Low-delay reconfigurable Viterbi decoding circuit

By adopting parallel processing and improved backtracking algorithms in the Viterbi decoding circuit, a decoding circuit with low delay and low power consumption is realized, which solves the problems of high delay and high power consumption in the existing technology, and improves the decoding circuit flexibility and adaptability.

CN119051672APending Publication Date: 2024-11-29CHONGQING UNIV OF POSTS & TELECOMM
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411219850.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

The existing Viterbi decoding circuits have problems with high delay, low flexibility and high power consumption, making it difficult to adapt to high-speed communication and power-sensitive application scenarios.

Method used

A low-latency reconfigurable Viterbi decoding circuit is designed, and parallel processing and improved backtracking algorithm is adopted to use control unit, reconfigurable branch metric generation unit, reconfigurable path metric storage unit, and comparison and selection unit. Surviving path memory unit and backtracking unit realize circuit reconstruction and decoding, reducing decoding delay and power consumption.

Benefits of technology

It effectively reduces the decoding delay of the Viterbi decoding circuit, improves flexibility and availability, and can adapt to a variety of communication standards and power-sensitive application scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119051672A_ABST
    Figure CN119051672A_ABST
Patent Text Reader

Abstract

The invention relates to a low-delay reconfigurable Viterbi decoding circuit, and belongs to the technical field of wireless digital communication. The circuit comprises a control unit, a reconfigurable branch metric generation unit, a reconfigurable path metric storage unit, an add-compare-select unit, a surviving path storage unit and a backtracking unit. The control unit receives code rate, constraint length and truncation length information and generates reconstruction information according to the received information so as to control reconstruction of a subsequent decoding circuit. And the reconfigurable branch metric generation unit selects different branch metric generation units according to the code rate information, calculates an Euclidean distance by adopting a soft decision method, and generates a branch metric value. According to the invention, through the reconfigurable design, the support for various code rates and constraint lengths is realized, the decoding delay is reduced through the improved backtracking algorithm, and the flexibility and availability of the Viterbi decoding circuit are improved, so that the Viterbi decoding circuit can meet the application requirements of high-speed communication and sensitive power consumption.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of wireless digital communications and relates to a low-delay reconfigurable Viterbi decoding circuit. Background Art

[0002] With the rapid development of wireless communication technology, new communication standards and services emerge in an endless stream, and terminal equipment needs to be able to adapt to multiple communication standards. As an important channel coding technology, convolutional code is widely used in various digital communication systems. As a decoding algorithm for convolutional codes, the Viterbi algorithm is widely used in practical communication systems due to its excellent performance and low complexity.

[0003] However, the existing Viterbi decoding circuit has the following problems:

[0004] Existing Viterbi decoding circuits usually use a serial processing method, which results in a long decoding delay and is difficult to meet the needs of high-speed communication.

[0005] Existing Viterbi decoding circuits are usually designed for specific code rates and constraint lengths, and are difficult to adapt to different communication standards.

[0006] The existing Viterbi decoding circuit has a complex structure and high power consumption, and is difficult to apply to power-sensitive application scenarios.

[0007] In order to overcome the above problems, the present invention proposes a low-delay reconfigurable Viterbi decoding circuit to improve the flexibility, availability and decoding efficiency of the Viterbi decoding circuit. Summary of the invention

[0008] In view of this, an object of the present invention is to provide a low-latency reconfigurable Viterbi decoding circuit to solve the problems of high latency, low flexibility and availability of the current Viterbi decoding circuit.

[0009] In order to achieve the above object, the present invention provides the following technical solutions:

[0010] A low-latency reconfigurable Viterbi decoding circuit, comprising:

[0011] A control unit, for receiving information about a code rate, a constraint length, and a truncation length, and generating reconstruction information according to the received information to control reconstruction of a subsequent decoding circuit;

[0012] Reconfigurable branch metric generation unit, selects different branch metric generation units according to the bit rate information, and uses the soft decision method to calculate the Euclidean distance to generate the branch metric value;

[0013] Reconfigurable path metric storage unit, dynamically adjusts the number of storage units according to constraint length information, and is used to store path metric values ​​and perform iterative calculations;

[0014] An add, compare and select unit performs add, compare and select operations according to the path metric value and the branch metric value to generate a surviving path and a decision bit;

[0015] A survivor path storage unit, used for storing the decision bits generated by the addition, comparison and selection unit;

[0016] The backtracking unit uses an improved backtracking algorithm to backtrack according to the decision bit information of the surviving path storage unit and outputs the decoding result.

[0017] Further, the improved backtracking algorithm includes:

[0018] While writing the decision bits, a register is used to store the key decision bit values;

[0019] In the backtracking stage, the backtracking read operation is omitted, and the backtracking is performed directly according to the key decision bit value stored in the register, and the backtracking start state is provided.

[0020] Further, the reconfigurable branch metric generation unit includes:

[0021] Three branch metric generation units corresponding to code rates 1 / 2, 1 / 3 and 1 / 4 respectively;

[0022] The selector selects a corresponding branch metric generation unit according to the reconstruction information sent by the control unit.

[0023] Furthermore, the reconfigurable path metric storage unit adopts an in-situ storage update method to reduce memory usage and circuit power consumption.

[0024] Furthermore, the add, compare and select unit adopts a butterfly structure to reduce circuit delay.

[0025] Furthermore, the survivor path storage unit adopts a register exchange method to reduce decoding delay.

[0026] The beneficial effects of the present invention are:

[0027] (1) By adopting parallel processing and an improved backtracking algorithm, the present invention effectively reduces the decoding delay of the Viterbi decoding circuit, improves the decoding efficiency, and can meet the needs of high-speed communication.

[0028] (2) The present invention adopts a reconfigurable design and can be reconfigured according to different code rates and constraint lengths, thereby adapting to a variety of communication standards and improving the flexibility of the Viterbi decoding circuit.

[0029] (3) The present invention reduces circuit power consumption by optimizing the circuit structure and storage method, so that it can be applied to power-sensitive application scenarios.

[0030] (4) The present invention improves the availability of the Viterbi decoding circuit by reducing decoding delay and power consumption, so that it can be widely used in various wireless communication systems.

[0031] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0033] Figure 1 A schematic diagram of the overall architecture of a low-latency reconfigurable Viterbi decoding circuit provided by the present invention;

[0034] Figure 2 This is the ACS circuit structure diagram;

[0035] Figure 3 This is the structure diagram of the ACS butterfly unit;

[0036] Figure 4 This is the schematic diagram of the register exchange method;

[0037] Figure 5 Schematic diagram of the traditional backtracking algorithm. DETAILED DESCRIPTION

[0038] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0039] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0040] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0041] The present invention proposes a low-delay reconfigurable Viterbi decoding circuit, the overall framework of which is as follows: Figure 1 As shown, it includes a control unit (Control Unit), a reconfigurable branch metric generation unit (Configurable Branch Metric Generation Unit), a reconfigurable path metric storage unit (Configurable Path Metric Unit) plus a comparison and selection unit (Add Compare and Select Unit), a survivor path storage unit (Survivor Path Memory Unit), and a traceback unit (Traceback Unit).

[0042] The control circuit, when working, sends reconstruction information to the subsequent decoding circuit after updating the code rate, constraint length, and truncation length to control the reconstruction of the decoding circuit. The code rate R determines the control of the control circuit on the selector in the subsequent decoding circuit. When the input code rate R is 1 / 2, the selector selects 00, when the code rate is 1 / 3, the selector selects 01, and when the input code rate R is 1 / 4, the selector selects 10. The constraint length K will cause the structure of the addition and comparison selection unit and the reconfigurable path metric storage unit to change. After the control circuit receives the required constraint length K, the numerical value of the data is added to the reconstruction information, and the subsequent decoding circuit receives the data to reconstruct the circuit in real time. The traceback depth L is usually 5-10 times the numerical value of the constraint length K. In the reconfigurable Viterbi decoding circuit, the value is set to 5L. When the control unit receives the truncation length L, the numerical value of the data is added to the reconstruction information. The numerical value of the traceback length affects the specific parameters of the subsequent traceback unit storage unit, so the bit width and depth of the traceback unit memory are designed to be variable, and the specific size changes according to the reconstruction information value sent by the control unit.

[0043] The reconfigurable branch metric generation unit is affected by the code rate R. Since the reconstruction algorithm is only used for the three code rates of 1 / 2, 1 / 3 and 1 / 4, three branch metric generation units are designed to adapt to the above three code rates respectively. The specific selection of the two branch metric generation units is controlled by the selector. The selector selects the circuit after receiving the reconstruction information sent by the control information. In the branch metric generation unit, soft decision is used instead of traditional hard decision, that is, the Euclidean distance is calculated instead of the Hamming distance. The specific calculation method is as follows. Taking R = 1 / 2 as an example, assume that the soft decision adopts 8-level uniform quantization, and assume that the received code element is x = {x 0 ,x 1 ,…,x n ,…}, the codeword received if the transmission is correct should be y={y 0 ,y 1 ,…,y n ,…}, then the Euclidean distance calculation formula of this string of code elements is In hardware circuits, this operation consumes a lot of resources, so the formula is simplified to When R=1 / 2, the number of code elements received without error is (000,000), (000,111), (111,000), (111,111), so the module can only output four values: BM 00 , B.M. 01 , B.M. 10 , B.M. 11 According to the Euclidean distance calculation formula, we can get:

[0044] BM 00 =|x 1 -0|+|y1 -0|=x 1 +y 1

[0045] BM 01 =|x 1 -0|+|y 1 -7|=x 1 +~y 1

[0046] BM 10 =|x 1 -7|+|y 1 -0|=~x 1 +y 1

[0047] BM 11 =|x 1 -7|+|y 1 -7|=~x 1 +~y 1

[0048] Here, ~ represents the bitwise inversion of y, which further simplifies the calculation process, reduces the amount of calculation, and makes the circuit easier to implement.

[0049] The circuit structure of the addition, comparison and selection unit is related to the constraint length K, which is 2 K-1 The maximum constraint length supported by the invention is 9, so the circuit structure uses a total of 256 add-compare-select units, and the number of add-compare-select units is reconstructed in real time by receiving the reconstruction information of the control circuit. The specific number is 2 K-1 And ensure that the remaining unused addition and comparison selection units will not affect the circuit function.

[0050] ACS circuit structure diagram Figure 3 The adder performs an addition operation, which is to add the path metric at time t-1 to the newly generated branch metric value to obtain the path metric value at time t. The comparator performs a comparison operation, which is PM i+1 Get PM i +BM i,3 With PM j +BM j,3 The smaller value of the two, in addition, the information of the path that is not selected is deleted. The selector performs a selection operation, which is to select the path with the smaller total path value as the surviving path. Repeat the above steps using the updated path metric value and the newly generated branch path metric value. In addition, after each comparison by the comparator of the ACS unit, a 1-bit decision bit will be output for the subsequent surviving path storage unit. This decision bit can be used to determine which branch of the t-1 state the t state is transferred from.

[0051] ACS butterfly unit such as Figure 2 As shown, S 1 and S 2 are two states at time t-1, PM i , PM j They are S 1 , S 2 The path metric of the state, S 3 and S 4 are two states at time t, PM i+1 , PM j+1 They are S 3 and S 4 The path metric of the state, BM i,3 , B.M. i,4 ,BM j,3 ,BM j,4 is S at time t-1 1 and S 2 The newly generated branch metric of the two states, BM i,3 , B.M. i,4 , corresponding to S 1 Transfer to S 3 and S 4 Two paths, BM j,3 ,BM j,4 Corresponding to S 2 Transfer to S 3 and S 4 Two paths, where the path metrics satisfy PM i+1 Get PM i +BM i,3 With PM j +BM j,3 The smaller of the two, PM j+1 Get PM i +BM i,4 With PM j +BM j,4 The smaller of the two.

[0052] The reconfigurable path metric storage unit circuit structure is related to the constraint length K, and the specific number is 2 K-1 The circuit receives the reconstruction information sent by the reconstruction control circuit, transforms the structure in real time, and realizes the iterative calculation of the path metric. At time t-1, after the ACS unit adds, compares and selects, a new path metric value is calculated, and the value is stored in the path metric storage unit. At time t, the path metric value is sent back to the adding, comparing and selecting unit for calculation, and then the step is repeated. The present invention adopts a parallel method, using 2 K-1 path metric storage units to match 2 K-1In the path storage unit, each time a new path metric value is stored in the ACS output, the path metric value at the previous moment is no longer needed. Therefore, the module adopts the in-situ storage update method to store the newly received 2 K-1 The path metric value covers the previous moment 2 K-1 The method can reduce the usage of memory and thus reduce the power consumption of the circuit.

[0053] The survivor path storage unit is used to store the decision bits generated by the addition, comparison and selection unit for use by the subsequent backtracking unit. The backtracking unit adopts an improved backtracking algorithm to obtain the correct survivor path through the decision bit information stored in the survivor path storage unit and output the decoding result.

[0054] The decision bit output by the ACS unit can be used to determine which path is used to transfer to the current state, so the surviving path at time t-1 plus the transfer path is the surviving path at time t. The register exchange method is to transfer the surviving path information stored in the register at time t-1 to the new register and add the transfer path to obtain the new storage information. The principle of the register exchange method is as follows Figure 4 As shown in the figure, each row of registers stores the surviving path of a state, and the selector before each path is determined by the decision bit of the current state. With each new decision bit input, the process will shift the surviving path of the previous moment to the right and transfer it to the correct row, while outputting the decoding result from the right end. The advantage of the register exchange method is that the decoding delay is short, but when the constraint is large, the internal connection relationship of the decoder will become too complicated.

[0055] Traditional backtracking algorithms such as Figure 5 As shown, it includes a write operation, a backtracking read operation, and a decoding read operation. The write operation is to write the decision bits generated by the addition, comparison and selection unit to prepare for the subsequent read operation. The backtracking read operation is that when the number of decision bits written meets the backtracking depth, the backtracking unit starts backtracking. When performing the backtracking operation, a pointer with a bit width of k-1 will eventually be output. This pointer can be used as the initial backtracking starting point for the second backtracking operation. The pointer obtained after the second backtracking read operation is completed is used to complete the subsequent decoding read operation. The decoding read operation described is to backtrack and generate a decoding result, which is consistent with the backtracking read process. While the decoding read operation outputs the decoding result, the memory simultaneously releases storage space for writing new data. The algorithm requires 3 read operation pointers and 6 storage units with a depth of L / 2. The write direction is from left to right, and the read direction is from right to left, as shown in Figure 5As shown, each row is a state, and each state lasts for L / 2. In one state, the write operation, the decoding read operation, and the two backtracking read operations are performed in parallel, and the delay of the backtracking algorithm can be obtained as 3L. This method is easy to control and manage, but the disadvantage is that the delay is relatively long. In order to achieve low delay without increasing the overall circuit power consumption, the present invention proposes an improved backtracking algorithm based on the traditional backtracking algorithm. The delay of the traditional backtracking algorithm is mainly caused by the write operation, one decoding read operation and two backtracking read operations. The purpose of the backtracking read operation is to find the initial state required for the decoding read, and provide a pointer with a bit width of K-1 for the decoding read for subsequent use. The pointer value is ultimately determined by the decision bit. Therefore, when the amount of written data meets the backtracking read depth L / 2 each time, the register is used to directly store the key decision bit value. In the subsequent backtracking stage, the backtracking read operation is omitted, and the backtracking starting state is directly provided for the decoding read operation. The application of this method can reduce the delay of the circuit in the backtracking stage and achieve low delay of the Viterbi decoding circuit.

[0056] Embodiment 1

[0057] This embodiment provides a Viterbi decoding circuit with K=7, R=1 / 3, and L=42.

[0058] 1. Control unit:

[0059] The receiving code rate information R=1 / 3, the constraint length information K=7, and the truncation length information L=42.

[0060] Reconstruction information is generated, including a rate selection signal, constraint length information, and truncation length information.

[0061] 2. Reconfigurable branch metric generation unit:

[0062] According to the code rate selection signal, a corresponding branch metric generation unit is selected.

[0063] The soft decision method is used to calculate the Euclidean distance and generate the branch metric value.

[0064] 3. Reconfigurable path metric storage unit:

[0065] According to the constraint length information, the number of path metric storage units is set to 128.

[0066] The in-situ storage update method is adopted to overwrite the old path metric value with the new path metric value.

[0067] 4. Add comparison unit:

[0068] According to the constraint length information, the number of addition, comparison and selection units is set to 128.

[0069] Perform addition, comparison, and selection operations to generate survivor paths and decision bits.

[0070] 5. Survivor path storage unit:

[0071] 128 decision bits are stored for use by the traceback unit.

[0072] 6. Retrospective unit:

[0073] An improved backtracking algorithm is used to perform backtracking based on the decision bit information of the surviving path storage unit.

[0074] Output the decoding result.

[0075] 7. Decoding process:

[0076] The control unit receives the code rate, constraint length and truncation length information.

[0077] The reconfigurable branch metric generation unit generates branch metric values.

[0078] The add, compare and select unit performs add, compare and select operations to generate survivor paths and decision bits.

[0079] The reconfigurable path metric storage unit stores the path metric value.

[0080] The survivor path storage unit stores the decision bits.

[0081] The backtracking unit performs backtracking and outputs the decoding result.

[0082] Embodiment 2

[0083] This embodiment provides a Viterbi decoding circuit with K=9, R=1 / 2, and L=45.

[0084] 1. The structures of the control unit, the reconfigurable branch metric generation unit, the reconfigurable path metric storage unit, the addition, comparison and selection unit, the survivor path storage unit and the backtracking unit are the same as those in the first embodiment.

[0085] 2. Decoding process:

[0086] The control unit receives the code rate, constraint length and truncation length information.

[0087] The reconfigurable branch metric generation unit generates branch metric values.

[0088] The add, compare and select unit performs add, compare and select operations to generate survivor paths and decision bits.

[0089] The reconfigurable path metric storage unit stores the path metric value.

[0090] The survivor path storage unit stores the decision bits.

[0091] The backtracking unit performs backtracking and outputs the decoding result.

[0092] The present invention can also be implemented in the following ways:

[0093] Use different soft decision methods: For example, more level uniform quantization methods or other soft decision methods can be used to improve decoding performance.

[0094] Use different backtracking algorithms: For example, you can use a pointer-based backtracking algorithm or other backtracking algorithms to improve backtracking efficiency.

[0095] Use different memory structures: For example, SRAM, DRAM, or other types of memory can be used to meet different application requirements.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A low-latency reconfigurable Viterbi decoding circuit, characterized in that: include: A control unit, for receiving information about a code rate, a constraint length, and a truncation length, and generating reconstruction information according to the received information to control reconstruction of a subsequent decoding circuit; Reconfigurable branch metric generation unit, selects different branch metric generation units according to the bit rate information, and uses the soft decision method to calculate the Euclidean distance to generate the branch metric value; Reconfigurable path metric storage unit, dynamically adjusts the number of storage units according to constraint length information, and is used to store path metric values ​​and perform iterative calculations; An add, compare and select unit performs add, compare and select operations according to the path metric value and the branch metric value to generate a surviving path and a decision bit; A survivor path storage unit, used for storing the decision bits generated by the addition, comparison and selection unit; The backtracking unit uses an improved backtracking algorithm to backtrack according to the decision bit information of the surviving path storage unit and outputs the decoding result.

2. The low-latency reconfigurable Viterbi decoding circuit according to claim 1 is characterized in that: The improved backtracking algorithm includes: While writing the decision bits, a register is used to store the key decision bit values; In the backtracking stage, the backtracking read operation is omitted, and the backtracking is performed directly according to the key decision bit value stored in the register, and the backtracking start state is provided.

3. The low-latency reconfigurable Viterbi decoding circuit according to claim 1 is characterized in that: The reconfigurable branch metric generation unit comprises: Three branch metric generation units corresponding to code rates 1 / 2, 1 / 3 and 1 / 4 respectively; The selector selects a corresponding branch metric generation unit according to the reconstruction information sent by the control unit.

4. The low-latency reconfigurable Viterbi decoding circuit according to claim 1 is characterized in that: The reconfigurable path metric storage unit adopts an in-situ storage update method to reduce memory usage and circuit power consumption.

5. The low-latency reconfigurable Viterbi decoding circuit according to claim 1 is characterized in that: The add-compare-select unit adopts a butterfly structure to reduce circuit delay.

6. The low-latency reconfigurable Viterbi decoding circuit according to claim 1 is characterized in that: The survivor path storage unit adopts a register exchange method to reduce decoding delay.