CN109474412B - General filtering multi-carrier method based on selective mapping - Google Patents

Abstract

The invention provides a general filtering multi-carrier method based on selective mapping, which divides sub-carriers into B sub-bands, expands each sub-band into U different candidate sub-bands, multiplies each candidate sub-band by a group of randomly generated twiddle factors for the candidate sub-bands, filters the sub-carriers of each sub-band by an FIR filter, selects one sub-band with the minimum peak-to-average power ratio from the candidate sub-bands of a sending end for transmission, superposes the transmission signals in the B sub-bands selected from the sub-bands in the time domain to form a new sending signal, carries out fast Fourier transform on the receiving signal, removes interference signals, directly restores the sending signal from the receiving signal by a linear equalization method, carries out the inverse operation of the twiddle factors on the equalized signal, and can reconstruct the original sending signal. The PAPR performance of the invention is effectively improved, and the invention has certain engineering application value.

Description

General filtering multi-carrier method based on selective mapping

Technical Field

The invention relates to the technical field of wireless communication, in particular to a general multi-carrier filtering method.

Background

In the 4G era, the commercialization of LTE-Advanced basically meets the requirements of people on multi-element transmission of voice, video, network and the like, and greatly enriches the ever-increasing spiritual culture life of people. However, with the continuous improvement of image resolution and the increase of online duration of users, mobile data traffic is explosively increased in an exponential manner, massive internet of things devices are continuously emerging, and new application scenarios are continuously emerging, such as: holographic projection, unmanned driving, robotics, artificial intelligence, augmented reality, mobile social, and the like. The existing 4G network can not meet the requirement of future communication, and 5G is generated in the background. Unlike the single application scenario of 4G, IoT and M2M communication will become the main driving force for 5G communication, in other words, 5G will mainly solve the performance challenge problem faced in multi-scenario. The 5G technology needs to comprehensively consider multiple technical indexes such as peak rate, spectrum efficiency, network energy efficiency, delay and the like, effectively support multiple types of services, and meet the characteristics of high capacity, multiple accesses, high mobility, high transmission rate and low time delay.

The OFDM is widely applied to various wired and wireless occasions such as LTE, WiMAX, WIFI, HIPERLAN/2, DVB, ADSL and the like by the advantages of multipath fading resistance, lower complexity and seamless fusion of multi-antenna technology. But the higher peak power and out-of-band radiation, strict time synchronization limit the discontinuous spectrum of carrier aggregation, while the looser synchronization and localized spectrum characteristics will be one of the most important requirements of the physical layer of the future wireless network, which also makes IMT-2020 seek more diversified transmission technologies. The Generalized Frequency Division Multiplexing (GFDM) technique has attracted the attention of scholars, because GFDM has a higher bandwidth efficiency than OFDM, since only one Cyclic Prefix (CP) is used for a group of symbols, instead of using a CP for each symbol. The variable filter design and the sparsity of the signal make the GFDM robust to synchronization errors and more suitable for interactive scenes of spectrum fragmentation. However, the use of large-sized FFT and Successive Interference Cancellation (SIC) algorithms results in high complexity and decoding delay at the receiving end. Non-orthogonal waveform designs also face the same dilemma with GFDM and FBMC: complex pilot design and incompatibility with multiple antenna techniques.

Through the development of communication technologies of previous generations, the demand of the eMBB is basically met, but the wireless technology meeting the uRLLC and mMTC communication scenes is not developed enough. That is, it is very urgent to research a low latency wireless transmission technology for packet transmission that satisfies IoT and M2M communication. UFMC is just a new filtering transmission mechanism that satisfies this requirement. The UFMC reduces the length and out-of-band power of the filter by dividing the whole frequency band into a plurality of groups of sub-bands and filtering the carrier wave of each group of sub-bands. The modulation mode of QAM is adopted to be seamlessly connected with the MIMO technology, and the method is very suitable for short uplink burst communication or low-delay communication. However, the higher PAPR affects the energy efficiency of UFMC, and does not meet the requirement of 5G green communication.

The conventional solution to the PAPR problem in multi-carrier systems is to lower the operating point of the non-linear power amplifier, which usually results in a significant power efficiency loss. To solve this problem, many researchers have proposed different solutions. Such as: slicing (ACF), constellation extension (ACE), selective mapping (SLM), and Partial Transmit Sequence (PTS). Among them, SLM attracts attention of a large number of scholars due to its superior performance and characteristics of being free from interference and distortion. Although PAPR of OFDM is well known, PAPR research of UFMC has not been paid sufficient attention. According to the current state of the literature, only the W RONG proposes a low complexity PTS (LC-PTS) UFMC to reduce the PAPR of the system, unfortunately the system performance is degraded.

Disclosure of Invention

In order to overcome the defects of the prior art, the invention provides a general filtering multi-carrier method based on selection mapping.

The technical scheme adopted by the invention for solving the technical problem comprises the following steps:

(1) at a sending end, firstly, sub-bands are divided according to the number of carriers, and the sub-carriers with the total number of N are divided into B sub-bands;

(2) each sub-band is expanded into U different candidate sub-bands, a sending end generates BU different candidate sub-bands in total, and the candidate sub-bands all contain data information originally transmitted by the sending end;

(3) for the BU candidate subbands, each candidate subband is multiplied by a set of randomly generated twiddle factors B^(u)，B^(u)＝[b_u,0,b_u,1,…,b_u,N/B-1]^TWherein U is 1,2, …, U, b_u,N/B-1Is belonged to +/-1 and +/-j, and the rotated signal is expressed as S_i,j＝[b_i,0S₀,b_i,1S₁,...,b_i,N/B-1S_N/B-1]Wherein i is 1,2, …, B, j is 1,2, …, U;

(4) the signal in the frequency domain is transformed into the time domain by an inverse fast fourier transform, denoted as:

wherein S is_i,j(k) Is a candidate signal in the frequency domain, O_iIs the subcarrier index set in the ith subband;

(5) filtering the sub-carrier of each sub-band by an FIR filter, wherein the time domain impulse response of the filter of the ith sub-band is fi, i is 1,2, …, B, and obtaining a filtered signal x_i,j(l) For simplicity of analysis, the same filtering process is applied to the U candidate subbands, which are expressed as:

x_i,j(l)＝s_i,j(l)*f_i(l)l＝0,1,...,N+L-2 (2)

wherein, denotes a linear convolution operation, s_i,j(l) Representing a time domain signal;

(6) selecting one sub-band with the minimum peak-to-average power ratio from every U candidate sub-bands of a sending end for transmission, wherein the sub-band signals x are B sub-band signals_iAnd transmitting, wherein the selected subband index and the twiddle factor are transmitted to a receiving end in the form of sideband information to reproduce the received signal information:

x_i(l)＝min{x_i,j(l)}l＝0,1,...,N+L-2 (3)

(7) the transmission signals in B subbands selected from BU subbands are superimposed in the time domain to form a new transmission signal, which is expressed as:

(8) at the receiving end, the 2N point received signal is subjected to fast fourier transform, and the time domain signal is converted into a frequency domain signal Y, which is expressed as:

Y＝WPHFVS+WPn∈C^2N×1 (5)

wherein, W is a 2N-point FFT transformation matrix, P is an extended unit matrix, H is a Toeplitz matrix related to channel impulse response, F is a matrix formed by the Toeplitz matrix, V is a diagonal matrix formed by an inverse Fourier transformation matrix, and S is a signal matrix formed by frequency domain signals to be transmitted after selection;

(9) in the received signal Y, the signal at even number position is the signal to be demodulated, the signal at odd number position is the interference signal, the signal at odd number position is cut off, and only the signal at even number position is demodulated, the demodulated signal Y after the interference signal is removed_iExpressed as:

Y_i＝P_i ^TWPHFVS+P_i ^TWPn∈C^N×1 (6)

let phi equal to P_i ^TWPHFV, phi is a diagonal matrix, P_iIs a 2N × N matrix, and thus, the signal Y of the k-th subcarrier_i(k) I 1, …, B, depending only on the transmitted signal S_i,j(k) And a diagonal matrix Φ;

(10) directly recovering a transmitting signal from a receiving signal by a linear equalization method, wherein zero forcing equalization (ZF) and minimum mean square error equalization (MMSE) are used for recovering the signal;

where Φ (k) is an equalization coefficient of the k-th subcarrier, σ is a ratio of signal to noise power, and denotes a conjugate transpose operator;

(11) assuming that the receiving end has a transmission copy of the twiddle factor, the equalized signal is subjected to the inverse operation of the twiddle factor, and the original transmission signal can be reconstructed.

The twiddle factor in the step (3) can adopt a cyclic shift sequence, a Hadamard sequence, a Riemann sequence or a Helmert sequence.

The invention has the beneficial effect that compared with the traditional UFMC, the PAPR performance is effectively improved. Simulation results show that SLM-UFMC can be used in Prob (PAPR)>PAPR0)＝10^-4The PAPR of the UFMC is reduced by as much as 1.8 dB. The PAPR performance of the system is further improved as the number of candidate subbands increases. Under the condition that some hardware is not limited, the invention has certain engineering application value, improves the energy efficiency of the UFMC, better meets the requirement of 5G green communication, and is easy to be combined with the non-orthogonal multiple access technology to further improve the capacity of the system. Meanwhile, the method lays a foundation for the large-scale application of the UFMC system in a future 5G small-packet low-delay transmission scene and in the communication of the Internet of things and a machine to a machine.

Drawings

Fig. 1 is a block diagram of an exemplary general filtering multi-carrier system of the present invention.

Fig. 2 is a block diagram of a general filtering multi-carrier system based on selection mapping according to the present invention.

FIG. 3 is a PAPR performance comparison of different modulation modes of OFDM, UFMC, SLM-OFDM and SLM-UFMC systems of the present invention.

FIG. 4 shows the PAPR performance of SLM-UFMC under different candidate sub-band numbers according to the present invention.

Detailed Description

The invention is further illustrated with reference to the following figures and examples.

Due to the sub-band filtering of the UFMC, the SLM and the UFMC cannot be simply combined, but a part of modules need to be adjusted, and a general filtering multi-carrier system based on selective mapping is provided through reasonable design. Through the combination of UFMC and SLM, the PAPR performance of the system is effectively improved. Simulations have also shown that SLM-UFMC can be used in Prob (PAPR)>PAPR0)＝10^-4The PAPR of the UFMC is reduced by 1.8dB, so that the SLM-UFMC meets the requirement of 5G green communication, the capacity of the system is further improved by combining the non-orthogonal multiple access technology easily, and a foundation is laid for the large-scale application of the UFMC in a future 5G small-packet low-delay transmission scene and the communication of the Internet of things and machines to machines.

(1) At a sending end, firstly, sub-bands are divided according to the number of carriers, and the sub-carriers with the total number of N are divided into B sub-bands;

(2) each sub-band is expanded into U different candidate sub-bands, a sending end generates BU different candidate sub-bands in total, and the candidate sub-bands all contain data information originally transmitted by the sending end;

(3) for the BU candidate subbands, each candidate subband is multiplied by a set of randomly generated twiddle factors B^(u)，B^(u)＝[b_u,0,b_u,1,…,b_u,N/B-1]^TWherein U is 1,2, …, U, b_u,N/B-1Belongs to { +/-1, +/-j }, different twiddle factors bring different system performances, other types of twiddle factors can also be adopted, such as cyclic shift sequences, Hadamard sequences, Riemann sequences or Helmert sequences, and the signal after rotation is expressed as S_i,j＝[b_i,0S₀,b_i,1S₁,...,b_i,N/B-1S_N/B-1]Wherein i is 1,2, …, B, j is 1,2, …, U;

(4) the signal in the frequency domain is transformed into the time domain by an inverse fast fourier transform, denoted as:

wherein S is_i,j(k) Is a candidate signal in the frequency domain, O_iIs the subcarrier index set in the ith subband;

(5) filtering the sub-carrier of each sub-band by an FIR filter, wherein the time domain impulse response of the filter of the ith sub-band is fi, i is 1,2, …, B, and obtaining a filtered signal x_i,j(l) For simplicity of analysis, the same filtering process is applied to the U candidate subbands, which are expressed as:

x_i,j(l)＝s_i,j(l)*f_i(l)l＝0,1,...,N+L-2 (2)

wherein, denotes a linear convolution operation, s_i,j(l) Representing a time domain signal;

(6) selecting one sub-band with the minimum peak-to-average power ratio from every U candidate sub-bands of a sending end for transmission, wherein the sub-band signals x are B sub-band signals_iThe selected subband index and the twiddle factor are transmitted to a receiving end in the form of sideband information to reproduce the received signal information：

x_i(l)＝min{x_i,j(l)}l＝0,1,...,N+L-2 (3)

(7) The transmission signals in B subbands selected from BU subbands are superimposed in the time domain to form a new transmission signal, which is expressed as:

(8) at the receiving end, for the received signal, the conventional practice is to perform zero padding operation on the received signal, that is, to pad a certain number of zeros behind the received signal, so that the length of the received signal is 2N, the present invention performs fast fourier transform on the 2N-point received signal, and converts the time domain signal into a frequency domain signal Y, which is expressed as:

Y＝WPHFVS+WPn∈C^2N×1 (5)

wherein, W is a 2N-point FFT transformation matrix, P is an extended unit matrix, H is a Toeplitz matrix related to channel impulse response, F is a matrix formed by the Toeplitz matrix, V is a diagonal matrix formed by an inverse Fourier transformation matrix, and S is a signal matrix formed by frequency domain signals to be transmitted after selection;

(9) in the received signal Y, the signal at even number position is the signal to be demodulated, the signal at odd number position is the interference signal, the signal at odd number position is cut off, and only the signal at even number position is demodulated, the demodulated signal Y after the interference signal is removed_iExpressed as:

Y_i＝P_i ^TWPHFVS+P_i ^TWPn∈C^N×1 (6)

let phi equal to P_i ^TWPHFV, phi is a diagonal matrix, P_iIs a 2N × N matrix, and thus, the signal Y of the k-th subcarrier_i(k) I 1, …, B, depending only on the transmitted signal S_i,j(k) And a diagonal matrix Φ;

(10) directly recovering a transmitting signal from a receiving signal by a linear equalization method, wherein zero forcing equalization (ZF) and minimum mean square error equalization (MMSE) are used for recovering the signal;

where Φ (k) is an equalization coefficient of the k-th subcarrier, σ is a ratio of signal to noise power, and denotes a conjugate transpose operator;

(11) assuming that the receiving end has a transmission copy of the twiddle factor, the equalized signal is subjected to the inverse operation of the twiddle factor, and the original transmission signal can be reconstructed.

The examples of the invention are as follows:

for further explanation and understanding of the embodiments of the present invention, the basic principle of the conventional UFMC is first analyzed, and its functional block diagram is shown in fig. 1.

The N total number of sub-carriers are divided into B sub-bands, i.e. each sub-band has N/B sub-carriers, the ith sub-band has N sub-carriers_iI is 1,2, …, B. At the transmitting end, a certain amount of bit information is mapped onto the symbols S by gray code M-ary quadrature amplitude modulation (M-QAM). Then these symbols are allocated to each sub-band by means of carrier division, wherein the frequency domain symbol divided by the ith sub-band is S_i(i ═ 1,2, …, B). These symbols S are then_iConversion to time-domain signal s by N-point Inverse Fast Fourier Transform (IFFT)_i. The UFMC then performs a filtering process on each subband, summing the superposition of all subbands.

For a received signal, it is conventional to complement the received signal with a certain number of zeros, so that the received signal length reaches 2N. Then, fast Fourier transform is carried out on the 2N point received signal, the time domain signal is converted into the frequency domain signal Y, and the transmitted signal Y can be directly recovered from the received signal by an equalization method_i，i＝1,2,…,B。

The general filtering multi-carrier system based on selective mapping is implemented as follows, and the block diagram is shown in fig. 2.

In SLM-UFMC, N total number of subcarriers are first divided into B subbands, and then each subband is expanded into U different candidate subbands, so that the transmitting end generates BU different candidate subbands, which all contain originally transmitted data information, and after IFFT and filtering, selects B subbands with the smallest PAPR from the BU candidate subbands for superposition transmission. Fig. 2 presents a system block diagram of a generic filtering multi-carrier system based on selection mapping, each candidate subband being multiplied by a set of twiddle factors B for BU candidate subbands^(u)＝[b_u,0,b_u,1,…,b_u,N/B-1]^TWhere U is 1,2, …, U. Different twiddle factors may lead to different system performance, such as cyclic shift sequences, Hadamard sequences, Riemann sequences or Helmert sequences, etc. The rotated signal can be expressed as:

S_i,j＝[b_i,0S₀,b_i,1S₁,...,b_i,N/B-1S_N/B-1] (9)

wherein, i is 1,2, …, B, j is 1,2, …, U.

The signal in the frequency domain is transformed into the time domain by an inverse fast fourier transform, which is represented as:

wherein S is_i,jIs a candidate signal in the frequency domain, O_iIs the subcarrier index set in the ith subband. By filtering each group of sub-carriers f_iI-1, 2, …, B, a filtered signal x can be obtained_i,j(l) In that respect To simplify the analysis, the same filtering process is applied here for the U candidate subbands. Can be expressed as:

x_i,j(l)＝s_i,j(l)*f_i(l)l＝0,1,...,N+L-2 (11)

wherein, denotes a linear convolution operation, s_i,j(l) Denotes a time domain signal, where i is 1,2, …, B, j is 1,2, …, U. The sending end selects B sub-bands x with the minimum PAPR from the BU candidate sub-bands_iAnd transmitting, i is 1,2, … and B. The selected subband index and the twiddle factor information are transmitted to the receiving end in the form of sideband information for reproducing the received signal information.

x_i(l)＝min{x_i,j(l)}l＝0,1,...,N+L-2 (12)

Wherein i, j is as above, and min {. cndot } represents the PAPR minimum. The formation of a new transmission signal after the selected signals are superimposed in the time domain can be represented as:

if the above formula is rewritten in a matrix manner, it can be rewritten as:

wherein F ═ F₁，F₂,…,F_B],V＝Λ[V₁,V₂,…,V_B]Where Λ represents a diag matrix,

represents the frequency domain signal, S, to be transmitted after minimum PAPR selection_B,jIs and min { x_B,1,x_B,2,…,x_B,UThe corresponding transmission signal. V_i(i ═ 1,2, …, B) is an N × N_iA matrix of dimensions which extracts n from the inverse Fourier transform matrix according to the corresponding sub-band position over the entire available frequency range_iThe column vectors constitute the transform matrices of all sub-bands constituting exactly one complete inverse fourier transform matrix. V is a diagonal matrix, and the sub-matrices on the main diagonal are respectively V₁，V₂，…，V_B。F_iIs a Toeplitz matrix (L is the length of a linear filter, the filter lengths of different sub-bands can be different, and the carrier wave intervals can also be different) with (N + L-1) multiplied by N dimensions, each column of the Toeplitz matrix is obtained by circularly shifting the impulse response coefficient of the filter on the corresponding sub-band position, F_iCan be adjusted according to the propagation condition and time-frequency offset of the system.

Assuming that the system has perfect synchronization measures, the received signal y after passing through the channel is regarded as the convolution form of the signal x and the channel h, and is expressed as:

where H is the time domain impulse response of a channel of length r, representing a linear convolution with the signal x, and H is a Toeplitz matrix associated with the channel impulse response, the first column of which is [ H [ ]^T,0_1×(N+L-2)]^TIts first row element is [ h (0),0_1×(N+L-2)]^TAnd N is complex white Gaussian noise of (N + L + r-2) dimension. If the above formula is rewritten to a matrix form, it can be rewritten as:

y＝HFVS+n (16)

h is a Toeplitz matrix related to channel impulse response, F is a matrix formed by the Toeplitz matrix, V is a diagonal matrix formed by an inverse Fourier transform matrix, S is a signal matrix formed by frequency domain signals to be transmitted after selection, and N is complex Gaussian white noise with dimension of (N + L + r-2).

For the received signal y, it is conventional to perform a zero padding operation, i.e. to pad a certain number of zeros behind the received signal, so that the received signal has a length of 2N. Then, we perform fast fourier transform on the 2N point received signal to convert the time domain signal into the frequency domain signal Y. This process can be expressed as:

Y＝WPHFVS+WPn∈C^2N×1 (17)

wherein W is 2NA point FFT transformation matrix, P being an extended unit matrix, which can be written as

The signal at even number in the received signal Y is the signal we want to demodulate, the signal at odd number is the interference signal, and it is generally necessary to discard the signal at odd number and demodulate only the signal at even number. This operation may be performed using a 2N matrix P_iTo realize that the method is used for realizing,

demodulated signal Y after removal of interfering signals_iCan be expressed as:

Y_i＝P_i ^TWPHFVS+P_i ^TWPn∈C^N×1 (18)

let phi equal to P_i ^TWPHFV, Φ is a diagonal matrix. Thus, the signal Y_i(k) I 1, …, B, depending only on the transmitted signal S_i,j(k) And a diagonal matrix phi. The transmitted signal is directly reproduced from the received signal by a linear equalization method, and ZF and MMSE equalization can be used for restoring the signal without loss of generality.

Wherein the content of the first and second substances,

representing the equalized signal. Assuming that the receiver has a transmitted copy of the twiddle factor, the equalized signal is subjected to the inverse operation of the twiddle factor to reconstruct the original transmitted signal.

The peak-to-average power ratio of the SLM-UFMC system is defined as:

wherein E [ | x (l) & gtdoes not burn²]Representing the average power of the signal, max [ | x (l)²]Representing the maximum power of the signal. Because the PAPR of a single signal does not reflect the statistical properties of the signal, the PAPR of a system is usually characterized by a Complementary Cumulative Distribution Function (CCDF), i.e., the PAPR is greater than a given threshold PAPR₀Probability of (c):

CCDF_PAPR＝Prob(PAPR>PAPR₀) (22)

the PAPR performance of OFDM, UFMC and SLM-OFDM systems under different modulation modes is compared with that of SLM-UFMC. Computer simulation is used to verify the superiority of PAPR performance of SLM-UFMC system, and FIG. 3 shows the comparison of PAPR performance of OFDM, UFMC, SLM-OFDM and SLM-UFMC systems under different modulation modes. To fully evaluate the system performance, we used two experimental controls, CCDF was used to evaluate the PAPR performance of the system. The first group adopts 512 subcarriers and a 16QAM modulation mode, and the second group adopts 128 subcarriers and a QPSK modulation mode.

As can be seen from fig. 3, the conventional UFMC faces more severe PAPR than OFDM, which may be because the subband filter impulse response is not equal to 1, which increases the symbol length and thus affects the statistical properties of the system PAPR. This also presents a pressing problem for UFMC applications in IoT and M2M communication, that is how to solve UFMC high PAPR problem? Meanwhile, it can be seen from the figure that the PAPR performance of the UFMC is respectively reduced by 0.3dB and 0.5dB relative to the OFDM in two groups of experiments Prob (PAPR)>PAPR₀)＝10^-3. By adopting the SLM method, the PAPR of OFDM can be reduced, and compared with OFDM, the PAPR of SLM-OFDM is in Prob (PAPR)>PAPR₀)＝10^-4The performance is improved by about 3dB and 4dB, respectively. Moreover, compared with UFMC, the SLM-UFMC performance is respectively improved by 0.5dB and 1.8dB in Prob (PAPR)>PAPR₀)＝10^-4. Also, it is noted that the performance of SLM-OFMD is better than that of SLM-UFMC under the same conditions. At 10^-4Performance of bothWith a difference of 2.8dB and 2.3dB, respectively. This is consistent with the problems discussed above, i.e., UFMC faces more severe PAPR problems than OFDM under the same conditions. From different modulation schemes, QPSK modulation has better performance than 16QAM modulation, which may be due to the fact that the number of carriers is smaller, so that the number of subcarriers in lockstep is smaller, and the peak power superposition of the system is smaller.

The relation between the PAPR performance of the SLM-UFMC and the number U of different candidate subbands is shown in fig. 4, where the number U of candidate subbands takes on values of 4, 8, 16, and 64, respectively. It can be seen from the figure that as the number of candidate subbands increases, the number of candidate signal subbands that can be selected becomes greater, the range of selecting a transmission signal increases, and the performance of the system becomes better as the U value increases. In Prob (PAPR)>PAPR₀)＝10^-4The PAPR of the SLM-UFMC can be improved by nearly 1dB when U-64 compared to U-4. Of course, the performance is improved at the cost of sacrificing part of the complexity, and the larger the number U of candidate subbands is, the more the number of needed IFFTs is, and the higher the system complexity is. However, as hardware technology evolves, complexity may not be a bottleneck limiting the application of a new technology. Therefore, the system proposed herein is also an effective way to solve the high PAPR problem of UFMC without some hardware limitations. Meanwhile, the method lays a foundation for large-scale application of UFMC in IoT and M2M communication.

Aiming at the problem of high PAPR of UFMC, a universal filtering multi-carrier system SLM-UFMC based on selective mapping is provided. Through the ingenious combination of the SLM and the UFMC, the PAPR performance of the UFMC is effectively improved. Simulation results show that SLM-UFMC can be used in Prob (PAPR)>PAPR0)＝10^-4The PAPR of the UFMC is reduced by 1.8dB, the number U of the candidate sub-bands is increased, and the PAPR performance of the SLM-UFMC can be further improved. This also makes SLM-UFMC more compatible with the requirement of 5G green communication, and the invention can be easily combined with non-orthogonal multiple access technology to further increase the capacity of the system. Meanwhile, the method lays a foundation for large-scale application of UFMC in future 5G and machine-to-machine communication and Internet of things.

Claims (2)

1. A general filtering multi-carrier method based on selection mapping, characterized by comprising the steps of:

(1) at a sending end, firstly, sub-bands are divided according to the number of carriers, and the sub-carriers with the total number of N are divided into B sub-bands;

(2) each sub-band is expanded into U different candidate sub-bands, a sending end generates BU different candidate sub-bands in total, and the candidate sub-bands all contain data information originally transmitted by the sending end;

(3) for the BU candidate subbands, each candidate subband is multiplied by a set of randomly generated twiddle factors B^(u)，B^(u)＝[b_u,0,b_u,1,…,b_u,N/B-1]^TWherein U is 1,2, …, U, b_u,N/B-1Is belonged to +/-1 and +/-j, and the rotated signal is expressed as S_i,j＝[b_i,0S₀,b_i,1S₁,...,b_i,N/B-1S_N/B-1]Wherein i is 1,2, …, B, j is 1,2, …, U;

(4) the signal in the frequency domain is transformed into the time domain by an inverse fast fourier transform, denoted as:

wherein S is_i,j(k) Is a candidate signal in the frequency domain, O_iIs the subcarrier index set in the ith subband;

(5) filtering the sub-carrier of each sub-band by an FIR filter, wherein the time domain impulse response of the filter of the ith sub-band is fi, i is 1,2, …, B, and obtaining a filtered signal x_i,j(l) For simplicity of analysis, the same filtering process is applied to the U candidate subbands, which are expressed as:

x_i,j(l)＝s_i,j(l)*f_i(l)l＝0,1,...,N+L-2 (2)

wherein, denotes a linear convolution operation, s_i,j(l) Representing the time domain signal, L is the length of the linear filter;

(6) selecting a sub-band with the minimum peak-to-average power ratio from every U candidate sub-bands of a sending endLine transmission, with a total of B subband signals x_iAnd transmitting, wherein the selected subband index and the twiddle factor are transmitted to a receiving end in the form of sideband information to reproduce the received signal information:

x_i(l)＝min{x_i,j(l)}l＝0,1,...,N+L-2 (3)

(7) the transmission signals in B subbands selected from BU subbands are superimposed in the time domain to form a new transmission signal, which is expressed as:

(8) at the receiving end, the 2N point received signal is subjected to fast fourier transform, and the time domain signal is converted into a frequency domain signal Y, which is expressed as:

Y＝WPHFVS+WPn∈C^2N×1 (5)

w is a 2N-point FFT transformation matrix, P is an expanded unit matrix, H is a Toeplitz matrix related to channel impulse response, F is a matrix formed by the Toeplitz matrix, V is a diagonal matrix formed by an inverse Fourier transformation matrix, S is a signal matrix formed by frequency domain signals to be transmitted after selection, and N is complex Gaussian white noise;

(9) in the received signal Y, the signal at even number position is the signal to be demodulated, the signal at odd number position is the interference signal, the signal at odd number position is cut off, and only the signal at even number position is demodulated, the demodulated signal Y after the interference signal is removed_iExpressed as:

Y_i＝P_i ^TWPHFVS+P_i ^TWPn∈C^N×1 (6)

let phi equal to P_i ^TWPHFV, phi is a diagonal matrix, P_iIs a 2N × N matrix, and thus, the signal Y of the k-th subcarrier_i(k) I 1, …, B, depending only on the transmitted signal S_i,j(k) And a diagonal matrix Φ;

(10) directly recovering a transmission signal from a receiving signal by a linear equalization method, wherein zero forcing equalization (ZF) and minimum mean square error equalization (MMSE) are used for recovering the signal, and the formula is as follows:

where Φ (k) is an equalization coefficient of the k-th subcarrier, σ is a ratio of signal to noise power, and denotes a conjugate transpose operator;

(11) assuming that the receiving end has a transmission copy of the twiddle factor, the equalized signal is subjected to the inverse operation of the twiddle factor, and the original transmission signal can be reconstructed.

2. A general filtering multi-carrier method based on selection mapping according to claim 1, characterized in that:

the twiddle factor in the step (3) can adopt a cyclic shift sequence, a Hadamard sequence, a Riemann sequence or a Helmert sequence.

Images