Compensation of Nonlinear Distortion in OFDM Systems Using an Efficient Evaluation Technique

Document Sample
Compensation of Nonlinear Distortion in OFDM Systems Using an Efficient Evaluation Technique Powered By Docstoc
					                                                               (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                           Vol. 9, No. 6, June 2011

     Compensation of Nonlinear Distortion in OFDM
     Systems Using an Efficient Evaluation Technique
                   Dr. (Mrs.).R.Sukanesh,                                                           R.Sundaraguru,
               Professor, Department of ECE,                                              Research Scholar, Department of ICE,
             Thiagarajar College of Engineering,                                               Anna University Chennai,
                    Madurai - 15, India.                                                           Chennai-25, India.

Abstract— Orthogonal Frequency Division Multiplexing (OFDM)                 proposed in [6], which compensates the nonlinearity at the
signal with larger peak to average power ratio (PAPR) will cause            receiver, but the channel response isn’t accurate. The
the undesirable spectrum re-growth and performance                          algorithm proposed in [7] can mitigate the nonlinear distortion
degradation in bit error rate (BER), both due to the inter-                 and gives better BER performance with the assumption of
modulation products occurring in the nonlinear amplifier at the
transmitter. This paper proposes a new approach to compensate
                                                                            attenuation coefficient is equal to 1, which is not true
the nonlinearity introduced by the HPA. By approximating the                according to Bussgang’s theorem. In this paper a new adaptive
attenuation coefficient of HPA model, the distortion is estimated,          method is proposed, in which the BER performance improved
and then it is subtracted from the received symbol at the receiver.         with moderate complexity in the system.
By performing several iterations, the estimation of the distortion              The remainder of this paper is organized as follows. In
becomes more accurate, and cancels the nonlinear distortion.                Section II, the OFDM transmission system model with
Simulation results show that the presented scheme is more                   nonlinearity is discussed. The proposed compensation
efficient to compensate the nonlinear distortion in OFDM                    technique is introduced in Section III. Section IV presents the
systems.                                                                    simulation results. Conclusions are drawn in section V.
Keywords— Orthogonal Frequency division Multiplexing (OFDM),
Nonlinear Distortion (NLD), High Power Amplifier (HPA), Bit
Error Rate (BER), Peak to Average Power Ratio (PAPR).                                   II.      OFDM SYSTEM MODEL

                 I.    INTRODUCTION

    OFDM has attracted considerable interest among
communication system designers because of its high spectrum
efficiency and robustness to severe multipath fading and it is
widely used in high speed digital communications such as                                      Fig.1 Baseband equivalent OFDM system
digital video broadcasting (DVB), digital audio broadcasting
(DAB), digital subscriber line (DSL) and digital HDTV                           Fig.1 shows the baseband-equivalent functional block
broadcasting systems [1], [2], [3]. However, due to the large               diagram of the OFDM transmission system. The QAM signal
dynamic range of the modulated signal, OFDM is very                         generator produces complex symbols with independent,
sensitive to nonlinear distortions both in the high power                   identically distributed random in-phase and quadrature
amplifier (HPA) stages of the transmitter and in the channel.               components from the finite alphabet set. The serial-to-parallel
The nonlinearity causes (i) spectral-spreading of the OFDM                  block converts the QAM input data stream into a block of N
signal and (ii) intermodulation between subcarriers which                   symbols, which in turn modulate the corresponding subcarrier.
seriously degrade the system performance. To overcome the                   The Nyquist rate sampled OFDM signal is described as,
linearization challenges at the transmitter, several digital
predistortion schemes have been proposed [4], [5]. The basic

idea behind these techniques relies on modeling the                                               ⎛ j 2πkn⎞
                                                                                                  ⎜       ⎟
nonlinearity in HPA and its inverse function first and then                         1             ⎝ N ⎠
                                                                             sn =             Sk e            ,      n = 0,1,L,N − 1,
passing the transmitted signal (before HPA) through the                             N                                                                    (1)
inverse nonlinearity (pre-distorter). However, in order to                              k=0
implement the adaptive predistortion technique in OFDM
systems, a large amount of RAM is required, whose contents                     According to the central limit theorem if the number of
are updated with low convergence speeds. One recent solution                subcarriers is large, the signal can be approximated as a
of this problem is decision-added compensation method                       Gaussian distributed random variable. Using Bussgang’s

                                                                                                                  ISSN 1947-5500
                                                            (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                        Vol. 9, No. 6, June 2011
theorem the signal at the output of nonlinearity can be written          linear model of the OFDM transmission with nonlinearity
as the sum of an attenuated input replica and an uncorrelated            consists of a complex gain ‘α’, and an uncorrelated additive
distortion term [8], [9].                                                Gaussian distortion [10], [11]. The performance of this system
                                                                         is evaluated in the same way as an AWGN channel.
~ = αs + d
  n   n    n                                                (2)                       III.    PROPOSED MODEL

where dn is the distortion term, and ‘α’ is the attenuation
                                                                           Fig. 2 shows the block diagram of a proposed compensation
coefficient, which is described as,
                                                                         technique. The receiver works in an iterative fashion that the

        { }
                                                                         attenuation coefficient ‘α’ of transmitting HPA model is
    E ~n s n
      s                                                                  estimated using the training sequence, which gives the
α=                                                                       imitation of nonlinear distortion components, at last use the
     ⎧ 2⎫                                                                replica to cancel the nonlinear distortion components in the
   E⎨ sn ⎬
                                                                         received symbols.
       ⎩       ⎭                                            (3)

The transmitter and receiver shaping filters have the frequency
response Gt and Gr respectively,

 Gt ( f ) = G r ( f ) =     G ( f ),

where G(f) denotes a raised-cosine Nyquist pulse. The
                                                                                 Fig.2 Proposed Model to Compensate Nonlinear Distortion
spectrally shaped signal at the output of the transmit filter is
fed through the HPA and the channel. The auto-correlation
                                                                            Based on the proposed system the nonlinear signal can be
function of the output signal can be written as,
                                                                         expressed as the sum of the attenuated linear signal αsn and the
                                                                         nonlinear distortion dn.
R ~~ = α R + R
  ss      ss   dd                                           (5)          d n = ~n − αs n
Equation (5) can be used to derive the power of distortion for
different subcarriers. At the receiver, the output of the FFT                The estimated nonlinear distortion term dn is subtracted
block gives a set of decision variables.                                 from the current channel observation to obtain the refined
                                                                         channel signal. By taking the advantage of training sequence,
~                                                                        it is possible to get more accurate channel response. So the
S = αS + D
 k    k    k                                                (6)          output after nonlinear compensation is represented as,

                                                                         s n = rn − d n ∗ h(n )
            N −1

       1                 −j
Sk =               sn   e N                                                 Finally the proposed adaptive algorithm will be more
       N                                                                 effective and compensate the nonlinear distortion.
            n =0
                                                                                   IV.       SIMULATION RESULTS

            N −1

                               2 πkn                                                Only AWGN is assumed to be present in the channel.
        1                 −j
                                                                         The numbers of IFFT and sub-carriers points are 1024 and 512
Dk =               dne           N
        N                                                                respectively. A widely accepted HPA model is a nonlinear
            n =0                                                         memoryless model, in which transformation carried between
                                                                         the complex envelope of the input and output signals [11],
                                                                         [12]. It can be defined as f[ρ] =A[ρ] ·e jΦ[ρ], where the function
~ = αs ∗ h(n ) + d ∗ h(n ) + n
r                                                                        A [·] and Φ [·] represents the AM/AM and AM/PM
 n    n           n            k                            (9)          conversions, respectively. Two nonlinear HPA models have
                                                                         been adopted for simulation. A travelling wave tube amplifier
where h(n) is the channel response assumed to be perfectly               (TWTA) with strong AM/AM and AM/PM conversions, are
known, and nk is the channel noise. Therefore the equivalent             given by [12],

                                                                                                      ISSN 1947-5500
                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                         Vol. 9, No. 6, June 2011
  [ ]   2
Α ρ = Α sat
                         2        2
                     ρ        + Α sat

         π       ρ
  [ ]
Φ ρ =                                                      (13)
         3 ρ 2 + A sat

and for solid state power amplifier (SSPA)

  [ ]=                       ρ                             (14)
         [1 + ( ρ                         ]
A ρ                                            ,
                                     2ρ 1 2ρ
                         Ao      )                                               Fig. 4 BER versus SNR for 16QAM when IBO=8dB

Φ [ρ ] = 0                                                 (15)             Fig. 4 shows the BER performance under the ideal AWGN
                                                                        channel without TWTA. With TWTA, the SNR performance
                                                                        is improved more than 6 dB compared with algorithm in [7].
   where Asat is the input saturation voltage, Ao is the output
saturation voltage, and ‘ρ’ is the parameter that controls the
smoothness of the transition from linear region to saturation
region. In the case of TWTA, Ao = Asat / 2 and for SSPA, Ao =
Asat / 2 . The effect of the nonlinear amplifier depends on the
operating point, which is the average power of the input
signals. Input backoff (IBO) and output backoff (OBO) [8] are
two common parameters to verify the nonlinear distortion.

                As                                         (16)
IBO = 10 log

                 Ao                                        (17)
IBO = 10 log                                                                     Fig. 5 BER versus IBO for 16QAM when SNR =20dB

where As is the input power at the saturation point, Pin is the            In Fig. 5, it is observed that the IBO of the compensated
                                                                        signals with SSPA can be improved more than 1 dB compared
average input power, Ao is the maximum output power, and                with the algorithm proposed in [7].
Pout is the average output power.

             Fig.3 BER versus SNR for 16QAM when IBO=1dB                         Fig. 6 BER versus IBO for 16QAM when SNR =25dB

  Fig.3 shows the BER performance under the ideal AWGN                     Fig.6 shows the IBO of the compensated signals with
channel without SSPA. With SSPA, the performance of the                 TWTA can be improved about 2 dB compared with algorithm
SNR is improved about 5 dB compared with algorithm in [7].              proposed in [7].

                                                                                                   ISSN 1947-5500
                                                                        (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                    Vol. 9, No. 6, June 2011
                   V.      CONCLUSIONS                                                    for MIMO-OFDM systems with nonlinear channel” Proceedings of
                                                                                          ‘International Symposium on Intelligent Signal processing and
   In this paper, a new adaptive algorithm is proposed at the                             Communication Systems,’ pp.113-116, Dec 2005.
receiver to compensate the nonlinearity of the HPA in OFDM                           [7] Jing Yang, Xiaomin Mu, Shouyi Yang, Lin Qi, “Algorithm of nonlinear
systems. By performing several iterations, the estimated                                  compensation in OFDM system,” IEE International conference on
                                                                                          Signal Processing for Communications. Hangzhou, China, pp. 939-942,
distortion becomes more accurate and it is subtracted from the                            Nov 2006.
received signal. This paper presented various computer                               [8] P. Banelli and S. Cacopardi “Theoretical analysis and performance of
simulation results to verify the effectiveness of proposed                                OFDM signals in nonlinear AWGN channels,” in IEEE Transactions on
method. From the computer simulation results, it is confirmed                             Communications, vol. 48, no. 3, pp. 430–441, March 2000.
that the presented method could achieve the higher                                   [9] A. Papoulis Probability, random variables, and stochastic processes,
transmission data rate with better BER performance.                                       3rd ed. New York: McGraw-Hill, 1991.
                                                                                     [10] A. Behravan, F. Munier, T. Svensson, M. Flament,T. Eriksson, A.
  For future works, these techniques will be applied to more                              Svensson and H. Zirath “System implications in designing a 60 GHz
complex MIMO systems.                                                                     WLAN RF frontend,” in Proc. GHz2001 Symposium, Lund, Sweden,
                                                                                          Nov. 2001.
                            REFERENCES                                               [11] E. Costa and S. Pupolin “M-QAM-OFDM system performance in the
                                                                                          presence of a nonlinear amplifier and phase noise,” in IEEE
[1]   H. Sari, G. Karam and I. Jeanclaude, “Transmission techniques for
                                                                                          Transactions on Communications, vol. 50, no.3, pp.462–472, March
      digital terrestrial TV broadcasting," IEEE Commun. Mag., vol. 33, pp.
      100-109, Feb. 1995.
                                                                                     [12] A.A.M. Saleh, “Frequency-independent and frequency-dependent
[2]   B. Le Floch, R. Halert-Lasalle and D. Castellain, “Digital audio
                                                                                          nonlinear models for TWT amplifiers,” IEEE Transactions on
      broadcasting to mobile receivers," IEEE Trans. Consumer Electronics.,
                                                                                          Communications, vol-COM -29, no.11, pp.1715–1720, Nov.1981.
      vol. 35, pp. 493-503, Aug. 1989.
[3]   R. Van Nee and R. Prasad, “OFDM Wireless Multimedia
      Communications," Artech House, 1999.                                                                   AUTHORS PROFILE
[4]   H. Kang, Y. Cho and D.Youn, “On compensating nonlinear distortions
                                                                                     Dr.R.Sukanesh- Professor, Department of Electronics and Communication
      of an OFDM system using an efficient adaptive predistorter," IEEE
                                                                                     Engineering, working at Thiagarajar College of Engineering, affiliated to
      Trans. Commun., vol. COM-47, no. 4, pp. 522-526, Apr. 1999.
                                                                                     Anna University of Technology, Madurai. She is doing research in the area of
[5]   G. Karam and H. Sari, “A data predistortion techniques with memory for         neural network based parameter identification applied to bio-medical systems.
      QAM radio systems," IEEE Trans. Commun., vol. COM-39, pp. 336-
      344, Feb. 1991.                                                                R.Sundaraguru - PhD research scholar in Information and Communication
[6]   Shouyi Yang, Jiangtao Xi, Fang Wang, Xiaomin Mu and Hideo                      Engineering field, doing research at Anna University Chennai. His area of
      Kobayashi “Decision aided compensation of residual frequency offset            interest is interference suppression in wireless OFDM systems.

                                                                                                                      ISSN 1947-5500