RU2280953C2 - Method for adaptive control of signal peak factor - Google Patents

Abstract

FIELD: electrical communications; adaptive control of signal peak factor.

SUBSTANCE: each modulating character is adaptively controlled to minimize envelope peaks (PAPR); reference signals are generated and peak power is optimally controlled through subcarriers; output OFDM character is interpolated beyond base band; then optimal control is effected for all data and beyond-the-band subcarriers which are then subjected to digital filtration in frequency region, transformed by means of digital backward Fourier transformation into time readings, and the latter are then used for generating reference signals and optimal controls according to mixed RMS criterion thereby minimizing peak factor and modulating-character distortions at a time.

EFFECT: enhanced effectiveness of control.

2 cl, 3 dwg

Description

The invention relates to the field of telecommunications, in particular to methods for controlling a signal, and can be used, for example, in cellular communication systems.

The following specific terms and abbreviations are used to review the prior art and to describe the invention.

- PAPR (Peak-to-Average Power Ratio) - ratio of peak power to average;

- OFDM (Orthogonal Frequency Division Multiplexing) - orthogonal frequency division multiplexing;

- DMT (Discrete MultiTone) - discrete multi-tone (transmission);

- FFT (Fast Fourier Transformation) - fast Fourier transform;

- IFFT (Inverse Fast Fourier Transformation) - inverse fast Fourier transform;

- QAM (Quadrature Amplitude Modulation) - quadrature amplitude modulation;

- QPSK (Quadrature Phase Shift Keying) - 4-position (quadrature) phase shift keying;

- BER (Bit Error Ratio) - bitwise error coefficient or probability of error per bit;

- SNR (Signal Noise Ratio) - signal to noise ratio.

- CCDF (Complementary Cumulativ Distribution Function) - selective probability distribution;

- SPD (Spectrum Power Density) - spectral power density;

- OBR (Out-Band Radiation) - region of out-of-band radiation.

It is known that one of the main disadvantages of DMT and OFDM systems is the high peak factor of the transmitted signals. It arises due to the fact that the OFDM / DMT signal consists of a large number of harmonics that are independently modulated in amplitude and phase. With their coherent (or quasicoherent) addition, “peaks” of the envelope arise, which are characterized by the PAPR (Peak-to-Average Power Ratio), i.e. the ratio of the peak power of the signal to its average power. This effect, if you do not deal with it, requires an increase in the dynamic range of the ADC, DAC and output power amplifier. This leads to their unjustified complication, which means an increase in the cost of the equipment as a whole. An even more important consideration is that PAPR is the main parameter that determines the level of inter-channel interference. For OFDM systems, this is the most vulnerable indicator. Therefore, an effective solution to the problem of reducing PAPR will significantly expand the field of practical application of OFDM technology in cellular communication systems and facilitate their coexistence with other technologies. With a large number of subcarriers and a random nature of the phase changes of individual harmonics (which is usually found in practice), the probability of their quasicoherent addition, and hence the probability of significant envelope peaks, is always non-zero and equal to about 10 ^-6 -10 ^-5 . Therefore, large PAPR values in the OFDM signal are observed quite rarely. On this important fact, most of the implemented methods for reducing PAPR are based.

A known method of controlling the output signal, involving the use of soft or hard clipping (Clipping). Such a method is described in articles by M. Pauli, H. P. Kuchenbecker, "Minimization of the Intermodulation Distortion of a Nonlineary Amplified OFDM Signal". Wireless Personal Communications, Vol.4, No.1, pp. 93-101, Jan. 1997 [1] and Richard van Nee and Ramjee Prasad, "OFDM Wireless Multimedia Communications". - (Artech House Universal Personal Communications Library, 1999) [2].

The essence of such a known method is the nonlinear limitation of the real OFDM / DMT signal at a predetermined level using a special limiter or power amplifier with a saturation characteristic. However, the following side effects occur:

- Violation of orthogonality between subcarrier signals (mutual interference of channels). This leads to energy losses of the order of 0.25 dB (with a decrease in PAPR to 6 dB).

- Increase the level of out-of-band radiation by 8 ... 10 dB. As a result, additional measures are required to filter the signal. However, if such filtering is performed, the PAPR level will increase sharply, and additional clipping measures will be required. The latter will lead to an increase in energy loss to 1.25 or more dB (i.e., we will get a change in the out-of-band emission level for energy efficiency).

A block diagram illustrating a clipping method is shown in FIG.

There is also a more effective way to reduce PARP, it is based on the compensation of the envelope peaks due to the reservation of part of the subcarrier frequencies (Peak Reduction by Tone Reservation). The method is described in published reports by Tellado, J., Cioffi, JM, "PAR Reduction in Multicarrier Transmission Systems", Delayed Contribution ITU-T 4/15, D.150 (WP 1/15), Geneva, February 9-20, 1998 [3] and A. Gatherer and M. Polley, "Controlling clipping probability in DMT transmission," in the Proceedings of the 31st Asilomar Conference on Signals, Systems and Computers (Pacic Grove, CA), Nov. 1997 [4]. The method consists in the fact that the OFDM signal from the output of the IFTT block is subjected to special iterative processing. Namely, from the signal at the moments corresponding to large PARP values, the reduced (scaled) copies of the peaks, which are formed from the same OFDM signal by a special filter circuit, are subtracted. Compensation is made to the required PARP level. Characteristically, with accurate implementation, this method does not lead to the expansion of the signal spectrum and nonlinear distortion of the signal (violation of the orthogonality of the subcarriers). This is possible due to the fact that the compensating signal is formed from that part of the orthogonal subcarriers that are not involved in the transmission of information. They are specially reserved for peak factor management.

However, this method also has its drawbacks:

- If the compensation is performed according to the RMS criterion, and the reserved subcarrier numbers are fixed, then the maximum decrease in PAPR is small. In particular, in [3] it is noted that with a reserve of 5% and a probability of CCDF of 10 ^-5 , you can reduce the PAPR from 15 dB to 11.6 dB (ie, the absolute gain is 3.4 dB) .

- If the numbers of reserved subcarriers for each OFDM symbol are not fixed, but determined in the process of additional optimization, then this gain can be increased to 6.2 dB [3]. However, this procedure complicates the algorithm and has serious limitations in its practical implementation.

Indeed, in this case, on the receiving side, one must either know or very accurately estimate the numbers of all reserved subcarriers. Under the conditions of the Rayleigh channel and QAM modulation, such an estimation task is practically impossible. If the transmission of these subcarrier numbers in each OFDM symbol is envisaged, this will lead to an additional decrease in spectral and energy efficiency.

- Despite the absence of non-linear distortions, the method of compensating the envelope peaks due to the reservation of a part of the subcarrier frequencies (Peak Reduction by Tone Reservation), in its most effective version (the compensation coefficient - “PAPR Reduction Gain” reaches 6 dB or more), leads to a decrease energy efficiency by 0.5-1.0 dB (see [3], p. 5). This does not take into account the above difficulties of practical implementation, which will lead to even greater losses.

Thus, each of the considered methods for reducing PAPR has a number of disadvantages that limit their practical application.

Closest to the claimed method is the Peak Reduction by Tone Reservation algorithm described in [3] (p. 7) for the root mean square quality criterion and a fixed set of reserved subcarriers. A similar algorithm based on peak compensation is also used in the draft international standard OFDMA PHY Proposal for the 802.16.3 PRY Layer, 2001/01/18 [5]. A block diagram of an algorithm of this type is shown in FIG. 2.

The claimed invention solves the problem of increasing the efficiency of signal control in comparison with the known technical solutions, in particular, in such parameters as

- PAPR restriction level;

- level of nonlinear distortion of modulating symbols;

- out-of-band emission level.

This result is achieved due to the fact that, although the claimed method, like the prototype, uses the ideas of clipping and optimal control of peak power through subcarriers, however, instead of backup and control of individual subcarriers, as provided in [3], the claimed method is formed and fed optimal control for all information and out-of-band subcarriers, due to which a high degree of control efficiency is achieved for all the quality indicators listed above and at the same time there is the possibility of taking into account negative effect of interpolation on the peak factor of the output signal. To achieve the desired result, control is added to each modulating symbol of the signal, which minimizes the PAPR of the signal, but at the same time leads only to a slight controlled distortion of the transmitted information. Control optimization is performed according to a mixed rms criterion, in which PAPR and the amount of introduced distortion are simultaneously minimized.

The signal is controlled according to the specified criterion in the expanded spectral region [0, F ₀ ], which includes the base frequency band of the OFDM signal [0, F _max ) and the out-of-band emission region OBR [F _max, F ₀ ], where F ₀ = k _int F _max . The coefficient k _int = 3 ... 4 takes into account the effect of signal interpolation in the DAC on the increase in PAPR. Such a control method has the structure of an optimal controller, in which clipped (in the material domain) subcarrier signals are used as reference signals. By optimizing the management and filtering of out-of-band radiation, it is possible to overcome the disadvantages of the conventional clipping procedure, which were mentioned in the analysis of the prototype.

For a better understanding of the proposed method, Fig. 3 is a structural diagram explaining the principle of operation of the proposed method in the form of an algorithm for reducing PAPR.

Compared with the prototype [3], the circuit contains the following new functional blocks:

- Block 10 control additional OFDM signal (Oversampled Controller). This procedure reduces to adding optimal controls (i.e., the components of the vector u _opt ) to all components of the additional OFDM symbol (with the possible exception of the pilot signal component).

- Block 11 of an out-of-band digital filter (Out of Band Digital Filter), providing attenuation of the spectral components of the OFDM signal lying outside the main band, in accordance with a given mask at each control step. Moreover, at the last control step (i.e., before applying the signal to the DAC input at time t = K), filtering with a softer mask is provided.

- Block 14 additional fast inverse discrete Fourier transform (Oversampled IFFT). Unlike the prototype, it is performed on an additional OFDM symbol having the dimension N _car _ = k _int N _car .

- Block 8 minimizing the peak factor and band distortion (PAPR & In-Band Distortion Minimization) on the basis of the input additional OFDM signal generates reference clipped signals and performs K-step formation of the optimal control vector u _opt . The control algorithm is constructed according to a mixed rms criterion, which simultaneously minimizes the peak factor of the signal and distortion of the modulating symbols. The mathematical description of the algorithm has the form

- расширенный вектор состояния модели OFDM сигнала, включающий опорные клипированные сигналы и отвечающий (i-1)-му шагу управления. На данной схеме приведены лишь те блоки, которые непосредственно участвуют в процедуре уменьшения PAPR, остальные блоки, предшествующие этой процедуре и показанные на Фиг.2 (кодер, модулятор и т.д.), остаются без изменения.where P is the matrix of optimal control,

- extended state vector of the OFDM signal model, including reference clipped signals and corresponding to the (i-1) th control step. This diagram shows only those blocks that are directly involved in the PAPR reduction procedure, the remaining blocks preceding this procedure and shown in Figure 2 (encoder, modulator, etc.) remain unchanged.

When implementing the proposed method carry out the following sequence of operations:

1. The transmitted sequence of information symbols (information signal) is subjected to coding, interleaving and modulation operations, as a result of which a sequence of modulated symbols is formed.

2. The sequence of modulated symbols is divided into separate blocks of length N _car , called OFDM symbols.

3. Each OFDM symbol is complemented by (k _int -1 ...) N _car null characters, thus forming extended OFDM symbols, each of which consists of k _int N _car elements.

4. To each element of the OFDM symbol, control signals generated by the envelope peak compensation unit due to the reservation of part of the subcarrier frequencies within and outside the band ("Reduction by Tone In-Band & Out-Band") are added. As a result of this, a corrected extended OFDM symbol is generated.

5. This adjusted extended OFDM symbol is subjected to additional filtering in the Out of Band Digital Filter block to reduce the out-of-band emission level.

6. The sequences of extended OFDM symbols obtained after filtering are subjected to fast Fourier transform operations and then, by parallel-serial conversion, they are converted into a sequence of time samples.

7. The resulting sequence of time samples on the feedback circuit is sent to the PAPR & In-Band Distortion Minimization block, in which the optimal controls coming to the Reduction by Tone In-Band & Out-Band block are also generated.

The additional advantages of the proposed method include the fact that he

- limits out-of-band emission,

- does not reduce spectral efficiency,

- takes into account the effect of increasing PAPR due to interpolation,

- does not require any changes to the normal structure of the demodulation of the OFDM signal.

To verify the practical operability of the proposed method, statistical modeling was carried out, the description of which is given below. The results of statistical modeling of the implementation of the method showed that, in comparison with the prototype, the inventive adaptive control method can further reduce the PAPR OFDM signal level by an average of 4.5 dB with an average energy loss of the order of 0.1-0.15 dB (according to the BER characteristics on level 0.01) and an out-of-band emission level not exceeding 35 dB.

An analysis of the operation of the algorithm as part of a straight line OFDM simulator was carried out according to three indicators: energy efficiency, envelope peak suppression efficiency and out-of-band emission level. Interference immunity was assessed by BER (SNR) characteristics. PAPR Compensation Efficiency - Based on Selective Probabilistic Distributions CCDF (Complementary Cumulativ Distribution Function)

where m is the number of favorable outcomes (exceeding a given threshold - s), M is the total number of trials. The out-of-band emission level was estimated from the “tails” of the power spectral density (SPD) of the interpolated OFDM signal. SPM was built on empirical data using the autoregressive method. To characterize BER, CCDF, and SPD, the OFDM forward link simulator sim_AdapPAPR.m was developed. It allows you to compare three different PAPR reduction algorithms:

- Algorithm of hard clipping OFDM signal, as in Fig.1.

- The main prototype is the optimal version of the Peak Reduction by Tone Reservation algorithm [3] for the RMS criterion and a fixed set of subcarriers. Its structural diagram is shown in Fig.2.

- Adaptive PAPR-Control Algorithm, described above, with the block diagram shown in Fig.3.

In addition, the program allows you to compare all these algorithms with the classic version of the OFDM signal (without PAPR compensation).

The simulator program used a six-beam radio link model with slow Rayleigh fading and additive white Gaussian noise. The main parameters of the program:

- Dimension OFDM symbol N _car = 64, the number of reserved control characters N _rez = 4.

- The interpolation coefficient OFDM signal K _int = 4.

- Type of QPSK modulation.

- The number of recursive iterations of the control algorithm K = 2.

- OBR mask with frequency response of the raised cosine.

- The individual parameters of each algorithm (clipping thresholds, maximum distortion level of information symbols p, weighting coefficients q, p ∈ [0,1] ∈ control criteria, etc.) were chosen for reasons of the most efficient operation.

The average power of the OFDM samples in the experiment was maintained equal to unity. The values of PAPR and DSP were normalized with respect to the maximum value and determined in decibels.

Compared with the adopted prototype [3], the proposed adaptive PAPR control method allowed to achieve the following results:

1. PAPR peaks with a probability of occurrence were reduced by 4.5 dB.

At the same time, energy losses (according to BER characteristics at the level of 0.01) do not exceed 0.15 dB.

2. The level of out-of-band radiation is almost the same as that of the prototype [3].

Note that due to the PAPR interpolation effect, the gain of the prototype [3] turned out to be almost 2 dB less than stated in the publication.

Claims (2)

1. The method of adaptive control of the peak factor of the signal, which consists in the fact that generate information symbols in the form of a sequence of modulated symbols in the base frequency band [0, F _max ], generate reference clipped signals, characterized in that the sequence of modulated symbols is divided into separate blocks with a length of N _CAR , called OFDM symbols, each OFDM symbol is complemented by (K _int -1) N _CAR , zero symbols, forming extended OFDM symbols, with an extended frequency band [0, F ₀ ], carry out the correction of extended OF DM symbols by supplying additional control signals to each OFDM element symbols, the corrected extended OFDM symbols are digitally filtered in the extended frequency band [0, F ₀ ], the filtered sequence of the extended OFDM symbols is converted using the inverse discrete Fourier transform into a sequence of time samples, which used to generate clipped reference signals intended for optimal control when adjusting extended OFDM symbols. 2. The method according to claim 1, characterized in that they perform multi-step optimal control based on dependencies:

where P is the optimal control matrix;

- extended state vector of the OFDM signal model, containing reference signals and corresponding to the (i-1) -th control step; K is the total number of control steps.

Images