Digital signal overall scaling method, device and equipment based on thermometer coding

Abstract

The invention discloses a digital signal overall scaling method, device and equipment based on thermometer coding, relates to the field of digital signal processing, and solves the problems that hardware is high in complexity and long in critical path in overall scaling calculation. Comprising the following steps: performing XNOR gate operation and AND gate logic operation on each element in an input vector to generate a zoom coefficient of a thermometer coding format; carrying out AND gate logic operation on the zoom coefficient, calculating a minimum value, and obtaining a minimum zoom coefficient under the coding representation of the thermometer; converting the minimum zoom coefficient into a binary format to obtain a binary zoom coefficient; performing numerical scaling and bit width reduction on each element in an input vector according to the binary scaling coefficient to obtain a scaled vector; the technical effects of low overall zooming complexity and short key path are achieved.

Description

Technical Field

[0001] The present invention relates to the field of digital signal processing, and more particularly to a method, device and apparatus for overall scaling of digital signals based on thermometer coding. Background Art

[0002] In modern communications and digital signal processing systems, representation and calculation methods based on binary fixed-point numbers are often used. Fixed-point numbers have a fixed decimal point position. To ensure computational accuracy, a sufficient fractional portion is required. To accommodate the large dynamic range in the system, a wide integer portion is often required. This results in a large bit width for fixed-point numbers, leading to greater storage and computational complexity.

[0003] In practical communications and digital signal processing systems, the required absolute computational accuracy is often dynamic, while the relative error is relatively fixed. Therefore, some systems employ floating-point numbers for calculations. Floating-point numbers represent values ​​as a fixed-point number multiplied by a power of 2, essentially scaling the fixed-point number. However, floating-point numbers also suffer from extremely high computational complexity, limiting their application.

[0004] Considering that in systems such as communications, the main source of dynamic range is power variation, and the frequency of power regulation is low, this means that the dynamic range of data remains roughly consistent over a short period of time. Therefore, a compromise can be made between fixed-point and floating-point representations, eliminating the need to scale individual numbers and instead scaling a range of values ​​as a whole. However, the maximum range of overall scaling should be determined by the number with the largest absolute value in the set of data. Therefore, some existing traditional implementations require a binary comparison tree to find the number with the largest absolute value. This processing method is highly complex and has a long critical path, making it difficult to implement in hardware and limiting the application of the overall scaling concept.

[0005] Therefore, there is an urgent need for a scaling method that can solve the above technical problems. Summary of the Invention

[0006] In order to address the deficiencies in the prior art, the purpose of the present invention is to provide a method, device and equipment for overall scaling of digital signals based on thermometer coding. The present invention combines the calculation of the scaling coefficient with the calculation of the minimum value of the thermometer code to ensure the correct calculation while greatly reducing the complexity and critical path of the circuit, thereby expanding the application scope of the overall scaling technology. It adopts a processing architecture of thermometer code generation, minimum value calculation and binary conversion, and reduces the complexity of the critical path by converting the traditional binary comparison tree into an AND gate logic operation in the thermometer domain. It is particularly suitable for ASIC (application-specific integrated circuit) implementation of large-scale vector processing.

[0007] The above technical objectives of the present invention are achieved through the following technical solutions:

[0008] In a first aspect, a method for overall scaling of a digital signal based on thermometer coding is provided, comprising the following steps:

[0009] Perform XNOR and AND logic operations on each element in the input vector to generate scaling coefficients in thermometer code format;

[0010] Performing an AND gate logic operation on the scaling factor to calculate a minimum value, thereby obtaining a minimum scaling factor under thermometer coding representation;

[0011] Converting the minimum scaling factor into a binary format to obtain a binary scaling factor;

[0012] Numerical scaling and bit width reduction are performed on each element in the input vector according to the binary scaling coefficient to obtain a scaled vector.

[0013] Furthermore, performing an XNOR gate operation and an AND gate logic operation on the input data to generate a scaling factor in a thermometer coding format includes:

[0014] Add one bit to the maximum number of scaled bits in each element of the input vector to generate a binary number with a length of one bit plus the maximum number of scaled bits;

[0015] Generate an intermediate signal with a maximum number of scaled bits plus one bit length by performing an exclusive OR operation on the sign bit of the binary number and each bit other than the sign bit;

[0016] The intermediate signal is then subjected to an AND gate logic operation to generate a scaling factor represented by a thermometer code.

[0017] Furthermore, the maximum number of scaling bits is equal to the number of consecutive bits equal to the sign bit in the most significant bits excluding the sign bit.

[0018] Furthermore, converting the minimum scaling factor into a binary format to obtain a binary scaling factor includes:

[0019] The decoding calculation of the thermometer code is realized by a multiplexer to convert the thermometer code into the scaling bit number of binary representation.

[0020] Furthermore, the numerical scaling and bit width reduction of each element in the input vector according to the binary scaling coefficient to obtain a scaled vector includes:

[0021] Selecting a corresponding input vector according to the minimum scaling factor of the binary system;

[0022] Adjusting the numerical range of the input vector according to the maximum scaling factor and the maximum scaling bit to obtain a scaled numerical range;

[0023] The reduced bit width is the difference between the total number of bits and the scaled number of bits;

[0024] After numerical scaling and bit width reduction, an integer part and a data part to be truncated are obtained, the integer part is retained, and the data part to be truncated is truncated to obtain a scaled vector.

[0025] Furthermore, the portion of the data to be truncated is rounded, where the rounding includes rounding to five, banker's rounding, or random rounding, to obtain the corrected scaled vector.

[0026] In a second aspect, a digital signal overall scaling device based on thermometer coding is provided, comprising:

[0027] A scaling factor calculation module is used to perform an XNOR operation and an AND logic operation on each element in the input vector to generate a scaling factor in a thermometer code format;

[0028] A minimum value calculation module, configured to perform an AND gate logic operation on the scaling factor to calculate a minimum value, thereby obtaining a minimum scaling factor under thermometer coding representation;

[0029] a decoder module, configured to convert the minimum scaling factor into a binary format to obtain a binary scaling factor;

[0030] The scaling module is configured to perform numerical scaling and bit width reduction on each element in the input vector according to the binary scaling coefficient to obtain a scaled vector.

[0031] Furthermore, the scaling factor calculation module includes:

[0032] an XNOR gate unit, configured to perform an XNOR operation on the sign bit of the binary number and each bit other than the sign bit, to generate an intermediate signal having a maximum number of scaled bits plus one bit length;

[0033] The AND gate unit is used for performing an AND gate logic operation on the intermediate signal to generate a scaling coefficient represented by a thermometer code.

[0034] Furthermore, the scaling module includes:

[0035] a vector selection unit, configured to select a corresponding input vector according to the minimum binary scaling factor;

[0036] The numerical scaling unit is used to adjust the numerical range of the input vector according to the maximum scaling coefficient and the maximum scaling bit to obtain a scaled numerical range.

[0037] According to a third aspect, an electronic device is provided, comprising the digital signal overall scaling device based on thermometer coding as described in the second aspect.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] 1. The present invention provides a thermometer-coded digital signal overall scaling method that combines scaling coefficient calculation with thermometer-coded minimum value calculation to ensure correct calculation while significantly reducing circuit complexity and critical path, thus expanding the application scope of overall scaling technology. Furthermore, the invention adopts a processing architecture that combines thermometer code generation, minimum value calculation, and binary conversion, and converts the traditional binary comparison tree into a thermometer-domain AND gate logic operation, thereby reducing the complexity of the critical path. The invention is particularly suitable for implementation in ASICs (application-specific integrated circuits) for large-scale vector processing.

[0040] 2. This invention breaks through the bottleneck of traditional dynamic range detection by constructing thermometer code through the use of sign bit XNOR operation. In the XNOR gate operation, the intermediate signal is generated by the XNOR operation of the sign bit and each bit, and the consistency of the highest bit continuous sign can be detected in parallel. Compared with the traditional circuit, the number of logic gates is saved by about 40%;

[0041] 3. By using thermometer coding to represent the scaling factor, the present invention enables the minimum value calculation module to complete the full vector extreme value extraction with only a single-stage AND gate array. Compared with the O(L)-level gate delay of the binary comparison tree, the processing delay is reduced to a fixed 1 logic gate delay.

[0042] 4. The present invention achieves a dynamic balance between storage resources and computing accuracy by establishing a bit width adaptive adjustment mechanism. The bit width reduction scheme dynamically controls the scaling factor through a thermometer encoder / decoder, and cooperates with a rounding module to achieve a small bit width compression error, which can reduce storage resource usage by 30%. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0044] Figure 1 This is a flowchart of Example 1 of the present invention;

[0045] Figure 2 This is a schematic diagram of Example 1 of the present invention;

[0046] Figure 3 This is a module block diagram in Example 2 of the present invention;

[0047] Figure 4 This is a schematic diagram of a scaling factor calculation module in Example 2 of the present invention;

[0048] Figure 5 This is a schematic diagram of a minimum value calculation module in Example 2 of the present invention;

[0049] Figure 6 Schematic diagram of a decoder module in Example 2 of the present invention;

[0050] Figure 7 This is a schematic diagram of the scaling module in Example 2 of the present invention. DETAILED DESCRIPTION

[0051] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with examples and drawings. The exemplary embodiments of the present invention and their descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0052] Example 1: A digital signal overall scaling method based on thermometer coding, such as Figure 1 As shown, the following steps are included:

[0053] Step 1: Perform XNOR and AND logic operations on each element in the input vector to generate scaling coefficients in thermometer code format;

[0054] Step 2: Perform an AND gate logic operation on the scaling factor to calculate the minimum value and obtain the minimum scaling factor under the thermometer coding representation;

[0055] Step 3: Convert the minimum scaling factor into a binary format to obtain a binary scaling factor;

[0056] Step 4: Perform numerical scaling and bit width reduction on each element in the input vector according to the binary scaling coefficient to obtain the scaled vector.

[0057] In the above steps, the principle of the overall scaling method of digital signals based on thermometer coding is as follows: Figure 2 shown.

[0058] In the two's complement representation system, for an N-bit binary number x[N-1…0], the representation weight of each bit is:

[0059]

[0060] Where i is the i-th digit of the binary number; W i is the weight of the i-th bit; N is the number of bits, and x is a binary number.

[0061] Taking into account:

[0062]

[0063] That is, when the highest bit is 1, it represents a negative number, and when the highest bit is 0, it represents a positive number. Therefore, the highest bit is also called the sign bit. The range of binary complement representation is [-2 N-1 ,2 N-1 -1].

[0064] In step 1, an XOR gate operation and an AND gate logic operation are performed on the input data to generate a scaling factor in a thermometer code format. The steps include: adding one bit to the maximum scaling bit number in each element of the input vector to generate a binary number with a length of one bit plus the maximum scaling bit number; performing an XOR operation on the sign bit of the binary number with each bit other than the sign bit to generate an intermediate signal with a length of one bit plus the maximum scaling bit number; and performing an AND gate logic operation on the intermediate signal to generate a scaling factor represented by the thermometer code. The maximum scaling bit number is equal to the number of consecutive bits in the most significant bit other than the sign bit that are equal to the sign bit.

[0065] In step 2, considering that in the overall scaling, it is necessary to find the minimum value of the scaling coefficient to ensure that each element in the input vector does not exceed the numerical representation range after scaling, and solving the minimum value in the traditional binary domain has the problems of high complexity and long critical path; under the thermometer coding representation system, the minimum value can be calculated through a simple AND gate. Therefore, the present invention creatively utilizes this feature to solve the calculation of the overall scaling coefficient.

[0066] In step three, the minimum scaling factor is converted into a binary format to obtain a binary scaling factor, including: decoding calculation of the thermometer code to convert the thermometer code into a scaling bit number represented in binary through a multiplexer.

[0067] In step 4, each element in the input vector is numerically scaled and the bit width is reduced according to the binary scaling coefficient to obtain a scaled vector, including: selecting the corresponding input vector according to the minimum binary scaling coefficient; adjusting the numerical range of the input vector according to the maximum scaling coefficient and the maximum scaling bit to obtain a scaled numerical range; the reduced bit width is the difference between the total number of bits and the number of scaled bits; after numerical scaling and bit width reduction, an integer part and a part of data to be truncated are obtained, the integer part is retained, and the part of data to be truncated is truncated to obtain the scaled vector.

[0068] In order to consider the relationship between scaling factor and numerical representation, consider a binary number x[N-1…0], and set its maximum scaling factor to α, where α = 2 β , the number of scaling bits β=0,1,…,N-2. Then we have

[0069]

[0070] Among them, α is the maximum scaling factor; β is the number of scaling bits; β max is the maximum scaling bit; α max is the maximum zoom factor.

[0071] Since the value range of x is an integer, we can simplify it to get:

[0072]

[0073] From the definition of numerical representation, we can get:

[0074]

[0075] and:

[0076]

[0077] That is, regardless of whether x is positive or negative, the number of consecutive bits in its highest order, excluding the sign bit, that are equal to the sign bit is equal to the maximum number of scaling bits. This property is similar to thermometer coding, but with some minor differences: 1. Thermometer coding requires that the bit representing the value must be 1, while the sign bit may be 0 or 1; 2. In thermometer coding, except for the 1 representing the value, all other bits are 0, which is not guaranteed here.

[0078] In scaling, in order to save subsequent storage and computational complexity, the bit width needs to be reduced, so the maximum scaling factor α needs to be set. max and its corresponding maximum scaling bit β max =log2(α max ), then the bit width after scaling should be N-β.

[0079] In some embodiments, step four further includes: rounding the portion of the data to be truncated, where the rounding includes rounding to five, banker's rounding, or random rounding, to obtain a corrected scaled vector to reduce the error caused by direct truncation.

[0080] Example 2: A digital signal overall scaling device based on thermometer coding, such as Figure 3 As shown, it includes a scaling coefficient calculation module, a minimum value calculation module, a decoder module, and a scaling module. The scaling coefficient calculation module is used to perform an XNOR gate operation and an AND gate logic operation on each element in the input vector to generate a scaling coefficient in a thermometer-coded format; the minimum value calculation module is used to perform an AND gate logic operation on the scaling coefficient to calculate the minimum value and obtain the minimum scaling coefficient under the thermometer-coded representation; the decoder module is used to convert the minimum scaling coefficient into a binary format to obtain a binary scaling coefficient; and the scaling module is used to perform numerical scaling and bit width reduction on each element in the input vector according to the binary scaling coefficient to obtain a scaled vector.

[0081] For example, assuming the length of the input vector is L=4, the maximum scaling bit β max = 3. The input vector will be the highest β max The +1 bit is sent to the scaling coefficient calculation module to calculate the scaling coefficient under the thermometer code representation. Then, the minimum value calculation module calculates the minimum scaling coefficient under the thermometer code representation. The decoder module converts the thermometer code into binary, and finally the scaling module uses this coefficient to complete the scaling.

[0082] In some embodiments, the principle of the overall scaling factor calculation module based on thermometer code representation is as follows Figure 4 As shown. By performing an exclusive-or (XNOR) operation on the sign bit and each other bit, the bit that is the same as the sign bit can be represented as 1, and the bit that is different can be represented as 0. According to the conclusions obtained in formulas (5) and (6), the number of consecutive bits equal to the sign bit in the highest bit other than the sign bit is equal to the maximum number of scaled bits, that is, the consecutive 1s must be interrupted by at least one 0. In order to represent it as a thermometer code, it is necessary to perform an AND operation on each bit and change all the 1s after the at least one 0 to 0. The area of ​​this module is only β max XNOR gates and β max -1 AND gate, the critical path is only 1 XNOR gate and β max -1 AND gate with extremely low complexity and short critical path.

[0083] In some embodiments, the principle of the overall scaling factor minimum value calculation module based on thermometer code representation is as follows: Figure 5 As shown. The area of ​​this module is only β max The critical path is only one L-input AND gate.

[0084] In some embodiments, the principle of the decoder module is as follows Figure 6 As shown in FIG. , the decoding calculation of the thermometer code can be completed by a small number of multiplexers (MUX).

[0085] In some embodiments, reference Figure 7 ,The scaling module can use MUX to select the input corresponding to a specific scaling factor. ,Taking into account the change in bit width, the data to be truncated can be rounded. ,The rounding processing includes rounding off, banker’s rounding or random rounding, and a ,corrected scaled vector is obtained to reduce the error caused by direct truncation.

[0086] In some embodiments, the scaling factor calculation module includes: an XNOR gate unit, which is used to perform an XNOR operation on the sign bit in the binary number and each bit except the sign bit, to generate an intermediate signal with a maximum scaling bit number plus one bit length; and an AND gate unit, which is used to perform an AND gate logic operation on the intermediate signal to generate a scaling factor represented by the thermometer code.

[0087] In some embodiments, the scaling module includes: a vector selection unit for selecting a corresponding input vector according to a minimum binary scaling coefficient; and a numerical scaling unit for adjusting the numerical range of the input vector according to a maximum scaling coefficient and a maximum scaling bit to obtain a scaled numerical range.

[0088] Embodiment 3: An electronic device is provided, which includes the digital signal overall scaling device based on thermometer coding as described in Embodiment 2, and the electronic device is used to implement the digital signal overall scaling method based on thermometer coding as described in Embodiment 1.

[0089] Working Principle: The present invention provides an overall scaling method with low complexity and short critical path. By performing an XNOR gate operation and an AND gate logic operation on each element in the input vector, a scaling coefficient in thermometer code format is generated. An AND gate logic operation is performed on the scaling coefficient to calculate the minimum value, thereby obtaining the minimum scaling coefficient under the thermometer code representation. The minimum scaling coefficient is then converted into binary format to obtain a binary scaling coefficient. Finally, each element in the input vector is numerically scaled and bit width reduced according to the binary scaling coefficient to obtain a scaled vector. By combining the scaling coefficient calculation with the minimum value calculation of the thermometer code, the circuit complexity and critical path are greatly reduced while ensuring the correctness of the calculation, thus expanding the application scope of the overall scaling technology. In addition, the processing architecture of thermometer code generation, minimum value calculation and binary conversion is adopted, and the critical path complexity is reduced by converting the traditional binary comparison tree into an AND gate logic operation in the thermometer domain.

[0090] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0091] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0092] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0093] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0094] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A digital signal overall scaling method based on thermometer coding, characterized in that: The following steps are involved: Perform XNOR and AND logic operations on each element in the input vector to generate scaling coefficients in thermometer code format; Performing an AND gate logic operation on the scaling factor to calculate a minimum value, thereby obtaining a minimum scaling factor under thermometer coding representation; Converting the minimum scaling factor into a binary format to obtain a binary scaling factor; Numerical scaling and bit width reduction are performed on each element in the input vector according to the binary scaling coefficient to obtain a scaled vector.

2. The method for overall scaling of digital signals based on thermometer coding according to claim 1, characterized in that: The step of performing an XNOR gate operation and an AND gate logic operation on the input data to generate a scaling factor in a thermometer coding format includes: Add one bit to the maximum number of scaled bits in each element of the input vector to generate a binary number with a length of one bit plus the maximum number of scaled bits; Generate an intermediate signal with a maximum number of scaled bits plus one bit length by performing an exclusive OR operation on the sign bit of the binary number and each bit other than the sign bit; The intermediate signal is then subjected to an AND gate logic operation to generate a scaling factor represented by a thermometer code.

3. The method for overall scaling of digital signals based on thermometer coding according to claim 1, characterized in that: The maximum number of scaling bits is equal to the number of consecutive bits equal to the sign bit among the most significant bits excluding the sign bit.

4. The method for overall scaling of digital signals based on thermometer coding according to claim 1, characterized in that: Converting the minimum scaling factor into a binary format to obtain a binary scaling factor includes: The decoding calculation of the thermometer code is realized by a multiplexer to convert the thermometer code into the scaling bit number of binary representation.

5. The method for overall scaling of digital signals based on thermometer coding according to claim 1, characterized in that: The step of numerically scaling and bit-width reducing each element in the input vector according to the binary scaling coefficient to obtain a scaled vector includes: Selecting a corresponding input vector according to the minimum scaling factor of the binary system; Adjusting the numerical range of the input vector according to the maximum scaling factor and the maximum scaling bit to obtain a scaled numerical range; The reduced bit width is the difference between the total number of bits and the scaled number of bits; After numerical scaling and bit width reduction, an integer part and a data part to be truncated are obtained, the integer part is retained, and the data part to be truncated is truncated to obtain a scaled vector.

6. The method for overall scaling of digital signals based on thermometer coding according to claim 5, characterized in that: Also includes: The portion of the data to be truncated is rounded, where the rounding includes rounding to five, banker's rounding, or random rounding, to obtain a corrected scaled vector.

7. A digital signal overall scaling device based on thermometer coding, characterized in that: include: A scaling factor calculation module is used to perform an XNOR operation and an AND logic operation on each element in the input vector to generate a scaling factor in a thermometer code format; A minimum value calculation module, configured to perform an AND gate logic operation on the scaling factor to calculate a minimum value, thereby obtaining a minimum scaling factor under thermometer coding representation; a decoder module, configured to convert the minimum scaling factor into a binary format to obtain a binary scaling factor; A scaling module is configured to perform numerical scaling and bit width reduction on each element in the input vector according to the binary scaling coefficient to obtain a scaled vector.

8. The digital signal overall scaling device based on thermometer coding according to claim 6, characterized in that: The scaling factor calculation module includes: an XNOR gate unit, configured to perform an XNOR operation on the sign bit of the binary number and each bit other than the sign bit, respectively, to generate an intermediate signal having a maximum number of scaled bits plus one bit length; The AND gate unit is used for performing an AND gate logic operation on the intermediate signal to generate a scaling coefficient represented by a thermometer code.

9. The digital signal overall scaling device based on thermometer coding according to claim 6, characterized in that: The scaling module includes: a vector selection unit, configured to select a corresponding input vector according to the minimum binary scaling factor; The numerical scaling unit is used to adjust the numerical range of the input vector according to the maximum scaling coefficient and the maximum scaling bit to obtain a scaled numerical range.

10. An electronic device, characterized in that: The electronic device comprises the digital signal integral scaling device based on thermometer coding according to claim 7.

Images