International Journal of Electrical & Computer Sciences IJECS Vol: 9 No: 9

- 431 -

Performance Comparison of Wavelet Packet Modulation and OFDM over Multipath Wireless Channel with Narrowband Interference
Usman Khan, Sobia Baig and M. Junaid Mughal

AbstractOrthogonal Frequency Division Multiplexing (OFDM) suffers from high side lobes in transmitted signal which enhance its sensitivity to inter-carrier interference (ICI) and narrowband interference (NBI). In literature, Wavelet Packet Modulation (WPM) has been proposed as an alternative to OFDM. WPM offers much lower side lobes in transmitted signal, which reduce its sensitivity to ICI and NBI. This paper presents a performance comparison of OFDM and WPM over multipath wireless channel with NBI. WPM utilizes a time domain minimum mean square error (MMSE) equalizer for the equalization of multipath wireless channel. It is shown that WPM employing a time domain MMSE equalizer exhibits higher immunity to NBI compared to an OFDM system, over multipath wireless channel. Index TermsWavelet Packet Modulation, narrowband interference (NBI), MMSE equalization.

I. INTRODUCTION FDM is a Discrete Fourier Transform (DFT) based multicarrier modulation (MCM) scheme. It is robust against frequency selective fading as the subcarriers in an OFDM system transform a frequency selective channel into numerous flat fading channels [1]. OFDM is also computationally efficient due to Fast Fourier Transform (FFT) implementation [1]. However, it suffers from high side lobes in transmitted signal, due to rectangular pulse shape of sinusoidal carriers [1]. However, it suffers from high side lobes in transmitted signal, due to rectangular pulse shape of sinusoidal carriers [1]. High side lobes enhance the sensitivity of OFDM systems to ICI and NBI [1], [2]. Various types of windows can be applied to improve the pulse shape of OFDM symbol. Raised cosine filtering has also been considered for this purpose, however it increases the computational complexity of the OFDM system [3], [4]. Therefore, WPM, a Discrete Wavelet Packet Transform (DWPT) based MCM scheme, is proposed as an alternative to OFDM system [5]. WPM offers improved spectral shaping as compared to DFTbased MCM schemes, with comparable computational complexity [4]. It offers lower magnitude side lobes in
Manuscript received September 18, 2009. The authors are with the Faculty of Electronic Engineering, Ghulam Ishaq Khan Institute of Engineering Sciences and Technology, Topi - 23640, District Swabi, Pakistan (e-mails: usmanjadoon83@gmail.com, sobia@giki.edu.pk, junaid@giki.edu.pk).

transmitted signal, which in turn reduce ICI and NBI [5]. In literature, some comparisons of bit error rate (BER) performance have been made between OFDM and WPM in NBI [6], [2]. It has been shown that BER performance of both the schemes is highly dependent on NBI, however, WPM exhibits higher immunity to NBI as compared to an OFDM system [6], [2]. However, these comparisons analyzed the performance of both schemes merely over additive white gaussian noise (AWGN) link and did not evaluate their performance over a multipath wireless channel [6], [2]. Therefore, in order to accurately determine the sensitivity of both schemes to NBI in wireless communications, the effect of multipath wireless channel must be included. This paper presents a comparison of BER performance of WPM and OFDM based communication systems over multipath wireless channel in the presence of NBI. WPM utilizes a time domain MMSE equalizer for the equalization of multipath wireless channel. The OFDM system, utilizes cyclic prefix (CP) in guard band interval to cancel inter symbol interference (ISI). A frequency-domain zero-forcing equalizer in OFDM system compensates for channel distortion suffered by each subcarrier. Section II discusses the system model and section III focuses on the channel equalization in WPM. NBI is described in section IV while section V presents the simulation results of the BER comparison of the two schemes in NBI. Finally conclusions are presented in section VI.

II. SYSTEM MODEL System model of the DWPT based MCM system studied in this paper is shown in Fig.1. The transmitted symbol s ( t ) in a MCM scheme is the sum of amplitude modulated M waveforms m ( t ) as follows,
s (t ) =
M 1 m =0

m ( t )

(1) symbol

modulating the mth waveform m ( t ) [6]. The waveforms are mutually orthogonal and are expressed mathematically as, (2) m ( t ) n ( t ) = ( m n )

where d m is

constellation

encoded mth data

(k ) =

1 for k = 0 0 otherwise

1944091 IJECS-IJENS @ International Journals of Engineering and Sciences IJENS

International Journal of Electrical & Computer Sciences IJECS Vol: 9 No: 9

M Serial to Parallel Converter Parallel to Serial Converter M Constellation Encoding h0[n] h1[n]

- 432 -

packet index, 0 i ( M / 2 1) [7]. The sequences hlow [ n ] and

hhi [ n ] are the discrete impulse responses of the low pass and
Channel

Data In

. . . .

. . . .
M M

. . . .
hM-2[n] hM-1[n]

high pass filters of the QMF bank with perfect reconstruction. While, h j ,i [ n / 2] is the upsampled version of h j ,i [ n ] . The high pass filter can be found from low pass filter using relationship
hhi [ n ] = ( 1) hlow [ L 1 n ] ,
n

AWGN

NBI

(5)

M M Constellation Decoding

h0[-n] h1[-n]

Data Out

. . . .

. . . .
M M

. . . .
hM-2[-n] hM-1[-n]

where L is the length of the sequence. The impulse responses of the decomposition filters are the time reversed versions of the synthesis filters as shown in Fig. 1 and 2. Therefore, the decomposition filters act as matched filters to detect the original transmitted waveform.
Time Domain MMSE Equalizer

III. CHANNEL EQUALIZATION FOR WPM IN MULTIPATH WIRELESS CHANNEL A. Channel Model We have considered a multipath wireless channel for the performance evaluation of WPM and OFDM systems. The received signal in a multipath wireless channel consists of a series of attenuated, time delayed and phase shifted replicas of the transmitted signal [9]. Therefore, the baseband impulse response of a multipath wireless channel can be expressed as,
cim =
N m 1 k =0

Fig. 1.

Block diagram of WPM

where represents a convolution operation. OFDM, a DFT based MCM scheme, utilizes M complex exponentials w ( t ) e( j 2 mt / M ) as the MCM basis functions m ( t ) . These basis functions are limited in time-domain by the window function w ( t ) which corresponds to a sinc-shaped waveform in the frequency domain. WPM utilizes wavelet packets as the MCM basis functions. Wavelet packets are derived through the DWPT [6]. DWPT is implemented using multistage tree-structured quadrature mirror filter (QMF) bank [7]. DWPT decomposes the signal into wavelet packets using analysis side of multistage tree-structured QMF bank. The inverse DWPT (IDWPT) reconstructs the signal as the sum of the wavelet packets using synthesis side of the multistage treestructured QMF bank. In this paper, we have utilized a multistage tree-structured Daubechies QMF bank, which is implemented by Mallets fast algorithm [8]. The synthesis bank constructs a signal as the sum of M = 2 J wavelet packets. These wavelet packets are built by J successive iterations each consisting of filtering and upsampling operations. The wavelet packets are given by the convolution equation h j ,2i [ n ] = hlow [ n ] h j 1,i [ n / 2] (3)
h j ,2i +1 [ n ] = hhi [ n ] h j 1,i [ n / 2] ,

a ( t , ) e
k

2 f c k ( t )

(t k (t ))

(6)

where ak ( t , ) and i ( t ) are the amplitudes and propagation delays, respectively, of the kth multipath component at time t [9]. The exponent 2 f c k ( t ) represents the phase shift encountered due to free space propagation of the kth multipath component. N m is the number of multipaths of the channel and

( t ) is the Dirac delta pulse. The mathematical model of a time-varying multipath channel is represented by (6). The gain a0 ( t , ) at any time t
corresponds to the shortest path and is Rician distributed if there is a line of sight, and Rayleigh distributed if there is no line of sight [9].

(4)

where j is the iteration index, 1 j J and i is the wavelet

hlow[-n] x[n]

hlow[n] xp[n]

+
hhi[-n] Analysis

B. Time Domain Equalization for WPM A time domain linear equalizer is employed to compensate for the signal distortion caused by the multipath wireless channel. MMSE algorithm is utilized to train the equalizer coefficients. It provides a more robust solution than zero forcing equalizer as it considers the channel noise into account [10]. If x is the transmitted signal matrix and c is the channel impulse response vector, then the received signal y is given by, y = cT x + n 0 (7) where n 0 is the channel noise and cT is the transpose of c [10]. The MMSE approach finds the optimum weights w eq for the MMSE equalizer using channel correlation matrix R and channel impulse response vector P , as given below w eq = R -1 P (8)

hhi[n] Synthesis

Fig. 2. Wavelet Packet Elementary Block Decomposition and Reconstruction

1944091 IJECS-IJENS @ International Journals of Engineering and Sciences IJENS

International Journal of Electrical & Computer Sciences IJECS Vol: 9 No: 9

TABLE I

- 433 -

R = E yy
H

(9)

SIMULATION PARAMETERS

P = E yx , (10) where E denotes the expected value, y H is the conjugate

*

transpose of y and x* is the conjugate of x [10]. Then the equalizer output y eq is given by
H y eq = w eq y .

Parameter Data Rate Sampling frequency Modulation Wavelet DWPT Decomposition level No of sub-carriers
0

WPM 2-Mbps 4- MHz 2-ary QAM Daubechies-2 2 -

OFDM 2-Mbps 4-MHz 2-ary QAM 8

(11)

10

We have evaluated the performance of WPM and OFDM over multipath wireless channel in the presence of NBI. Therefore, NBI is briefly discussed here. IV. NARROWBAND INTERFERENCE The NBI usually arises in wireless communication due to interference of different signals at the receiver. It can be modeled by a sinusoidal wave which interferes with the signal of interest at the receiver [2]. The interfering signal I[n] is a sinusoid having frequency fi and power Pi = Ai2 , where Ai is the amplitude of the sinusoid. Then the received signal r[n] is the sum of the transmitted signal t[n] with the interfering signal I[n] [2]. r [ n] = t [ n] + I [ n] (12)
I [ n ] = Ai e j 2 fi n
BER

10

-1

10

-2

10

-3

OFDM WPM 10
-4

0.1

0.2

0.3

0.4 0.5 0.6 Interfering Frequency

0.7

0.8

0.9

(13)

Fig. 3. BER performance comparison of OFDM and WPM as a function of normalized frequency of NBI signal at power 0 dB
10
-1

The level of interference depends upon the power of interfering signal Pi and its frequency fi . OFDM and WPM are both MCM schemes. The energy of an individual symbol in MCM system is concentrated in its respective modulated subcarrier. Therefore, if the interfering frequency fi matches the centre frequency of a sub-carrier then BER increases for that sub-carrier, but it decreases when fi moves away from the centre frequency. In case, the sub-carriers waveforms have null out-of-band energy, the interference will be limited to the subcarrier whose band includes the frequency fi . But in actual systems, sub-carriers waveforms do not have a null out-of-band energy, therefore, interference is caused by the side lobes of the adjacent subcarriers [6]. DWPT based MCM scheme, offers lower magnitude side lobes in transmitted signal as compared to DFT based MCM scheme, and therefore, it may prove to be more robust against NBI. Henceforth, a performance comparison is made between the two schemes in the presence of NBI over multipath wireless channel.

BER

10

-2

OFDM WPM 10
-3

-20

-15

-10 -5 NBI power in dB

Fig. 4. BER performance comparison of OFDM and WPM as a function of power of NBI signal at normalized frequency of 0.5

V. PERFORMANCE COMPARISON OF WPM AND OFDM OVER MULTIPATH WIRELESS CHANNEL In this section, simulation results are presented which show the performance comparison of WPM and OFDM in terms of BER in the presence of NBI over multipath wireless channel with AWGN. The channel is modeled as a six-ray Rayleigh fading channel. WPM system utilizes a time domain MMSE equalizer for the channel equalization, while, OFDM utilizes a

frequency domain zero-forcing equalizer composed of a single tap per subcarriers for channel cancellation. Matlab is used as a simulation tool. Simulation parameters are specified in Table I. We have compared the BER performance for both the schemes as a function of the normalized interfering frequency. Moreover, BER performance is also compared for both the schemes as a function of the power of NBI signal. Signal to noise ratio (SNR) for both the schemes is 20 dB. Results obtained through simulation are shown in Fig. 3 and 4 respectively. The simulation results for both the modulation schemes show that the level of interference is dependent on the

1944091 IJECS-IJENS @ International Journals of Engineering and Sciences IJENS

International Journal of Electrical & Computer Sciences IJECS Vol: 9 No: 9 frequency and power of NBI signal. The simulation results in Fig. 3 present the BER performance as a function of normalized interfering frequency at the NBI power level of 0 dB. The results obtained have shown that BER performance of both modulation schemes is highly dependent on the frequency of NBI signal. However, OFDM system has shown higher BER when the interfering frequency corresponds to the center frequency of a sub-carrier. The WPM system shows the same behavior but with a smoother BER curve. The simulation results in Fig. 4 present the BER performance as a function of power of NBI signal while the normalized interference frequency is arbitrarily chosen to be 0.5. The results obtained have shown that as the power of the NBI signal is increased the BER increases for both the modulation schemes. However, WPM shows more robustness to the power of NBI signal as it provides better BER performance than OFDM. Hence, WPM employing a time domain MMSE equalizer has a better BER performance than OFDM in the presence of NBI over multipath wireless channel. In general, results obtained have shown that WPM system utilizing a time domain MMSE equalizer is capable of high immunity to NBI over multipath wireless channel as compared to an OFDM system. VI. CONCLUSION Performance comparison is made between WPM employing a time domain MMSE equalizer and OFDM in the presence of NBI over multipath wireless channel. It has been established that WPM systems employing a time domain MMSE equalizer exhibits higher immunity to NBI as compared to OFDM systems. NBI usually arises in wireless communication due to interference of different signals at the receiver. Therefore, future modulation techniques have to cope not only with channel distortion, but also with NBI originating from other sources as well. In this scenario, WPM with high immunity to NBI can emerge as a strong contender for modulation techniques in transceivers, for wireless communication. REFERENCES
[1] [2] J. A. C. Bingham, ADSL, VDSL, Multicarrier Modulation, WileyInterscience, 2000. M. Gautier, M. Arndt and J. Lienard, Efficient wavelet packet modulation for wireless communication, in Proc. AICT 2007, Mauritius, May 2007. J. Armstrong, Analysis of new and existing methods of reducing intercarrier interference due to carrier frequency offset in OFDM, IEEE Trans. Commun., vol.47, pp. 365-369, Mar. 1999. H. Murakami, Implementation and Perfect Reconstruction of a Maximally Decimated FIR Filter Bank Using Parallel Module Decomposition, IEEE Trans. Signal Processing, vol.45, no. 2, February 1997. B. G. Negash and H. Nikookar, "Wavelet-based multicarrier transmission over multipath wireless channels," Electronics Letters, vol. 36, no. 21, pp. 1787--1788, Oct. 2000. A. Jamin and P.Mahonen. "Wavelet packet modulation for wireless communication," Wireless Communication and Mobile Computing Journal, vol. 5, no. 2, pp 123--137, Mar. 2005. M. Vetterli, J. Kocacevic, Wavelets and Subband Coding, Prentice Hall, 1995. S. Mallat, A Wavelet Tour of Signal Processing, Published, 1999, ISBN 0-12-466606-X. [9]

- 434 -

T. S. Rappaport. Wireless Communications, Principles and Practice, 2nd edition, Prentice Hall, 2001. [10] J.G. Proakis, Digital Communications. 3rd edition, McGraw-Hill International Editions, 1995.

[3]

[4]

[5]

[6]

[7] [8]

1944091 IJECS-IJENS @ International Journals of Engineering and Sciences IJENS