This action might not be possible to undo. Are you sure you want to continue?
Hisham Alasady Department of Electrical and Computer Engineering, Queen’s University, Kingston, Ontario K7L 3N6, Canada
Mohamed Ibnkahla Department of Electrical and Computer Engineering, Queen’s University, Kingston, Ontario K7L 3N6, Canada
Asif Batada Altera Corporation, 101 Innovation Drive San Jose, CA 95134
This paper presents the design procedure and implementation results for a 16 quadrature amplitude modulation (16-QAM) digital data predistortion technique for satellite communications using an Altera digital signal processing (DSP) board. The implementation uses Matlab/Simulink, Altera DSP Builder, and Altera’s Stratix® EP1S80 DSP development board. The design is first implemented in the Matlab/Simulink environment. It is then converted to VHDL level using the signal compiler in the Altera® DSP Builder block. The design is synthesized and fitted with Quartus® II software, and downloaded to an Altera Stratix EP1S80 DSP development board. The results show that using data pre-distortion by modifying the transmitted constellation, we can get a pure (undistorted) constellation at the output of the traveling wave tube (TWT) amplifier. The paper also presents the symbol error rate (SER) performance of a 16-QAM modulation with and without using data pre-distortion. The results show that the SER performance can be improved significantly (about 5 dB) when data pre-distortion is used.
modulation techniques . In this case, multi-level modulation techniques (M-ary) are the favorable candidates. Among all the existing M-ary modulation techniques, M-QAM modulation offers the maximum power/spectral efficiency [3, 8] and appears to be a potentially attractive modulation scheme for future satellite communications. The major bottleneck in achieving high data-rate and high spectral efficiency in satellite communication systems is the presence of onboard, nonlinear high-power amplifiers (HPAs), such as traveling wave tube (TWT) amplifiers. These amplifiers generate nonlinear distortions and spectral re-growth in the communication system, which result in performance degradation, especially for multi-level modulation schemes . For these reasons, most satellite communication systems use constant envelope modulation schemes, such as binary phase shift keying (BPSK) or quadrature phase shift keying (QPSK). These modulation schemes are less affected by the TWT nonlinear distortions, but they are not spectrally efficient. We can increase the spectral efficiency in satellite communication systems by using higher data-rate constellations such as M-QAM schemes, combining them with a pre-distortion technique in order to overcome the TWT nonlinearities. Pre-distortion is a technique that allows communication designers to get a non-distorted constellation at the TWT output. This paper proposes a digital data pre-distortion technique in which the input signal constellation is modified such that we get an optimum constellation
The rapid evolution of global information technology demands high data-rate transmissions via satellites in the presence of available bandwidth, which, in turn, requires spectrally efficient CF-SAT031505-1.0
: where α a . it is clear that as the operating point of the amplifier moves close to the saturation region. respectively. i. α1 = R r1 (3) and α2 = R r2 (4) .e. SYSTEM MODEL We consider in this paper a simplified satellite system model composed of a 16-QAM transmitted sequence and a TWT amplifier. β a = 1 and α p = β p = 1 . and amplitudeto-phase (AM/PM) conversions. r2). TWT amplifiers are commonly used in satellite communication systems for their high power efficiency. β p are constants to be determined in order to fit with the TWT measured data. which can be defined as : 2 αar π α pr (2) (1) and A(r ) = P(r ) = 3 1+ β pr2 1 + βa r 2 -4 -3 -2 -1 0 In-Phase 1 2 3 4 -5 -5 -4 -3 -2 -1 0 In-Phase 1 2 3 4 5 -4 -4 (a) (b) Figure 2. The 16-QAM constellation can be defined by a two-RR and two-phase difference (PD). 2) is defined as the ratio between the outer circle radius R and the inner (resp. with α a = 2. 8. ring ratio (RR) and phase difference (PD)]. second. In this case. Figure 1. 3. Figure 1(a) and (b) show the normalized AM/AM and AM/PM conversions. Altera’s DSP Builder and an Altera DSP board.Quadrature 0 -1 -2 -3 Quadrature at the amplifier’s output [1. DESIGN & IMPLEMENTATION OF 16-QAM DATA PRE-DISTORTER 3. the non-linear distortions increase. It also focuses on the design and implementation of 16-QAM digital data pre-distorter using Matlab/Simulink. The RR 1 (resp. respectively . From these figures. where the transmitted data is pre-distorted in such a way that the output of the TWT provides a non-distorted constellation. respectively. When operated near the maximum power region.1 Design Procedure The 16-QAM constellation can be represented by a circular format with three concentric circles as shown in Figure 3. 2]. Figures 2(a) and (b) show the 16-QAM constellation at the input and at the output of the TWT.e. The remainder of this paper is organized as follows: Section 2 presents the system model and the nonlinear characteristics of TWT amplifiers and their effects on 16-QAM constellations. and 4 symbols. Section 3 describes the design procedure of the data predistorter based on the constellation parameters [i. The non-linear distortions are more severe for multi-level signal constellations (such as 16-QAM). these amplifiers cause non-linear distortions which are modeled in terms of amplitude-to-amplitude (AM/AM). (a) AM/AM and (b) AM/PM conversions 4 3 2 2 1 1 0 -1 -2 -3 16-QAM Constellation at TWT Input 5 4 3 16-QAM Constellation at TWT Output 2. 16-QAM constellation: (a) at the TWT input and (b) at the TWT output One way to overcome this problem is to use a data pre-distortion technique. mid) circle one r1 (resp. this section includes the implementation model using Simulink/Matlab and Altera DSP Builder blocks.. and third circles hold 4. In addition.α p . the first. βa . It is clear that the TWT amplifier severely distorts the transmitted constellation which results in a performance degradation.
For example. by using the model of the amplifier. the AM/AM model is given by: 16-QAM signal Data Pre-distorter TWT Amplifier + Channel output Figure 4(a). 16-QAM constellation and 2 define the PDs between the first outercircle symbol and the nearest center-circle symbol.34126. This operating point is set according to the back-off (BO) value of the TWT [the BO of a TWT is defined (in dB scale) as the difference between its input saturation power and the average input signal ′ and R′ we then find power] . r2 3 1+ r2 (8) θ p1 = 54.8284. given by: π can be used for calculating the phases P( R′ ) and P(rp1) for outer and inner circles. These parameters have been calculated for 16-QAM constellation as: α p1 = 5.5650 . the pre-distorted PD θ p1 . the pre-distorted inner In-Phase Figure 4(b). can be equated θ p1 = P ( R′ ) − ⎡ θ ′ − P ( rp1 )⎤ ⎣ 1 ⎦ P(r ) = (7) Where in (7). Channel model. 5 4 3 2 1 Pre-distorted 16-QAM Constellation Quadrature 0 -1 -2 -3 -4 -5 -5 -4 -3 -2 -1 0 1 2 3 4 5 circle radius would be: 2r A(r ) = (5) 1+ r With A(r) = r ′ and r = rp. For the undistorted (linear) 16-QAM constellation we have: α1 = 3. For the TWT. we calculate the radius R′ of the outer circle from the operating point of the TWT. the TWT AM/PM model. respectively. Pre-distorted 16-QAM constellation . and θ1 = θ 2 = 26. The TWT has been assumed to work at the minimum back-off for 16-QAM. These parameters will be changed when the transmitted data passes through the TWT amplifier. we first 1 (6) Similarly. we can pre-distort the transmitted constellation by changing its parameters such that the TWT output gives the original nondistorted data. and θ p 2 = 46. respectively. α 2 = 1. This can be done based on the TWT model.rp1 = 1 1 − −1 r1′ ′2 r1 R rp1 (5) while the pre-distorted RR α p1 would be: α p1 = as: Figure 3. To overcome this problem. and between the first mid-circle symbol and the nearest inner-circle symbol. See Figure 4(a). Subsequently. the pre-distorted inner circle radius can be found. Figure 4(b) shows the pre-distorted 16QAM constellation.5650 . With the aid of α1 the value of the radius of the inner circle r1′ for the nonlinear case. α p2 = 2.236.84920 . to calculate the pre-distorted parameters α p1 and θ p1 . Additive Noise ′ and PD θ1′ at the output of a calculate the RR α1 linear channel case. The same procedure can be used to calculate the predistorted parameters α p 2 and θ p2 . Here for the TWT.
The input/output black color blocks belong to the Simulink library and are used to generate/display the 16-QAM symbols. Figure 6. The data pre-distorter function is based on the amplitude of the incoming data as follows: 1. r2 = 10 . As shown in Figure 3. the pre-distorter model consists of three sections. then the pre-distorted data will have an amplitude R′ = 18 and an additional phase of 30 degrees. Note that the data predistorter has two inputs/ outputs. The blue blocks belong to the Altera DSP Builder library. Figure 5. Each section predistorts its input amplitude and phase with the pre-calculated values. First. They represent the data pre-distorter itself. the 16-QAM constellation has 12 different phases and three different amplitudes. Hardware compilation steps using signal compiler block . Because the 16-QAM constellation has three different amplitudes. we implemented the data pre-distorter of the 16-QAM constellation using Matlab/Simulink and the Altera DSP Builder blocks. 3.If the input symbol amplitude is r1 = 2 . The predistorted 16-QAM constellation is displayed using a Matlab/Simulink scatter diagram. Matlab/Simulink and Altera DSP Builder model of 16-QAM data pre-distorter The data pre-distorter takes the amplitude and phase of the 16-QAM symbol (generated by Matlab/Simulink blocks) and pre-distorts them using Altera DSP Builder blocks.999 degrees.If the input symbol amplitude is R = 18 . 2.If the input symbol amplitude is r2 = 10 . then the pre-distorted data will have an amplitude r2′ = 1.715 degrees. The incoming data amplitudes are: R = 18 .897 and an additional phase of 9.2 Data Pre-Distorter Implementation This section presents the implementation steps and results of 16-QAM data pre-distorter.728 and an additional phase of 1. Figure 5 shows the block diagram of the Matlab/Simulink model.3. and r1 = 2 . then the pre-distorted data will have an amplitude r1′ = 0.
that the input symbols have three different amplitudes and 12 different phases. 16-QAM With Data Pre-distortion -5 -6 -7 -8 8 10 12 14 16 18 20 SNR in dB 22 24 26 28 30 Figure 8. Acknowledgements This work has been supported in part by the Communications and Information Technology Ontario (CITO) and Altera Corporation. while the second two hexadecimal waveforms represent the equivalent amplitude and phase of the pre-distorted 16-QAM constellation. Conclusion This paper presented the design procedure and implementation results of a 16-QAM data predistorter. 16-QAM Without Data Pre-distortion -3 -4 Rec. After providing the equations to calculate the pre-distorted constellation parameters. In this model we used 8 bits to represent each amplitude and phase. for a SER of 10-7) when data pre-distortion is employed. It is clear that the SER performance improves significantly (about 5 dB. The first two hexadecimal waveforms represent the input amplitude and phase of the 16-QAM constellation. as well as the SER performance results. respectively.Figure 7. Simulation waveform of 16-QAM data pre-distorter 10 0 Figure 6 displays the hardware compilation steps using signal compiler block. SER performance. Figure 7 shows the simulation waveform using Quartus II software. enabled from our predistortion techniques. it presented the steps required to implement a 16-QAM pre-distorter on an Altera DSP board. 10 10 10 SER 10 10 10 10 10 -1 -2 Rec. The results show that satellite communication systems can benefit from 16QAM modulation schemes. while the output symbols have three different amplitudes and 16 different phases. . This result enables the 16-QAM modulation scheme to be efficiently used in satellite communication systems without suffering from non-linear distortions. Figure 8 shows the system SER performance with and without data pre-distortion.3 Symbol Error Rate Performance This section presents the advantages of using data pre-distortion in terms of SER performance. It is clear from Figure 7. 4. 3.
May 2003. New York: McGraw Hill. Sulyman A. 312-339. Vol...  Webb W. T. Nov. 4th ed. Special issue on Gigabit wireless communications: Technologies and challenges. ConferenceMontreal – QB – Canada. Vol. and Ibnkahla M. IEEE CCECE. AlAsady H.10. pp.. “Analysis of Predistortion. M. Oct. 29. 1981. "High Speed Satellite Mobile Communications: Technologies and Challenges".. 45. No. vol. July 1995. February 2004. IEEE CCECE. and Ibnkahla M. 37. and ISI Cancellation Techniques in Digital Radio System with Nonlinear Transmit Amplifiers”.  Ibnkahla M. and Steel R. pp..2223-2231... pp. NO.  Proakis J. Conference. IEEE Trans. Vol. No. 11. Vol. 7. . pp. 1245-1254. “Comparison Between Digital & Analog Predistortion” Proc.  Saleh A.1989.. Dec. Comm. IEEE Trans.. J. 1715-1721. Jun Y.  Karam G. 2. “VariableRate Variable-Power MQAM for Fading Channels”. and Sari H.. G. and Chua S. Rahman Q.REFERENCES  Alasady H. IEEE Trans.Niagara Falls – ON – Canada.. The Proceedings of The IEEE.. 2002.43. pp. and Safwat A.... Commun. Comm.1997. “Performance Evaluation Of A Total Degradation of 16-QAM Constellations Over Satellite Channel” Proc. No. Equalization. “Frequency Independent and Frequency Dependent Nonlinear Model of TWT Amplifier”. Boutros R. G.  Alasady H. “Variable Rate QAM for Mobile Radio”. May 2004. IEEE Trans Comm.1218-1231. 92. Digital Communications.  Goldsmith A. A.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.