r个符号所潜在的全部路径与接收符号序列中前r个接 收符号之间的测度；
    步骤二、对计算得到的各测度进行排序，并存储其中较小的Rn个测度及其各自对应的路径；
    步骤三、对当前存储的最小测度对应的路径进行扩展，对扩展出的路径计算其与对应接收符号之间的瞬时测度，并将各瞬时测度与前一时刻对应的累加测度相加，得到当前时刻相加后的各扩展路径的累加测度；
    步骤四、对所述各扩展路径的累加测度与存储的其余未进行扩展的Rn-1个测度进行排序，并存储其中较小的Rn个测度及其各自对应的路径；
    步骤五、当步骤三中当前存储的最小测度对应的路径扩展后到达接收符号序列的深度时，则对扩展出的路径计算其与对应接收符号之间的瞬时测度，并比较各瞬时测度，将最小的瞬时测度对应的路径作为译码路径；否则重复步骤三和步骤四；
    其中Rn为一正整数，且小于高重OvXDM***对应的格状图的节点数。
  12. 如权利要求11所述的OvXDM***的快速译码方法，其特征在于，在步骤三中，将各瞬时测度与前一时刻对应的累加测度相加时，累加测度先与权重因子相乘后再与瞬时测度相加。
  13. 如权利要求2所述的OvXDM***的快速译码方法，其特征在于，所述权重因子的取值为大于0且小于等于1。
  14. 如权利要求1所述的OvXDM***的快速译码方法，其特征在于，r为logMRn的数值向下取整，其中M表示***的维度，取值是大于等于2的整数。
  15. 一种OvXDM***的快速译码装置，其特征在于，包括：
    第一计算模块，用于分别计算前r个符号所潜在的全部路径与接收符号序列中前r个接收符号之间的测度；
    第一排序模块，用于对计算得到的各测度进行排序；
    Rn个距离存储器及对应的Rn个路径存储器，分别用于存储第一排序模块中 得到的较小的Rn个测度及其各自对应的路径；
    扩展模块，用于对当前存储的最小测度对应的路径进行扩展；
    第二计算模块，用于对扩展模块扩展出的路径计算其与对应接收符号之间的瞬时测度，并将各瞬时测度与前一时刻对应的累加测度相加，得到当前时刻相加后的各扩展路径的累加测度；
    第二排序模块，用于对第二计算模块中计算得到的所述各扩展路径的累加测度与存储的其余未进行扩展的Rn-1个测度进行排序，其中较小的Rn个测度及其各自对应的路径用于更新所述Rn个距离存储器及对应的Rn个路径存储器的值；
    比较输出模块；当扩展模块对当前存储的最小测度对应的路径进行扩展达到接收符号序列的深度时，第二计算模块对扩展模块扩展出的路径计算其与对应接收符号之间的瞬时测度，比较输出模块比较各瞬时测度，并将最小的瞬时测度对应的路径作为译码路径；否则扩展模块、第二计算模块和第二排序模块重复进行工作；
    其中Rn为一正整数，且小于OvXDM***对应的格状图的节点数。
  16. 如权利要求15所述的OvXDM***的快速译码装置，其特征在于，还包括权重因子模块，用于在第二计算模块将各瞬时测度与前一时刻对应的累加测度相加时，先对前一时刻累加测度乘以权重因子，从而使得前一时刻的累加测度先与权重因子相乘后再与瞬时测度相加。
  17. 如权利要求16所述的OvXDM***的快速译码装置，其特征在于，所述权重因子模块的权重因子的取值为大于0且小于等于1。
  18. 如权利要求15所述的OvXDM***的快速译码装置，其特征在于，r为logMRn的数值向下取整，其中M表示***的维度，取值是大于等于2的整数。
PCT/CN2017/092067 2016-07-22 2017-07-06 一种适用于OvXDM***的快速译码方法、装置及OvXDM*** WO2018014738A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
JP2019503461A JP6744982B2 (ja) 2016-07-22 2017-07-06 ＯｖＸＤＭシステムに適用される一種類のファストデコード方法、装置及びＯｖＸＤＭシステム
KR1020197005393A KR102203029B1 (ko) 2016-07-22 2017-07-06 OvXDM시스템에 적용되는 일종의 쾌속적 디코딩 방법, 장치 및 OvXDM 시스템
EP17830372.3A EP3490177A4 (en) 2016-07-22 2017-07-06 QUICK DECODING METHOD AND DEVICE SUITABLE FOR AN OVXDM SYSTEM, AND OVXDM SYSTEM
US16/254,548 US20190229838A1 (en) 2016-07-22 2019-01-22 Fast decoding method and device suitable for ovxdm system, and ovxdm system

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN201610587517.4A CN107645363B (zh) 2016-07-22 2016-07-22 一种适用于OvXDM***的快速译码方法、装置及OvXDM***
CN201610584674.X 2016-07-22
CN201610587517.4 2016-07-22
CN201610584674.XA CN107645359B (zh) 2016-07-22 2016-07-22 一种OvXDM***的快速译码方法、装置及OvXDM***

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/254,548 Continuation US20190229838A1 (en) 2016-07-22 2019-01-22 Fast decoding method and device suitable for ovxdm system, and ovxdm system

Publications (1)

Publication Number Publication Date
WO2018014738A1 true WO2018014738A1 (zh) 2018-01-25

Family

ID=60991982

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/092067 WO2018014738A1 (zh) 2016-07-22 2017-07-06 一种适用于OvXDM***的快速译码方法、装置及OvXDM***

Country Status (5)

Country Link
US (1) US20190229838A1 (zh)
EP (1) EP3490177A4 (zh)
JP (1) JP6744982B2 (zh)
KR (1) KR102203029B1 (zh)
WO (1) WO2018014738A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112087263B (zh) * 2019-06-14 2022-06-14 华为技术有限公司 一种光通信中信号收发的方法、光收发机和***
CN115250222B (zh) * 2021-04-26 2024-04-02 ***通信有限公司研究院 一种信号处理方法、装置、设备和可读存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6912257B1 (en) * 1999-04-28 2005-06-28 Samsung Electronics Co., Ltd. Apparatus and method for normalizing metric values in a component decoder in a mobile communication system
CN101510813A (zh) * 2008-02-15 2009-08-19 ***通信集团公司 通信***中传输、处理数据的方法、通信***及通信设备
CN101557364A (zh) * 2009-05-12 2009-10-14 山东大学 Turbo-OvCDM***的联合迭代信道估计和译码方法
CN101632248A (zh) * 2007-02-14 2010-01-20 北京清深技术开发中心有限公司 一种编码复用与多址传输方法
CN102185618A (zh) * 2011-01-27 2011-09-14 华为技术有限公司 一种维特比译码算法的改进方法和卷积码译码器
CN103780349A (zh) * 2014-01-22 2014-05-07 北京工业大学 一种基于有限距离译码的解随机化采样格译码方法

Family Cites Families (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR940012973A (ko) * 1992-11-26 1994-06-24 정용문 데이타 통신용 비터비 디코딩 방법
JPH08279765A (ja) * 1995-04-05 1996-10-22 Hitachi Ltd 畳込み符号ならびにトレリス符号用の復号アルゴリズムとそれを用いる受信装置
JP3280834B2 (ja) * 1995-09-04 2002-05-13 沖電気工業株式会社 符号化通信方式における信号判定装置および受信装置ならびに信号判定方法および通信路状態推定方法
US5905742A (en) * 1995-12-27 1999-05-18 Ericsson Inc. Method and apparauts for symbol decoding
JP2004056724A (ja) * 2002-07-24 2004-02-19 Toko Electric Corp 誤り訂正復号化方法
JP2004248216A (ja) * 2003-02-17 2004-09-02 Sony Ericsson Mobilecommunications Japan Inc 信号処理装置及び方法
JP4854378B2 (ja) * 2006-05-01 2012-01-18 ソフトバンクＢｂ株式会社 無線伝送システムおよび無線伝送方法
JP4482835B2 (ja) * 2008-03-14 2010-06-16 ソニー株式会社 データ処理装置、データ処理方法、及び、プログラム
US20110075649A1 (en) * 2009-09-13 2011-03-31 Research Institute Of Tsinghua University In Shenzhen Method and system of frequency division multiplexing
US20110103236A1 (en) * 2009-09-13 2011-05-05 Research Institute Of Tsinghua University In Shenzhen Transmission method of code division multiplexing and multiple access
US9537759B2 (en) * 2012-01-31 2017-01-03 Massachusetts Institute Of Technology Multi-path data transfer using network coding
CN106911424A (zh) * 2015-12-22 2017-06-30 深圳超级数据链技术有限公司 译码方法、装置及其***
CN106911421B (zh) * 2015-12-22 2019-09-13 深圳光启合众科技有限公司 译码方法、装置及其***

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6912257B1 (en) * 1999-04-28 2005-06-28 Samsung Electronics Co., Ltd. Apparatus and method for normalizing metric values in a component decoder in a mobile communication system
CN101632248A (zh) * 2007-02-14 2010-01-20 北京清深技术开发中心有限公司 一种编码复用与多址传输方法
CN101510813A (zh) * 2008-02-15 2009-08-19 ***通信集团公司 通信***中传输、处理数据的方法、通信***及通信设备
CN101557364A (zh) * 2009-05-12 2009-10-14 山东大学 Turbo-OvCDM***的联合迭代信道估计和译码方法
CN102185618A (zh) * 2011-01-27 2011-09-14 华为技术有限公司 一种维特比译码算法的改进方法和卷积码译码器
CN103780349A (zh) * 2014-01-22 2014-05-07 北京工业大学 一种基于有限距离译码的解随机化采样格译码方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3490177A4 *

Also Published As

Publication number Publication date
KR20190032524A (ko) 2019-03-27
EP3490177A4 (en) 2020-03-25
JP2019522437A (ja) 2019-08-08
KR102203029B1 (ko) 2021-01-13
EP3490177A1 (en) 2019-05-29
US20190229838A1 (en) 2019-07-25
JP6744982B2 (ja) 2020-08-26

Similar Documents

Publication Publication Date Title
WO2018068540A1 (zh) 基于重叠复用的调制解调方法和装置
US11082153B2 (en) Signal sampling and recovery method and apparatus applicable to OvXDM system, and OvXDM system
WO2018019044A1 (zh) 编码方法和设备
WO2018076773A1 (zh) 基于重叠复用的译码方法、装置及调制解调方法和***
WO2018014738A1 (zh) 一种适用于OvXDM***的快速译码方法、装置及OvXDM***
US10771303B2 (en) Overlapped multiplexing-based decoding method and device, and modulation and demodulation method and system
CN108833321B (zh) 基于差分相位波形匹配的编码cpm信号码块同步方法
CN107911124B (zh) 一种非递归的sc译码部分和确定方法及装置
CN109639290B (zh) 一种半随机分组叠加编码及译码方法
CN107645359B (zh) 一种OvXDM***的快速译码方法、装置及OvXDM***
WO2018014733A1 (zh) 一种适用于OvXDM***译码方法、装置及OvXDM***
CN107645363B (zh) 一种适用于OvXDM***的快速译码方法、装置及OvXDM***
US10707896B2 (en) Forward and backward smooth decoding method, device, and system
JP6144846B2 (ja) 無線通信装置及び無線通信システム
WO2018019109A1 (zh) 基于重叠复用的译码方法、装置和***
Zhang et al. Decoding overlapped time division multiplexing system with Fano algorithm
ES2593093B1 (es) Método y dispositivo para sincronización de trama en sistemas de comunicación
CN104113499A (zh) 一种基于概率排序的低复杂度存储空间约束检测方法
Wang et al. Length-constrained MAP decoding of variable-length encoded Markov sequences
CN116527457A (zh) 一种降噪稀疏度自适应弱正交匹配追踪方法及***
Sánchez et al. An efficient parallel algorithm for list Viterbi decoding

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17830372

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2019503461

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 20197005393

Country of ref document: KR

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 2017830372

Country of ref document: EP

Effective date: 20190222