CN106789814A - A kind of low complex degree SLM algorithms of reduction FBMC OQAM systems PAPR - Google Patents

Abstract

The invention discloses a kind of low complex degree SLM algorithms of reduction FBMC OQAM systems PAPR, i.e. low complex degree dispersion selected mapping method (LD SLM) algorithm, comprise the following steps：S1, U length of generation are the phase place vector of N；S2, current data block is multiplied with rotating vector；S3, signal is sampled, filtered and is modulated；S4, computation interval are middle two continuous symbol cycle Ts_c；Rotating vector u when S5, preservation PAPR values are minimum_min；S6, according to u_minOptimal rotating vector is selected, input signal matrix is updatedSignal Xs of the return to step S2 to the next signal cycle_m+1Repeat the above steps, until m=M 1.The present invention, with the characteristic for reducing system PAPR, it is contemplated that the natural lap of FBMC OQAM signals, shortens search time while optimized algorithm complexity, specific dispersion selected mapping method (DSLM) algorithm in complexity advantageously.

Description

Low-complexity SLM algorithm for reducing PAPR of FBMC-OQAM system

Technical Field

The invention relates to an optimization method of a communication technology, in particular to a low-complexity SLM algorithm for reducing PAPR of an FBMC-OQAM system.

Background

Filter bank multi-carrier (FBMC) is a multi-carrier transmission scheme with high spectral efficiency, reasonable implementation complexity and no need for synchronization. Offset Quadrature Amplitude Modulation (OQAM) can eliminate inter-carrier interference (ICI) that would be caused by the overlap of sub-bands in an FBMC system. FBMC, in combination with OQAM, has many advantages not found in OFDM technology, such as excellent frequency location and low Power Spectral Density (PSD) sidelobes, which are more suitable for 5G Radio Access Technologies (RATs) than OFDM. FBMC-OQAM systems are increasingly becoming the leaders of the upcoming 5G Radio Access Technology (RAT) for radio waves. Like Orthogonal Frequency Division Multiplexing (OFDM), FBMC-OQAM has a high PAPR, which reduces the efficiency of a High Power Amplifier (HPA), resulting in signal distortion, spectrum spreading, and system performance degradation. Therefore, the PAPR reduction technique of FBMC-OQAM system is a major subject of current research.

PAPR reduction techniques can be divided into the following: clipping techniques, coding techniques, scrambling techniques, adaptive predistortion techniques, and Discrete Fourier Transform (DFT) spread spectrum techniques. The amplitude limiting technology is to reduce the PAPR by adopting amplitude limiting or nonlinear saturation near a peak value, and comprises a block scaling technology, a clipping and filtering technology, a Fourier mapping technology and a decision-assisted reconstruction technology; the coding technique is to select a code word capable of minimizing or reducing the PAPR, and can use a Colay complementary sequence, an M sequence and a Hadamard code; the scrambling technique is to scramble the input data and transmit a data block with the smallest PAPR; adaptive predistortion techniques may compensate for the non-linear effects of HPA in OFDM systems.

FBMC-OQAM cannot use a method of reducing PAPR well in OFDM due to its overlapping structure. Currently, some researchers discuss schemes for reducing PAPR of FBMC-OQAM system using PAM symbols, but these schemes are limited to PAM symbols and BER performance is poor. There are also researchers who have implemented a clipping scheme using iterative compensation to reduce PAPR of the FBMC-OQAM system, but the system needs to design a complicated receiver to satisfy the compensation of clipping noise. The multi-module joint optimization (MBJO) technique and the sliding window voice reservation (SWTR) technique are currently used in PAPR reduction of FBMC-OQAM systems, but they have a high complexity. The researchers have proposed an overlapping selective mapping (OSLM) method, which proposes that the assumption of independence between consecutive symbols does not hold, taking into account the natural overlap of FBMC-OQAM, but has a huge memory and computational complexity. Later, another scholars proposed a dispersion selective mapping (DSLM) method, which is similar to the conventional SLM, and considers the overlapping property of FBMC-OQAM, and solves the property of time dispersion of FBMC-OQAM signals, but the PAPR value in the [0, 4T ] interval is calculated, and the complexity is relatively high. Based on the statement, the invention improves on the basis of the SLM algorithm, reasonably utilizes the overlapping property and the power distribution rule of FBMC signals, and provides the low-complexity SLM algorithm, namely the LD-SLM algorithm, for reducing the PAPR of the FBMC-OQAM system.

Disclosure of Invention

As known from average power simulation of FBMC-OQAM signals, energy of each signal period is mainly and intensively distributed in [ T,3T ], and the invention provides a low-complexity SLM algorithm, namely an LD-SLM algorithm, for reducing PAPR of the FBMC-OQAM system. The PAPR value of the symbol concentrated in the interval [ T,3T ] is calculated according to the characteristic that the power distribution of the FBMC-OQAM signal is concentrated in the middle [ T,3T ] (T is a symbol period), and the minimum PAPR value in the interval [ T,3T ] is found out to select the optimal rotation symbol.

A low complexity SLM algorithm for reducing PAPR of an FBMC-OQAM system comprises the following steps:

s1, first, initializing the rotation vectors, and generating U phase rotation vectors with length N:

wherein:n∈[0,N-1]，u∈[0,U-1]；

s2, multiplying the current data block by the rotation vector, each input data block X_mAre multiplied by U different rotation vectors:

wherein: represents a matrix point-to-point multiplication;

s3, sampling, filtering and modulating the signal, wherein due to the overlapping nature of the FBMC-OQAM signal, the signal before the current data block needs to be considered, so as to obtain the following signals:

wherein: m belongs to [0,2 pi ], T belongs to [0, (m +1/2) T +4T ];

s4, calculating the signal xu (according to the following formula)^t) The calculation interval of the PAPR of (1) is Tc;

s5, according to the result of PAPR calculation, selecting the optimal rotation scheme to obtain the rotation vector number when the PAPR value is minimum, i.e. the rotation vector number is the minimumAnd will u_minStored as side information in U_SI＝[U_SIu_min]Performing the following steps;

s6, finally updating the input according to u_minSelecting the best rotation vector, multiplying the best rotation vector with the current data block, and updating the input signal matrixThen, the process returns to step S2, where the signal X of the next signal cycle is processed_m+1The above steps are repeated until M ═ M-1.

The LD-SLM algorithm only calculates the PAPR value of the symbol concentrated in the interval [ T,3T ], finds out the minimum PAPR value in the interval [ T,3T ] to select the best rotation symbol, which is not the [0, T ] interval of the SLM algorithm or the [0, 4T ] interval of the DSLM.

The low-complexity SLM algorithm for reducing the PAPR of the FBMC-OQAM system provided by the invention has the characteristic of reducing the PAPR of the system while optimizing the complexity of the algorithm, not only considers the natural overlapping part of FBMC-OQAM signals, but also shortens the search time, has more advantages than the DSLM algorithm in time complexity, and compared with the DSLM algorithm, because the steps before calculating the PAPR value are the same, namely the PAPR is inputThe incoming FBMC signal is sampled and filtered, requiring 2N (L) to pass through the filter_h+1)(MT₀+L_h-1) real multiplications, followed by multiplication of each data block with the generated rotation vector, require UN (L)_h+ T/2) complex multiplications; UN ((2m +1) T/2+4T) complex multiplications are needed to modulate the current data block and the previous data block; 2UNT is required for calculating PAPR value_cMultiplication of real numbers and NUT_cAddition of real numbers to obtain maximum value requires NUT_cThe real number division and UNT are respectively needed for 1 time for the comparison, the average value and the logarithm obtaining_cA sub-logarithmic operation, and 1 real multiplication, where 1 real multiplication, real addition, real division, logarithmic operation, and one look-up comparison operation are all written as 1 real operation; therefore, the difference in calculating the PAPR, i.e., the formula, is mainly comparedThe implementation algorithms of (a) are different, so the computation complexity of the DSLM algorithm is: c_DSLM16MNUT +4MNUT + M, i.e. C_DSLM20MNUT + M; the LD-SLM algorithm provided by the invention has the same processes of sampling, filtering and multiplying by the twiddle factor as the DSLM, and the PAPR calculation interval is [ T,3T ] when the optimal twiddle factor is obtained]Therefore, the computation amount when rotating and calculating the PAPR is different, and the computation complexity for the low complexity SLM algorithm is: c_LD-SLM8MNUT +2MNUT + M, i.e. C_LD-SLM10MNUT + M; in the formula: n is the number of subcarriers; m is the number of data blocks; t is the code element width/symbol period; k is a sampling factor/overlap factor; l is_hIs the filter impulse response length; t is₀For a sampling period, T_cThe interval taken for calculating the PAPR value, M being the current data block, and M ∈ [0, M-1](ii) a By analyzing and comparing the results, the LD-SLM algorithm provided by the invention reduces 10MNUT real number operations, C, in complexity compared with the DSLM algorithm_LD-SLMThe calculation complexity is reduced by about 50 percent by being approximately equal to 0.5CDSLM, and the low complexity SLM algorithm for reducing the PAPR of the FBMC-OQAM system, which is provided by the invention, comprehensively considers the overlapping property of FBMC-OQAM symbols, namely most energy of signals is concentrated in two continuous FBMC-OQAM symbol periods of the 2 nd and the 3 rd,therefore, the method can be well suitable for an FBMC-OQAM system, when the PAPR of the system is reduced, the performance of the LD-SLM algorithm provided by the invention is very close to that of the DSLM algorithm, and the calculation complexity is reduced by 10MNUT real number operations compared with that of the DSLM algorithm, namely, the calculation complexity is reduced by about 50%.

Drawings

FIG. 1 is an FBMC-OQAM signal model;

FIG. 2 shows the power distribution of 4 consecutive data blocks of FBMC-OQAM;

FIG. 3 is a block diagram of an SLM technology;

FIG. 4 shows PAPR under different U values of the OFDM-SLM system;

fig. 5 shows PAPR distribution under different schemes when U is 4;

fig. 6 shows a comparison of FBMC signals with OFDM signals.

Detailed Description

In order to further explain the technical scheme of the invention, the invention is explained in detail with reference to the figures and the specific embodiments.

Examples

The invention provides a low-complexity SLM algorithm for reducing PAPR of an FBMC-OQAM system, which comprises the following steps:

s1, first, initializing the rotation vectors, and generating U phase rotation vectors with length N:

wherein:n∈[0,N-1]，u∈[0,U-1]；

s2, multiplying the current data block by the rotation vector, each input data block X_mAre multiplied by U different rotation vectors:

wherein: represents a matrix point-to-point multiplication;

s3, sampling, filtering and modulating the signal, wherein due to the overlapping nature of the FBMC-OQAM signal, the signal before the current data block needs to be considered, so as to obtain the following signals:

wherein: m belongs to [0,2 pi ], T belongs to [0, (m +1/2) T +4T ];

s4, calculating the signal x according to the following formula^u(T) PAPR, the calculation interval being T_c；

Analyzing the simulation results of fig. 2, it can be known that the power of each signal is mainly distributed in the middle 2T interval of each signal period, so T here_c∈[mT+T,mT+3T]Spanning 2T, and T in DSLM technique_c∈[mT,mT+4T]Spanning 4T;

s5, according to the result of PAPR calculation, selecting the optimal rotation scheme to obtain the rotation vector number when the PAPR value is minimum, i.e. the rotation vector number is the minimumAnd will u_minStored as side information in U_SI＝[U_SIu_min]Performing the following steps;

s6, finally updating the input according to u_minSelecting the best rotation vector, multiplying the best rotation vector with the current data block, and updating the input signal matrixThen, the process returns to step S2, where the signal X of the next signal cycle is processed_m+1The above steps are repeated until M ═ M-1.

FIG. 1 shows an FBMC-OQAM signal model; in the FBMC-OQAM system, we use signal transmission based on OQAM modulation, and for a transmitting end containing M complex input signals, the transmitting end of N subcarriers can be written as follows:

whereinAndrespectively the real part and the imaginary part of the mth data block transmitted on the nth subcarrier, the real part and the imaginary part of the signal have a time-domain difference of T/2(T is a symbol period, also called a symbol width), and the signal is oversampled with a sampling period of T₀When the oversampling factor is K and is more than or equal to 4, the PAPR value of the sampled signal is very close to the PAPR value of the continuous signal, and K is 4 in the following simulation; after the signal passes through the filter, passes through a prototype filter h (t) and is modulated by N subcarriers, it can be obtained:

wherein N is 0,1, …, N-1; then, the user can use the device to perform the operation,and superposing the signals on the N sub-carrier signals to obtain a signal on an m-th data block:

wherein: l is the length of the prototype filter h (t), and X can be seen_m(T) has a length of (L + T/2); finally, the M data blocks are added together to obtain an FBMC-OQAM final signal x (t):

from (2) and (4) can be obtained:

wherein N is 0,1, …, N-1; m is 0,1, …, M-1, h (t) is the impulse response of the prototype filter, the PHYDYAS prototype filter is used in this document, the length of the filter is L kN-1, k is the overlapping factor, N is the number of subcarriers, wherein:

the impulse response of the filter is as follows:

wherein,is a standardized constant.

Fig. 2 shows power distribution of 4 adjacent data blocks, and it can be seen that each FBMC-OQAM signal lasts 4.5T and overlaps with the following 3 signals, and it can be seen that the power of the FBMC-OQAM signal is mainly distributed between 2 nd to 3 rd symbol periods of its signal duration period, i.e., concentrated between [ mT, (m +2) T ]. For the prototype filter, where the energy is mainly in the main lobe and its length affects the duration of the impulse response of the FBMC-OQAM signal, we define the power distribution of the FBMC-OQAM signal as follows:

P_avg[X(t)]＝|X(t)|²

fig. 3 is a block diagram illustrating a conventional SLM technique for reducing PAPR of an OFDM system. Inputting data X and sequence with U different phasesMultiplying to obtain a modified data block X^uWhereinN-0, 1, …, N-1, U-1, …, U. For U independent sequences { X^u[n]Get IFFT to get sequence x^u＝[x^u[0],x^u[1],…,x^u[N-1]]^TSelecting a sequence having a minimum PAPR among the sequencesEmission:

FIG. 4 shows PAPR under different U values of OFDM-SLM system, in order to make the receiver recover the original data block, it needs to store the Side Information (SI) in USI matrix, we choose 10⁵The simulation OFDM signal uses the SLM technology to obtain the PAPR performance when different U values are obtained, and the curve in the graph is drawn from the PAPR performance when different U values are obtainedThe right side and the left side are respectively origin, U is 4, U is 8, U is 16, and U is 32, the result of analysis and simulation shows that compared with the Original OFDM system, the PAPR value is effectively reduced after the SLM technique is applied, and the performance contrast is improved as the U value is increased.

Fig. 5 shows PAPR distribution under different schemes when U is 4, and 10 is selected⁵Simulating each FBMC symbol, wherein the parameters are selected as follows: the period T is 64, the number of subcarriers N is 64, 4-QAM modulation is adopted, the vector rotation vector range is that the oversampling factor K is 4, and the sampling period T is₀Taking 4T, the prototype filter is selected with a span of 4T. From the foregoing analysis, the energy of the FBMC-OQAM signal is mainly concentrated in the 2 nd and 3 rd symbol periods, so the simulation interval selected by the user during simulation is [ T,3T ]]Before calculating the value ratio of the PAPR in this section, [0, 4T ] is calculated]The number of calculations of PAPR values within the range is greatly reduced. Fig. 5 is a comparison between PAPR performance of original FBMC-OQAM signal and PAPR performance of FBMC-OQAM using DSLM and LD-SLM when N is 64. As can be seen from the simulation result in FIG. 5, after the LD-SLM algorithm is adopted, the PAPR performance of the FBMC-OQAM signal is improved by 3.5dB compared with the PAPR performance of the original FBMC signal; similar to the DSLM technique, the difference is only about 0.5dB performance. This shows that the LD-SLM technique proposed herein is similar to DSLM in reducing PAPR performance of FBMC-OQAM system, but has significant improvement compared to the original signal.

Fig. 6 is a diagram showing a comparison of FBMC signals and OFDM signals. We compare the PAPR of FBMC-OQAM signals using LD-SLM technique with the PAPR performance of OFDM signals using conventional SLM technique. As can be seen from fig. 6, the PAPR performance of the FBMC-OQAM signal using the LD-SLM algorithm is very close to the PAPR performance of the SLM algorithm, which means that the scheme proposed herein for reducing PAPR by LD-SLM can make reasonable use of the overlapping property of the FBMC-OQAM signal and can significantly reduce PAPR of the signal. As can also be seen from fig. 6, the PAPR performance of the system is further improved when the number of candidate signals increases.

The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.

Claims (2)

1. A low complexity SLM algorithm (LD-SLM algorithm) for reducing PAPR of an FBMC-OQAM system is characterized by comprising the following steps:

s1, first, initializing the rotation vectors, and generating U phase rotation vectors with length N:

$P^u={\[P_0^u,P_1^u,...,P_{N-1}^u\]}^T,$

wherein:n∈[0,N-1]，u∈[0,U-1]；

s2, multiplying the current data block by the rotation vector, each input data block X_mAre multiplied by U different rotation vectors:

$X_m^u=X_m\·P^u,$

wherein: represents a matrix point-to-point multiplication;

s3, sampling, filtering and modulating the signal, wherein due to the overlapping nature of the FBMC-OQAM signal, the signal before the current data block needs to be considered, so as to obtain the following signals:

wherein: m belongs to [0,2 pi ], T belongs to [0, (m +1/2) T +4T ];

s4, calculating the signal x according to the following formula_u(t) PAPR, the calculation interval being Tc;

${\mathrm{PAPR}}_{T_c}^u=\frac{\underset{0\≤t\≤T_c}{max}|x^u\left(t\right){|}^2}{\frac{1}{T_c}{\∫}_0^{T_c}|x^u\left(t\right){|}^{2}dt};$

s5, according to the result of PAPR calculation, selecting the optimal rotation scheme to obtain the rotation vector number when the PAPR value is minimum, i.e. the rotation vector number is the minimumAnd will u_minStored as side information in U_SI＝[U_SIu_min]Performing the following steps;

s6, finally updating the input according to u_minSelecting the best rotation vector, multiplying the best rotation vector with the current data block, and updating the input signal matrixThen, the process returns to step S2, where the signal X of the next signal cycle is processed_m+1The above steps are repeated until M ═ M-1.

2. The low complexity SLM algorithm for PAPR reduction of FBMC-OQAM system according to claim 1, wherein said LD-SLM algorithm only calculates PAPR values of symbols concentrated in the range of interval [ T,3T ], finding the smallest PAPR value in the interval [ T,3T ] to select the best rotated symbol.