Efficient Peak Cancellation Method for Reducing the Peak-To-Average Power Ratio in Wideband Communication Systems

Abstract

An efficient peak cancellation method for reducing the peak-to-average power ratio in wideband communication systems uses repeated clipping and frequency domain filtering to achieve a desired peak-to-average power ratio for wideband code division multiple access and orthogonal frequency division multiplexing signals. The maximum magnitude of the filtered pulse is determined by a scaling factor which permits eliminating several iterations while still achieving convergence to the targeted peak-to-average power ratio, thereby reducing computational load and saving hardware resources. This results in improved performance in terms of error vector magnitude, adjacent channel leakage ratio and peak-to-average power ratio.

Description

RELATED APPLICATION

This application incorporates by reference and claims the benefit of U.S. Provisional Patent Application Ser. No. 61/041,164, filed Mar. 31, 2008, and having the same inventors and title as the present application

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention generally relates to wideband communication systems using multiplexing modulation techniques. More specifically, the present invention relates to methods for reducing the peak-to-average power ratio for wideband code division multiple access and orthogonal frequency division multiplexing signals.

2. The Prior Art

As a result of the increasing importance of spectral efficiency in mobile communications, effective modulation techniques, such as wideband code division multiple access (WCDMA) and orthogonal frequency division multiplexing (OFDM), have been used. These modulations have large envelope fluctuations, since the transmitted signal is generated by adding a large number of statistically independent signals. The high peak-to-average power ratio (PAPR) sets strict requirements for the linearity of the power amplifier (PA) leading to low power efficiency, since it is desirable for the PA to operate in its linear region. The use of deliberate envelope clipping to digitally distort the signal while maintaining the signal quality at a sufficient level is a simple and practical way to decrease PAPR. Moreover, the reduced PAPR via clipping gives rise to the possibility of utilizing the dynamic range of the digital-to-analog-converter (DAC) more efficiently. The various PAPR techniques can be categorized into two groups depending on whether they use linear techniques (modulation-and-coding-dependent) or nonlinear techniques (modulation-and-coding-independent). Methods that use linear techniques for OFDM systems do not distort the signal in the time domain so that the spectral properties are not altered.

On the other hand, nonlinear techniques modify the envelope of the time domain signal and are mainly based on clipping-filtering (CF) and peak windowing (PW) clipping. The idea of the PW clipping method is to filter the clipped output signal using the window function with the coefficient weights. The windowed output signal must satisfy the inequality so as to achieve the desired clipping level. To minimize the resultant error in the time domain, the inequality must be as close to equality as possible. This is dependent on the type and length of the window. The resultant function is then multiplied by the delayed input signal [O. Vaananen, J. Vankka, and K. Halonen, “Effect of Clipping in Wideband CDMA System and Simple Algorithm for Peak Windowing,” World Wireless Congress, San Francisco, pp. 614-619, May 2002].

To suppress peak re-growth when filtering the out-of-band distortion of the clipped signal, iterative clipping and filtering for OFDM systems have been used. This approach has suggested iterative clipping and filtering of the clipped pulses, so as to reduce the convergence rate to the targeted PAPR. However, techniques based on repeated clipping and filtering that have been implemented for OFDM systems require several iterations to converge to the desired PAPR level, which implies that it is not an efficient algorithm for hardware implementation [J. Armstrong, “Peak-to-average power reduction for OFDM by repeated clipping and frequency domain filtering,” IEE Electronics Letters, vol. 38, no. 5, pp. 246-247, February 2002], [S. H. Leung, S. M. Ju, and G. G. Bi, “Algorithm for repeated clipping and filtering in peak-to-average power reduction for OFDM,” IEE Electronics Letters, vol. 38, no. 25, pp. 1726-1727, December 2002].

Hence, a need remains in the art for an improved method for reducing the PAPR in wideband communication systems that is able to eliminate several iterations to converge to the desired PAPR level and to simplify the hardware implementation for multi-carrier systems, such as OFDM and WCDMA.

SUMMARY OF INVENTION

Accordingly, the present invention has been made in view of the above problems, and it is an object of the present invention to provide a novel efficient method of peak cancellation (PC) for reducing the PAPR for wideband communication system applications. To achieve the above objects, according to an embodiment of the present invention, the technique is based on a method of repeated clipping and filtering. While conventional repeated peak cancellation (RPC) requires several iterations so as to converge into the targeted PAPR, since filtering causes peak re-growth, the present invention is able to eliminate several iterations, which subsequently saves hardware resources by means of the proper scaling factor.

BRIEF DESCRIPTION OF DRAWINGS

Both the foregoing and further objects and advantages of the invention can be more fully understood from the following detailed description taken in conjunction with the accompanying drawings in which:

FIG. 1. is a schematic diagram showing a multi-stage scaled repeated peak cancellation (SRPC) method.

FIG. 2. is a schematic diagram showing a preferred embodiment of the present invention.

FIG. 3A. is a schematic diagram showing a noise shaper for multi-carrier.

FIG. 3B. is a schematic diagram showing a noise shaper for single-carrier.

FIG. 3C is a schematic diagram showing an embodiment of a clipper

FIG. 4A. is a graph showing a peak cancellation pulse in time domain before filtering, after filtering at each stage, respectively (Prior Art).

FIG. 4B. is a graph showing peak cancellation pulse in time domain before filtering, after filtering, and after filtering and scaling at each stage, respectively.

FIG. 5. is a graph showing simulation results of the PAPR versus EVM for four WCDMA carriers using just clipping method, the PW method and the SRPC method of the present invention respectively.

FIG. 6. is a graph showing simulation results of the ACLR versus PAPR for four WCDMA carriers using the PW method, the RPC method, and the SRPC method of the present invention respectively.

FIG. 7. is a table showing performance comparisons of simulation results of the RMS EVM for different number of WCDMA carriers using the PW method, the RPC method, and the SRPC method of the present invention respectively.

FIG. 8. is a graph showing simulation results of the PDF for four WCDMA carriers using the SRPC method of the present invention respectively.

DESCRIPTION OF PREFERRED EMBODIMENTS

The conventional repeated peak cancellation (RPC) method can effectively reduce the PAPR. However, the RPC method requires several iterations to converge to the desired PAPR level, which implies that it is not an efficient algorithm for hardware implementation. Instead, the present invention applies a scaling factor to the peak cancellation pulse after the noise shaper but inside the peak cancellation loop. The objective is to achieve fewer iterations during processing and thereby reduce the PAPR and EVM. Compared to the conventional RPC method, an embodiment of the present invention achieves lower PAPR for, for example, four WCDMA carriers although approach is expandable into an unlimited number of carriers. The method provided by the present invention is therefore referred to hereinafter as Scaled Repeated Peak Cancellation (SRPC).

Various embodiments of the SRPC method according to the present invention are described in detail below with reference to the accompanying drawings.

FIG. 1. is a schematic diagram showing an embodiment of the multi-stage SRPC method. As illustrated, the baseband signal x(n) 101 goes through the first SPC 102 with a scaling factor α⁽⁰⁾ 107, and z_n⁽¹⁾ 105 is the output from the first iteration of the peak cancellation. After the i-th iteration, the resulting signal can be represented by Z 110.

In the SRPC method of the present invention, as illustrated in FIG. 2, the baseband signal x(n) 201 first passes through the clipper 202. The clipper 202 output, c_n, can be written as follows:

$c_n=\{\begin{array}{cc}\frac{A}{x_n},& x_n>A\\ 1,& x_n\le A\end{array}$

where A is the clipping threshold level. The clipped pulse or peak cancellation pulse, p_n can be written as

p_n=x_n−x_n·c_n

Finally the PAPR reduced signal, z_n 212 is described by

$\begin{array}{c}{z}_n=x_{n-d}-\alpha ·{\mathrm{pf}}_n\\ =x_{n-d}-\alpha ·p_n\star h_n\end{array}$

where pf_n, h_n, and α denote the output signal of the noise shaper 206, the impulse response of the low pass filter (LPF), and the scaler 208, respectively. * denotes the convolution operation.

As shown in FIG. 3a for multi-carrier operation, the peak cancellation pulse 301 is frequency translated by (On), filtered, frequency translated back to baseband and combined. This is because the out-of-band emissions reside between the different carriers and cannot be filtered out by line pass filter 304, as opposed to the single carrier applications in FIG. 3b where only one finite impulse response (FIR) filter 304 can be used. The FIR filters 304 for the multi-carriers have the same coefficients as that of a signal carrier FIR filter 304. There is peak re-growth beyond the clipped signal. This occurs because the resultant peak cancellation pulse (p_n) 301 is filtered by the noise shaper and subsequently subtracted from the delayed input signal. This has the net effect of increasing the peaks beyond that of the clipped signal. Let z_n 212 be the output signal and z_n⁽¹⁾ 105 be the output from the first iteration. After the i-th iteration, the resulting signal 110 can be represented by

$z_n^{\left(2\right)}=z_n^{\left(0\right)}-{\alpha }^{\left(1\right)}·{\mathrm{pf}}_n^{\left(1\right)}$ $\begin{array}{c}{z}_n^{\left(3\right)}=z_n^{\left(2\right)}-{\alpha }^{\left(2\right)}·{\mathrm{pf}}_n^{\left(2\right)}\\ =z_n^{\left(0\right)}-{\alpha }^{\left(1\right)}·{\mathrm{pf}}_n^{\left(1\right)}-{\alpha }^{\left(2\right)}·{\mathrm{pf}}_n^{\left(2\right)}\end{array}$ $⋮$ $\begin{array}{c}{z}_n^{\left(i\right)}=z_n^{\left(i-1\right)}-{\alpha }^{\left(i\right)}·{\mathrm{pf}}_n^{\left(i\right)}\\ =z_n^{\left(0\right)}-\sum _{j=1}^i{\alpha }^{\left(j\right)}·{\mathrm{pf}}_n^{\left(j\right)}\end{array}$

The scaler, α⁽ⁱ⁾, 109, at i-th iteration can be calculated as

${\alpha }^{\left(i\right)}=\frac{\max \left(p_m^{\left(i\right)}\right)}{\max \left({\mathrm{pf}}_n^{\left(i\right)}\right)}$

The envelope of the input signal has a Rayleigh distribution according to the central limit theorem, so that the maximum magnitude of the clipping pulse can be numerically found once the threshold level is set. This implies that the maximum magnitude of the filtered pulse can be accordingly determined.

Referring next to FIG. 3C, an embodiment of a clipper in accordance with the invention is shown in schematic block diagram form. In the embodiment shown, a clipper comprises an amplitude calculator 325 which receives the input signal and provides it to a comparator 327 and a lookup table (LUT) 329. A clipping threshold signal 331, which can be preset or variable according to the desired implementation, provides a second input to the second input to the comparator 327, and also provides an input to a multiplier 333. The output of the LUT provides the second input to the multiplier, the output of which is provided to a mux 335. The output of the comparator 327 provides a “select” input to the mux 335, while a constant 337 provides the second signal input to the mux. Thus, it can be appreciated that the mux selects either the output of the multiplier or a constant, depending on the comparison between the amplitude of the input signal and the clipping threshold. It will be appreciated by those skilled in the art that numerous alternatives and equivalents to the embodiment of FIG. 3C can be constructed given the teachings herein, and the illustrated embodiment is therefore not intended to be limiting and is just one of many that perform the requisite clipping function.

FIGS. 4a and 4b represent peak cancellation pulses in the time domain for the prior art and the present invention, respectively. As shown in FIG. 4b, applying the scaling factor results in less iteration when compared to FIG. 4a. Therefore, this scaling factor significantly reduces the computational load, which saves hardware resources in an implementation. According to numerical simulations, it has been found that two or three iterations of the SRPC is sufficient.

In examining the performance of an embodiment of the SRPC method, 3^rd Generation Partnership Project (3GPP) standard specifications state that the EVM and ACLR at 5 MHz offset should be less than 17.5% and −45 dBc, respectively. The scrambling codes and the time offsets of the time slot duration for multi-carriers test model 1 (TM1) of the WCDMA downlink system is based on 3GPP TS 25.141, Section 6.1.1 of Release 6 (2002-12). The numerical simulations used a signal that is TM1 with 64 dedicated physical channels (DPCH) and 614,400 input samples (one radio frame at 61.44 Msamples/sec) that are processed in MATLAB. A low pass FIR filter with 129 taps was designed to meet out-of-band distortions specifications of −77 dBc.

FIG. 5. is a graph showing simulation results of the PAPR with respect to EVM for four WCDMA carriers using the peak windowing method with an 85 tap Hamming window length, just clipping, and an embodiment of the present invention's SRPC method with three stages of the present invention respectively, through which the performance of the PAPR reduction of the three methods can be compared. In the figure, the solid line with diamond markers represents the performance with just clipping; this sets the lower bound on the PAPR and EVM. It obviously has a large out-of-band spectral radiation. The three-stage PC compressed the PAPR by 0.8 dB more than the single stage at an EVM of 10%. Using the SRPC technique, the PAPR can be suppressed to approximately 5.7 dB at a fixed 10% of EVM after only three stages, while 6.7 dB is achievable with the PW method based on four WCDMA carrier input signal. It should be noted that even a single stage of the proposed algorithm outperforms the PW technique and it requires only two iterations to obtain the same performance that is achieved by seven iterations of the conventional RPC method.

FIG. 6. is a graph showing simulation results of the ACLR versus PAPR for four WCDMA carriers using the peak windowing method, the conventional RPC method, and the SRPC method of the present invention respectively. In the figure, the PW technique has a critical disadvantage that degrades ACLR as opposed to conventional RPC and SRPC method. The original input signal has an ACLR of approximately −77 dBc. Another point to note is that the conventional RPC and SRPC methods deteriorate the ACLR up to approximately 2 dB as the clipping threshold is reduced. This is a result of the decrease in the average power as clipping becomes more significant.

FIG. 7. is a table showing performance comparisons of simulation results of the RMS EVM for different numbers of WCDMA carriers using the PW method, the RPC method, and the SRPC method of the present invention respectively. Simulations were performed for a different number of carriers. For a single carrier, all three techniques represent a similar ability in terms of EVM and PAPR. However, the PW method still allows the ACLR to be compromised, unlike the other two methods. The conventional RPC method requires more than five iterations which increase its complexity, while the proposed SRPC method only requires two iterations. It is not possible for the PW method to achieve a PAPR of 5.5 dB, for the three carrier and four carrier cases, even without considering EVM and ACLR. This is because the window significantly alters many input samples due to the large clipping, which significantly changes the average power.

FIG. 8. is a graph showing simulation results of the PDF for four WCDMA carriers using the SRPC method of the present invention respectively. In the figure, the solid line shows the PDF of the original input signal and the PDF at each stage of three stage SRPC method is illustrated. The PDF difference can be minimized in the region of samples with magnitude less than 1 V, as illustrated in FIG. 8.

In summary, the SRPC method of the present invention, compared to the conventional RPC method, could reduce PAPR more effectively since the SRPC method is able to eliminate several iterations, which subsequently saves hardware resources. In four WCDMA carriers, the present invention could achieve the state of the art performance for WCDMA applications.

Although the present invention has been described with reference to the preferred embodiments, it will be understood that the invention is not limited to the details described thereof. Various substitutions and modifications have been suggested in the foregoing description, and others will occur to those of ordinary skill in the art. Therefore, all such substitutions and modifications are intended to be embraced within the scope of the invention as defined in the appended claims.

Claims

1. A method for reducing peak-to-average power ratio in wideband communication systems using multiplexing modulation techniques comprising the steps of:

(a) clipping a baseband input signal;

(b) subtracting the baseband input signal from the result of said step (a);

(c) noise shaping the result of step (b);

(d) scaling the result of the said step (c); and

(e) subtracting from the result of step (d) the delayed baseband input signal.

2. The method of claim 1 wherein steps (a) to (e) are iterated until a desired clipping level is achieved.

3. The method of claim 1 wherein the clipping step includes using at least one of a group comprising an amplitude calculator, a comparator, a lookup table, a multiplier, a constant and a multiplexer.

4. The method of claim 1 wherein the noise shaping step is performed by converting the digitally clipped signal to a frequency domain signal, filtering by at least one finite impulse response filter, reconverting the filtered frequency domain signal to the baseband signal, and combining to yield an output signal.

5. The method of claim 1 wherein said step (c) is performed by converting the digitally clipped signals for multi-carrier such as WCDMA to frequency domain signals by (ωn), filtering by finite impulse response filters, reconverting the filtered frequency domain signals to the baseband signals, and combining the signals.

6. The method of claim 1 wherein said step (d) is performed by scaling in accordance with the following equation: α ( i ) = max  (  p m ( i )  ) max  (  pf n ( i )  ) wherein α(i) is a scaling factor at i-th iteration, pn the clipped signal or peak cancellation signal, and pfn the output signal of the noise shaper.

7. The method of claim 1 further comprising iterating steps (a) to (e) until a desired clipping level is achieved, and wherein the number of iterations needed to converge to the desired PAPR level is reduced by applying a scaling factor.

8. The method of claim 6 wherein applying a scaling factor reduces computational load for reducing PAPR.

9. The method of claim 6 wherein applying a scaling factor reduces hardware implementation complexity arising from the number of iterations.

10. The method of claim 6 wherein error vector magnitude is significantly improved.

11. The method of claim 6 wherein adjacent channel leakage ratio is significantly improved.

12. The method of claim 6 wherein peak-to-average power ratio is significantly improved.

13. The method of claim 1, further comprising the step of compensating for errors by combining power amplifier output with the signal resulting from step (d) through an additional digital-to-analog converter and an upconverter.