Up-sampling and shaping filtering combined signal modulation method

Abstract

The embodiment of the invention provides a signal modulation method combining up-sampling and shaping filtering in a modulation process. The method comprises the following steps of: sending a code element value lookup table (1), then accumulating an impulse response sequence (8) after I-path table lookup and an impulse response sequence (9) after Q-path table lookup at a T0-Tn moment according to an up-sampling mode (7) to obtain an I / Q value (10) at the T0-Tn moment, and then forming a sending waveform signal through subsequent digital up-conversion and DA signal output. In the table lookup process, a single code element value (2) is decomposed into an I sequence value and a Q sequence value (3), the modulation and shaping filtering joint lookup table (4) is looked up respectively, and an I-path modulated impulse response sequence (6) and a Q-path modulated impulse response sequence (5) are obtained. The method has the advantages that the modulation and shaping filtering combined lookup table is utilized, the modulation process, up-sampling and shaping filtering combined calculation is realized, multiplication in the modulation and filtering process is eliminated, and the operand is saved.

Description

Technical Field

[0001] This application relates to the field of digital communications, and in particular to signal modulation processes. Background Technology

[0002] In communication systems, the traditional signal modulation and transmission process is as follows: First, baseband data is mapped to symbol data; second, the symbol data is upsampled to match the transmission rate of the analog-to-digital converter (DA); then, in order to avoid inter-symbol interference, the upsampled symbols are convolved with a shaping filter to obtain the shaped and filtered data; finally, the shaped and filtered data is digitally upconverted and sent to the DA to form the final waveform signal.

[0003] Specific implementation as follows Figure 1 As shown, firstly, according to the modulation method, the baseband binary points d of length N are... n Mapped to form a series of code elements a n The sequence p(t) is represented as:

[0004]

[0005] Secondly, to adapt to the DA output sampling rate F s Perform L-fold upsampling, and after sampling, each symbol c n =[a n 0…0] 1×L The upsampled sequence q(t) represents:

[0006]

[0007] Among them, the multiplier LF s / R s , which represents the DA sampling rate / baud rate value.

[0008] Signal q(t) and g of length W T After the convolution filtering operation, we get:

[0009]

[0010] In this process, the filtering process is a convolution operation between N*L length waveform points and W*L length shaping filter. The total amount of computation is (N·LW·L 1)·W·L 2·sum(1:(W·L-1)) multiplications and (N·LW·L)·N·L 2·sum(1:(W·L 2)) additions, which is very computationally intensive.

[0011] To reduce computational load, the invention patent published in 2001 with patent number CN1365199A, entitled "Digital Shaping Filtering Method for Saving Computational Load", utilizes the characteristic of many zeros in the shaping filter to propose a digital shaping FIR filter for dedicated input signals. It only calculates the non-zero part of the value, reducing the computational load of convolution operations, but its calculation process still requires a certain amount of multiplication operations.

[0012] The invention patent published in 2013 with patent number CN103457888A, entitled "Communication Signal Transmission Method of Combined Upsampling and Digital Shaping", proposes a method to generate calculations in an FPGA using a lookup table. The coefficients of the shaping filter are generated based on a preset group delay and a preset upsampling factor N. However, the size of the lookup table is orders of magnitude larger than the group delay τ. When the group delay increases, a large table space is required. In addition, the size of the shaping filter is limited to N, which is not conducive to the design of shaping filters of a specific length.

[0013] Therefore, it is necessary to further reduce the amount of convolution operations and improve the system's computing performance to meet the requirement of rapid generation of transmitted waveforms. Summary of the Invention

[0014] The purpose of this invention is to provide a signal modulation method that combines upsampling and shaping filtering, such as... Figure 2 As shown, this addresses the problem of high computational complexity and high computational resource consumption in existing technologies for convolution multiplication operations.

[0015] During the modulation and transmission process, the binary baseband signal b to be transmitted k Its bit rate is R b The output signal s(t) is expressed as: Using M(M 2) K Modulation converts an M-bit binary stream into a sequence of symbol values ​​d. n (d n ∈[0, ...,M-1]), code rate is R s =R b / K, the output signal is

[0016] During transmission, in order to match the DA rate of the digital-to-analog converter, the signal is upsampled by a factor of L according to the mode and then converted to c. n =[d n 0…0] 1×L Or c n =[d n d n …0] 1×L ,a n The number is set to a value between 1 and L according to the waveform requirements, where L = F. s / R s This represents the DA output rate / code rate. Because c n Except for some values, all other values ​​are 0, so this property can be used to simplify the computational complexity.

[0017] During this process, for T0-T n When transmitting I / Q values, the modulated impulse response sequence is first obtained by looking up a table, and then the impulse response point value of each symbol at each time step is obtained according to the upsampling mode. Then, the values ​​from T0 to T... n The sum of the points at each time step yields T0-T. n The I / Q value.

[0018] In the single-symbol modulation process, a joint lookup table method is used for modulation and shaping filters. The I / Q values ​​are obtained from the symbol values, and the impulse response sequence after I / Q modulation is obtained after looking up the table. The values ​​in the lookup table are the shaping filter coefficients multiplied by the symbol modulation values. Due to the symmetry of the I / Q path values, the I / Q paths can share a single table.

[0019] This modulation process requires first looking up the coefficients of the raised cosine shaping filter in a lookup table, and then proceeding with the modulation process. The specific implementation is as follows:

[0020] (I) Constructing a joint lookup table for modulation and shaping filters

[0021] During the symbol modulation process, a lookup table method is used to obtain the corresponding I / Q sequence value (3) from the symbol value. Based on the I / Q sequence value (3), the I-path modulated impulse response sequence (6) and the Q-path modulated impulse response sequence (5) are obtained by looking up the table in the modulation and shaping filter joint lookup table (4).

[0022] The value in the lookup table is the shaping filter coefficient * symbol modulation value. Due to the symmetry of the I / Q path values, the I / Q paths can share a single table (4). The purpose of this table is to establish a mapping relationship between the sequence value index and the shaping filter coefficient * symbol modulation value.

[0023] Step 1: Based on the effect of reducing intersymbol interference, generate the raised cosine filter coefficients G′=[g0 … g WL-1 ], where the length of the raised cosine shaping filter is WL.

[0024] Step 2: According to M(M 2) K ) modulated symbol index sequence [0, ..., M-1] mapped to IQ point (A i A j Due to IQ symmetry, the real part ranges from [A0 … A M / 2-1 ], A0 and A M / 2-1 This represents the actual signal amplitude.

[0025] Step 3: Based on the mapping relationship of symbol I / Q index I, the real part filter coefficient lookup table G is:

[0026] correspond

[0027] The table size is WLM / 2. The lookup table for the imaginary part filter coefficients is symmetrical to the lookup table for the real part filter coefficients and can share the same table. A lookup can be performed using G[I,:] to return a single row of filter coefficients.

[0028] (II) Calculation process of symbol impulse response superposition based on upsampling mode

[0029] All symbols are obtained by looking up the table according to the sampling pattern (7) to obtain symbol T0-T. n The impulse response point value at time T0 is the sum of the impulse response point values ​​of all symbols at time T0. The same processing is applied to T1 until T... n To obtain T0-T, click here. n The sum of the impulse response point values ​​generated by the corresponding symbol at time T0-T is used as the T0-T value. n The I / Q value output at any given time.

[0030] This calculation process eliminates the multiplication operations in conventional modulation shaping filtering, and converts T0-T... n The calculation process of each point at each time moment is converted into the summation of the modulation and shaping filter values ​​of each symbol at the current time moment, that is, the summation of the impulse response sequence after the I-way lookup table at time T0-Tn (8) and the summation of the impulse response sequence after the Q-way lookup table at time T0-Tn (9). The total amount of calculation is the sum of the symbol transmission lengths, and the overall amount of calculation is the number of additions of the product of the number of symbols and the length of the shaping filter.

[0031] Assuming there are N symbols, the signal after being upsampled by L times becomes c. n =[d n 0…0] 1×L The number of symbol points is NL, and the shaping filter length is WL. The first point of the first symbol is d. n Generate g of length WL T (t) Impulse response, no impulse response is generated for the next L-1 points; the first point d of the second symbol n+1 Generate g of length WL T (t) Impulse response: no impulse response is generated for the next L-1 points, but the impulse responses from times L+1 to 2L of the first symbol are superimposed; the first point d of the third symbol. n+2 Generate g of length WL T(t) Impulse response: no impulse response is generated for the next L-1 points, but the impulse response of the first symbol from time 2L+1 to 3L will be superimposed, as well as the impulse response of the second symbol from time L+1 to 2L, and so on up to N symbols.

[0032] The calculation and superposition process is as follows: Figure 3 As shown.

[0033] In this process, the impulse response coefficients G[index,:] of the shaping filter are obtained by searching the modulation and shaping filter joint lookup table according to the upsampled symbol I / Q value sequence. Then, the value after shaping filtering at time t is the sum of the impulse response coefficients G[symbol index,t] of each generated symbol shaping filter at time t. Assuming the accumulation depth is N, the value after shaping filtering of the I-path at time t is:

[0034]

[0035] Then, the point value after shaping and filtering at time t can be obtained, and the I / Q value at all times can be obtained in this way.

[0036] (III) Definition of Upsampling Mode

[0037] In the upsampling process, the length of the transmitted symbol data after sampling is adapted to the DA by the number of sampling points L per symbol. The distribution of the symbol index sequence d(n) can be set according to the upsampling mode:

[0038] (1) The length L contains a d(n), whose position can be defined as any position in L.

[0039] Such as [d(n)…0…0], [0…d(n)…0] or [0…0…d(n)]

[0040] (2) The length L contains M d(n) elements, whose positions can be defined as positions 0-L / 2 in L.

[0041] For example, in the form of [d(n)…d(n)…0], [0…d(n)…d(n)…0], or [0…d(n)…d(n)], the number of d(n) M and the length are less than L / 2. The purpose of repeating multiple d(n) is to increase the signal content in the transmitted data.

[0042] The above-mentioned objective is achieved through the following technical solution: Detailed Implementation

[0043] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features described herein can be combined with each other unless otherwise specified.

[0044] (1) The following modulation parameters are used:

[0045]

[0046] (2) Construction of the joint lookup table for modulation and shaping filters

[0047]

[0048] For example, the actual table values ​​are as follows:

[0049] rrcFilter 0.005248 0.006031 0.006797 0.007539 0.008248 0.008916 0.009535 … tabRcos(1) 0.003711 0.004264 0.004806 0.005331 0.005832 0.006305 0.006743 … tabRcos(2) -0.00371 -0.00426 -0.00481 -0.00533 -0.00583 -0.0063 -0.00674 …

[0050] (3) Definition of I / Q modulation value

[0051]

[0052]

[0053] Note: MATLAB arrays start from 1.

[0054] (3) Definition of upper sampling mode

[0055] Assuming the [d 0 0…0] pattern is used, L = 40

[0056] (4) Sending process

[0057] TxData = randi([0MPsk-1], 1, 50); % Send data

[0058] % lookup method modulation

[0059] [sigOut_I,sigOut_Q]=mod_byRcosTab(TxData,sps,tabRcos,filterSpanSym);

[0060] For example, the transmitted code element is: [0 3 1 2 0 2 1 2 2 2 1…]

[0061] (5) Table lookup and accumulation process

[0062]

[0063] sigOut_I(k) = value_I;

[0064] sigOut_Q(k) = value_Q;

[0065] k = k + 1;

[0066] end

[0067] end

[0068] disp(k);

[0069] end

[0070] (5) Call the single data point accumulation process

[0071]

[0072] (6) Final Data Generation

[0073] The length of the I-value and Q-value sequence is 2160, which is N*sps + (filter symbols - 1)*sps (50*40 + 4*40)

[0074]

[0075]

[0076] (7) Comparison of Schemes

[0077] With N = 50 symbols, 4 psk, SPS L = 50, and filter length W = 5 symbols, the computational complexity of each implementation method is as follows:

[0078]

[0079] Notice:

[0080] Specific embodiments of the present invention have been described above. Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application.

[0081] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

[0082] Beneficial effects

[0083] This invention proposes a signal modulation method that combines upsampling and shaping filtering. It eliminates the multiplication operations in the conventional modulation and shaping filtering process and transforms the shaping filtering multiplication calculation process at each moment into the summation of the modulation and shaping filtering values ​​of each symbol at the current moment. The computational cost is the number of additions of the symbol transmission length, and the overall computational cost is the number of additions of the product of the number of symbols and the shaping filter length, which greatly saves computational cost. Attached Figure Description

[0084] Figure 1 A diagram illustrating a conventional signal generation process provided by this invention;

[0085] Figure 2 This is a diagram illustrating the signal generation process after computational optimization proposed in this invention.

[0086] Figure 3 This is a diagram illustrating the impulse response signal superposition process proposed in this invention.

[0087] Figure 4 This is a detailed diagram illustrating the modulation and transmission process of the transmitting symbols proposed in this invention.

Claims

1. A signal modulation method of upsampling and shaping filter joint, characterized in that : (1) Modulation and shaping filter joint look-up table The look-up table method is used in the symbol modulation process, and the corresponding I / Q sequence value (3) is obtained from the symbol value. According to the I / Q sequence value (3), the I-channel modulated impulse response sequence (6) and the Q-channel modulated impulse response sequence (5) are obtained after table lookup in the modulation and shaping filter joint look-up table (4). (2) Symbol impulse response superposition calculation process based on up-sampling mode All symbols are looked up according to the up-sampling pattern (7) to get symbols T0-T n The impulse response point value at time T0 is the sum of the impulse response point values of all symbols at time T0, T1 is the same processing until T n , and the superimposed value of the impulse response point values generated by the symbols corresponding to times T0-T n is taken as the I / Q value output at times T0-T n .

2. The modulator and shaping filter joint look-up table of claim 1, wherein : The values in the look-up table are shaping filter coefficients * symbol modulation values. Due to the symmetry of I / Q channel values, I / Q channels can share one table (4). The purpose of this table is to establish the mapping relationship between sequence value index and shaping filter coefficient * symbol modulation value.

3. The process of claim 1, wherein the process is based on an up-sampling pattern of the symbol impulse response superposition. : The multiplication operation in the conventional modulation shaping filter process is eliminated, T0-T n The calculation process of the point at each moment is converted into the sum of the current moment symbol modulation and shaping filter value, that is, the I path look-up table after impulse response sequence accumulation (8) at T0-Tn moment and the Q path look-up table after impulse response sequence accumulation (9) at T0-Tn moment, the total calculation amount is the addition of the symbol transmission length, and the overall calculation amount is the addition times of the multiplication value of the symbol number and the shaping filter length.

4. The up-sampling pattern (7) of claim 1, characterized by : The up-sampling process is to adapt the length of the sampled symbol transmission data to the digital-to-analog converter DA as the number of sampling points per symbol L. The distribution of the symbol index sequence d(n) can be set according to the up-sampling mode: (1) L length contains one d(n), its position can be defined as any position in L Such as [d(n)...0...0], [0...d(n)...0] or [0...0...d(n)] (2) L length contains M d(n), its position can be defined as 0-L / 2 position in L Such as [d(n)...d(n)...0], [0...d(n)...d(n)...0], or [0...d(n)...d(n)] form, the number of d(n) M length is less than L / 2, so the purpose of repeating multiple d(n) is to enhance the signal content in the transmission data.

Images