P ERFORMANCE A NALYSIS O F A DAPTIVE N OISE C ANCELLER E MPLOYING N LMS A LGORITHM

International Journal of Wireless & Mobile Networks (IJWMN) Vol. 7, No. 2, April 2015
DOI : 10.5121/ijwmn.2015.7204 45
PERFORMANCE ANALYSIS OF ADAPTIVE NOISE
CANCELLER EMPLOYING NLMS ALGORITHM
Farhana Afroz1
, Asadul Huq2
, F. Ahmed3
and Kumbesan Sandrasegaran1
1
Faculty of Engineering and Information Technology,
University of Technology, Sydney, Australia
2
Department of Electrical and Electronic Engineering,
University of Dhaka, Bangladesh
3
Department of Computer Science and Engineering, IUB, Bangladesh
ABSTRACT
In voice communication systems, noise cancellation using adaptive digital filter is a renowned technique
for extracting desired speech signal through eliminating noise from the speech signal corrupted by noise.
In this paper, the performance of adaptive noise canceller of Finite Impulse Response (FIR) type has been
analysed employing NLMS (Normalized Least Mean Square) algorithm. An extensive study has been made
to investigate the effects of different parameters, such as number of filter coefficients, number of samples,
step size, and input noise level, on the performance of the adaptive noise cancelling system. All the results
have been obtained using computer simulations built on MATLAB platform.
KEYWORDS
Adaptive Noise Canceller, NLMS, Number of Samples, Step Size, Filter Coefficients, SNR, NRR
1. INTRODUCTION
In communication system, generally different transformational operations are performed on a
signal during information transmission [1]. In signal processing, a signal containing useful
information is passed through a system (e.g. filter, modulator, adder etc.) to process the signal [2].
In noise cancelling, signal processing is concerned with filtering out the noise from the noise-
corrupted signal to recover the signal of interest. The statistics of the noise corrupting a signal is
unknown in many situations and changes with time. Moreover, the power of noise may be greater
than the power of the desired signal being transmitted. In these circumstances, conventional non-
adaptive digital filters may not show satisfactory performances and the noise cancelling should be
an adaptive process i.e. the noise canceller should be capable to adapt itself with changing
environments. Adaptive noise cancellation is an operation of suppressing background noise from
useful signals that is controlled in an adaptive manner in order to obtain improved SNR (Signal to
Noise Ratio) at the receiving end [3, 4]. In general, an adaptive noise canceling system consists of
an adaptive filter, two sensing systems and a subtracting unit. The primary concept of an adaptive
noise cancelling algorithm is to input the noise-corrupted signal to the digital filter which in turn
processes that noisy signal to remove the noise while leaving the useful signal unaffected [5]. The
adaptive filter coefficients get adjusted automatically according to the changes of the input signal

46
characteristics [6]. This paper provides a study of the performance of an adaptive noise canceller
employing NLMS (Normalized Least Mean Square) algorithm. The performance of the system is
analysed while varying a range of parameters such as step size, number of filter coefficients, input
noise level and number of samples.
The rest of this paper is organized as follows. A short literature review of related work is presented
in Section 2. An adaptive noise cancelling system is illustrated in Section 3 followed by
descriptions of adaptive algorithms in Section 4. Section 5 demonstrates the simulation parameters
and results. Finally, Section 6 concludes the paper.
2. RELATED WORK
The field of adaptive filter’s application is highly enriched with a very big volume of literature. In
this section, attempts will be made to report some of these works.
The initial works on adaptive filtering applications can be traced back in the 1950s. In 1959 the
Least Mean Square (LMS) algorithm was invented by Widrow and Hoff in their study of a pattern
recognition scheme known as the adaptive linear threshold logic element [7]. The LMS algorithm
and the idea of stochastic approximation method (developed by Robbins and Monro in [8]) in
statistics for solving sequential parameter estimation problems are closely related with each other.
The basic difference between them is that in case of LMS scheme, the algorithm parameter, (i.e.
step size) which is used to adjust the correction applied to the tap weight vector from one iteration
to the next, is held constant, whereas in stochastic approximation methods the step size is
maintained to be inversely proportional to time n [4]. Godard in 1974, contributed to the
development of adaptive filtering algorithms. Kalman filter theory was utilized in his work to
devise a novel class of adaptive filtering schemes for achieving fast convergence of the transversal
filter’s tap weights to their optimum values [9]. A comparison of the performances of RLSL
(Recursive Least Squares Lattice) and normalized step-size SGL (Stochastic Gradient Lattice)
schemes to that of the LMS transversal scheme for cancelling sinusoidal interferences was made in
[10]. The experimental results showed that the adaptive lattice filters is more advantageous than
LMS transversal filter, which makes them more preferable adaptive noise canceller (ANC) filter
structure if the increased computational cost can be accepted. A comparative study of the
performances of adaptive filtering algorithms was showed in [11]. In this work, the performances
of TV-LMS (time-varying LMS), LMS and RLS (Recursive Least Square) algorithms were
studied and compared in terms of algorithm execution time, required filter order and MSE (Mean
Square Error). A study of mean-square performance of adaptive filters employing averaging theory
was presented in [12]. This paper studied mean-square error and mean-square deviation
performance of adaptive filters along with the transient behavior of the corresponding partially
averaged systems. In [13], a simple neural network namely Adaline was utilized as adaptive filter
for cancelling engine noise in a car. The experimental results in [13] showed that the SNR gets
improved after passing through the noise cancelling system. A new class of nonlinear adaptive
filter namely ANFF (Adaptive Neural Fuzzy Filter) with adaptive fuzziness parameters and
adaptive learning ability was developed in [14]. In addition, an adaptive noise canceller was
simulated to verify the efficiency of the newly developed ANFF. A new approach in noise
cancellation was proposed in [15] in which two adaptive algorithms namely FAP (Fast Affine
Projection) and FEDS (Fast Euclidean Direction Search) algorithms are employed for cancelling

47
noise in speech signal. In addition, the obtained results are compared with the results obtained with
RLS, LMS, NLMS (Normalized LMS) and AP (Affine Projection) algorithms. A DSP-based
oversampling adaptive noise canceller employing RLS algorithm for reducing background noise
for mobile phones was presented as well as the performance of the system was analysed in [16]. A
novel adaptive noise cancelling scheme having low computation complexity was proposed in [17]
for cancelling different kinds of noise in ECG signal. In [18], an adaptive scheme namely
NTVLMS (New Time Varying LMS) has been proposed and the performance of the proposed
algorithm is compared with other well-known adaptive schemes such as NLMS, LMS, NVSSLMS
(New Variable step size LMS), TVLMS and RVSSLMS (Robust Variable step size LMS).
3. ADAPTIVE NOISE CANCELLER
Adaptive noise canceller is utilized to eliminate background noise from useful signals where a
signal of interest becomes submerged in noise. One basic element of an adaptive noise cancelling
system is adaptive filter. A digital filter having self-adjusting characteristics is known as adaptive
filter. An adaptive filter gets adjusted automatically to the changes occurred in its inputs [19]. The
coefficients of the adaptive filter are not fixed, rather these can be changed to optimize some
measure of the filter performance.
In many applications, a frequently encountered problem is the corruption of desired signal by
noise or other unwanted signals. Conventional linear filters, in which the filter coefficients are
fixed, generally can be utilized to extract the signal of interest in some situations when the
frequency bands occupied by the noise and signal are fixed and not spectrally overlapped with
each other. However, many situations exist when there is a spectral overlapping between the
desired signal and unwanted signal or the frequency band occupied by the noise is not known or
changes with time. The filter coefficients can not be defined in advance in such situations and it
must be a variable i.e. the filter characteristics need to be adjusted or altered intelligently according
to the changes in its input signal characteristics in order to optimize its performance.
Fig. 1 shows a model of adaptive noise cancelling system. As seen in figure, an adaptive noise
canceller consists of two inputs (known as primary input and reference input) and an adaptive
filter. The noise-corrupted signal (‫ݕ‬௞ = ‫ݏ‬݇ + ݊݇) is applied as primary input. Reference input is
the noise, ‫ݔ‬݇തതത which is correlated with the main input, ‫ݔ‬݇ in some way but uncorrelated with the
signal, ‫ݏ‬݇. The noise reference input is applied to the adaptive filter and an output, ݊ෝk is estimated
which is a close replica of ݊݇ as much as possible. The adaptive filter readapts itself incessantly so
that the error between ݊݇ and ݊ෝk is minimized during this process. Finally, the recovered signal is
obtained by subtracting the estimated noise, ݊ෝk from the primary input.
As shown in Fig. 1, the adaptive noise cancelling system model contains a band limit noise filter, a
noise reference filter and an adaptive filter. The band limit noise FIR (Finite Impulse Response)
filter is used to make the model more realistic to the environment. Noise is not always white in
nature. This filter allows passing a selected portion of the white noise spectra. As a result, a
colored noise (݊݇) is obtained in the output section. The output of the noise reference filter (‫ݔ‬݇തതത) is
simultaneously fed to the digital FIR filter section and the adaptive weight control mechanism
unit. Thus, a correlative behavior can be established between the noisy component of input and

48
reference noise. The adaptive filter has two parts: an FIR digital filter with adaptable tap weights
or coefficients, and an adaptive algorithm through which filter tap weights can be adjusted or
modified so that error can be minimized [3].
The desired output of the adaptive noise canceller is given by
sˆ k = ek = yk - ݊ෝk
= ‫ݏ‬݇ + ݊݇ - ݊ෝk (1)
where, ‫ݏ‬݇, ݊݇, yk and ek are termed as the useful signal, the band-limited noise, the noise-
corrupted signal and the error signal respectively.
The FIR filter output is [20]
nˆk = ik
N
i
i xw −
−
=
∑
1
0
For N-point filter (2)
Figure 1: A Model of Adaptive Noise Canceller
Where, iw = [wo,w1, …wN-1] are the adjustable filter coefficients [21] and, kx and kn
)
are the input
and output of the filter respectively.
4. ADAPTIVE ALGORITHM
Many adaptive algorithms have been proposed for implementing adaptive filter theory. In this
work, we have applied NLMS (Normalized Least Mean Square) algorithm for studying adaptive
noise cancelling system’s performance. The normalized LMS (NLMS) algorithm may be viewed

49
as a special implementation of the LMS algorithm. In the following sub-sections, LMS and NLMS
algorithms are illustrated.
4.1. Least Mean Square (LMS)
Widrow and Hoff first proposed the LMS algorithm in 1960. It is based on the steepest descent
algorithm. This algorithm modifies the filter coefficients in such a way that ek gets minimized in
the mean-square sense. It is a sequential mechanism which can be employed to adjust the filter tap
weights by continuously observing its input and desired output.
If the filter input vector is x(n) and the desired output vector is d(n), then the filter output, y(n)
and estimated error signal, e(n) can be written as equation (3) and (4) respectively [22].
‫ݕ‬ሺ݊ሻ = ‫ݓ‬்ሺ݊ሻ‫ݔ‬ሺ݊ሻ (3)
݁ሺ݊ሻ = ݀ሺ݊ሻ − ‫ݕ‬ሺ݊ሻ (4)
where, w(n) is the filter tap weight vector.
For LMS algorithm, at each iteration, the weight vector is updated by a small amount according to
the following equation [22]:
‫ݓ‬ሺ݊ + 1ሻ = ‫ݓ‬ሺ݊ሻ + 2ߤ݁ሺ݊ሻ‫ݔ‬ሺ݊) (5)
where, µ is called the algorithm step size. µ is a convergence factor, whose value decides by which
amount the tap weight vector will be changed at each iteration.
4.2. Normalized Least Mean Square (NLMS)
The NLMS (Normalized Least Mean Square) scheme can be seen as a special implementation of
the Least Mean Square (LMS) algorithm which considers the variations in the signal level at the
input of the filter and chooses a normalized step-size parameter that yields in a stable adaptation
algorithm having fast convergence rate. In NLMS algorithm the step size parameter µ is
normalized to µn and the tap weights are updated according to the following equation [22]:
‫ݓ‬ሺ݊ + 1ሻ = ‫ݓ‬ሺ݊ሻ +
ఓ௘ሺ௡ሻ௫ሺ௡ሻ
௫೅ሺ௡ሻ௫ሺ௡ሻାఝ
(6)
Here,
ߤ௡ =
ఓ
௫೅ሺ௡ሻ௫ሺ௡ሻାఝ
where, ߮ is a small constant, used to avoid the numerical instability of algorithm that may arise,
and x(n), e(n), w(n) and µ represent the filter input vector, the estimated error signal, the filter tap
weight vector and the step size respectively.
Normalized LMS algorithm can be viewed as an LMS algorithm with a time-varying step size
parameter. In addition, normalized LMS algorithm leads to faster rate of convergence as compared
with that of the standard LMS algorithm both for correlated and uncorrelated input data [23].

50
5. SIMULATIONS AND RESULTS
The performance evaluation of adaptive noise canceller employing NLMS algorithm is reported in
this section. The performance is analysed with varying some parameters such as step size, number
of filter coefficients, input noise level and number of samples. All the results are obtained using
computer simulations built on MATLAB platform. A recorded speech signal (shown in Fig. 2)
with following characteristics has been taken into consideration to study the system’s
performance.
Number of samples : 24000
Bit rate : 64 kbps
Audio sample size : 8 bit
Audio sampling rate : 8000 Hz
Audio format : PCM
Channels : 1(mono)
Figure 2: A speech signal
5.1. Effects of number of filter coefficients
Twenty observations were made to evaluate the variations of the system’s performance with the
number of filter coefficients. The performance of the system is measured by calculating Noise
Reduction Ratio (in dB). Noise-reduction-ratio (NRR) is the ratio of noise power to the error
power [24].

powerError
powerNoise
NRR = ሺ7ሻ

ܴܴܰ ሺ݀‫ܤ‬ሻ = 10݈‫݃݋‬ଵ଴ሺܴܴܰሻ (8)
The variations of NRR with changing number of filter tap weights are tabulated in Table 1. It is
seen in Fig. 3 that on an average the Noise Reduction Ratio of the system tends to decrease while

51
the number of filter coefficients is increased. It is also observed that NRR reaches its maximum
value of 29.9992dB when the number of filter coefficients is 7 after which it declines with an
increase of number of filter taps.
Table 1: Effects of number of filter coefficients
Simulation parameters
Number of samples: 20000
Noise Power: -16.8004
Step Size : 0.15
Frequency range of colored noise: 1200-2000Hz
Number of filter coefficients Noise Reduction Ratio (NRR) in dB
3 27.0656
4 28.7159
5 28.9636
6 29.1950
7 29.9992
10 29.1705
12 29.4239
14 29.1195
16 29.4878
18 28.6723
20 29.1335
23 29.0453
25 28.8233
28 28.8344
30 27.9594
32 28.1655
50 27.2856
64 27.5548
70 24.6840

52
Figure 3: Variation of NRR (in dB) with number of tap weights
5.2. Effects of step size
The effects of step size (adaptive algorithm parameter) on the performance of the system is
evaluated in this subsection. The step size is increased from 0.01 to 0.2 and the system’s
performance corresponding to respective step size is measured in terms of Noise Reduction Ratio
(NRR). The simulation parameters and the results obtained are tabulated in Table 2. A graphical
representation of tabular data is shown in Fig. 4. It is observed from Fig. 4 that above a particular
value of step size (0.03), the NRR gradually declines with increasing step size. Below that value,
NRR gradually increases with the increase in step size. The optimum step size [25] (at which the
best noise reduction is seen) is 0.03 for the given simulation parameters.
Table 2: Effects of step size
No. of filter coefficients: 32
Step Size Noise Reduction Ratio
(in dB)
.01 26.2926
.02 26.6716
.03 27.1593
.04 26.9859
.05 26.0503
.06 25.4877
.07 24.9614
.08 24.8289
.09 23.4080
0
5
10
15
20
25
30
35
0 10 20 30 40 50 60 70 80
NRR(dB)
Number of filter coefficients

53
0.1 23.9092
0.15 21.6609
0.2 20.3737
The pictorial representation of noisy signal i.e. the original speech signal contaminated with color
noise (upper part), and the recovered speech signal at optimum step size (lower part) are shown in
Fig. 5. It can be observed that the recovered speech signal is identical to the original speech signal.
Figure 4: Variations of NRR with step size
0
5
10
15
20
25
30
0.01 0.06 0.11 0.16 0.21
NRR(indB)
Step Size

54
Figure 5: Noisy signal and recovered signal for step size =0.03
Table 3: Optimum step size with different number of samples
Number of
Samples
Optimum Step Size NRR in dB
20000 0.02 28.4713
15000 0.03 27.1593
10000 0.03 26.9450
It is also observed from the Table 3 that the optimum step size is affected by the number of
samples of the signal while number of filter coefficients and noise power are kept identical in each
case. It is also seen that if the number of samples is decreased under identical simulation
parameters, the NRR of the system becomes worse.

55
5.3. Effects of the number of samples
The effect of number of samples on the performance of the adaptive noise cancelling system is
studied here. The performance is measured in terms of Noise Reduction Ratio (NRR). Results are
tabulated in following table (Table 4) and graphically represented in Fig. 6.
Table 4: Effects of number of samples
Step size: 0.15
Number of samples Noise Reduction Ratio (in dB)
4000 23.0838
6000 21.6214
8000 22.3855
10000 21.9996
12000 21.8909
14000 21.7266
18000 22.2465
20000 22.3013
22000 22.3687
24000 22.6737
Figure 6: Noise Reduction Ratio (in dB) versus number of samples
To study the system’s performance, several observations have been made by increasing the
number of samples from 4000 up to 24000 and measuring the corresponding NRR for respective
number of samples. It is seen from the Fig. 6 that after some initial fluctuations the NRR gradually
becomes stable as the number of samples is increased. The initial fluctuation might be due to the
inadequate number of samples employed for which the filter could not adapt to sudden changes in
the signal. However, it is somewhat difficult to make a firm conclusion regarding this trend at this
stage. More research works are needed to predict the correct reasons regarding this issue.
21
21.5
22
22.5
23
23.5
4000 9000 14000 19000
NRR(dB)
Samples

56
5.4. Effects of input SNR (Signal to Nose Ratio)
The impact of input Signal to Noise Ratio (the ratio of signal power to noise power) on the
performance of the system is analysed in this part. The performance is measured in terms of NRR.
The results obtained under the simulation parameters are given in Table 5 and graphically
represented in Fig. 7.
Table 5: Effects of input SNR on NRR
Step size: 0.03
Input SNR in dB Noise Reduction Ratio (NRR) in dB
3 26.0640
4 26.3715
5 27.9538
6 27.7578
7 27.7716
8 27.6028
9 27.4978
10 28.0481
11 30.0222
12 28.7785
Figure 7: Input SNR versus NRR
As seen from the table, ten observations were made to evaluate the effects of input SNR on the
system’s performance. The input SNR is increased from 3 dB to 12 dB and the respective NRR is
25.5
26
26.5
27
27.5
28
28.5
29
29.5
30
30.5
3 5 7 9 11 13
NRR(indB)
Input SNR (in dB)

57
calculated for each SNR value. It is seen from the Fig. 7 that if the SNR is gradually increased the
NRR is gradually increased till the input SNR is 5 dB followed by a drop of NRR on average with
increasing SNR up to 9 dB. Subsequently, NRR again starts rising with increasing SNR up to 12
dB after which it declines when the SNR is further increased by 1dB. It was also observed that the
system showed best performance at an input SNR of 11dB and worst performance at 3 dB input
SNR.
6. CONCLUSION
In this paper, the performance of adaptive noise canceller using NLMS algorithm has been
evaluated with varying different parameters of the system. The adaptation capability of the system
to any input noise situation as well as the effects of step size, number of filter coefficients, number
of samples and input noise level on the performance of the system are thoroughly studied
considering a speech signal as useful signal. It is evident that individually each of these parameters
has an optimum value at which the adaptive noise canceller showed best performance. Our future
work includes, making a comparison of the performance of adaptive noise cancelling system
employing RLS (Recursive Least Square) and LMS (Least Mean Square) algorithms through
computer simulations and then double check the obtained results by performing the real-time
experiments using DSP hardware.
REFERENCES
[1] Khanna, V. K. (2009), Digital Signal Processing, S. Chand and Company Limited, New Delhi.
[2] C. Durai, R. B. (2005), Digital Signal Processing, Laxmi Publications LTD, New Delhi.
[3] Ifeachor, E. C. and Jervis, B. W. (1993), Digital Signal Processing, Addison-Wesley Publishing
Company.
[4] Haykin, S. (1996), Adaptive Filter Theory, 3rd edition, Prentice-Hall International Inc.
[5] Widrow, B., Glover, J.R., Mccool, J.M., Kaunitz, J., Williams, C.S., Hearn, R.H., Zeidler, J.R., Eugene
Dong, Jr. and Goodlin, R.C. (1975), “Adaptive noise cancelling: Principles and Applications”,
Proceedings of the IEEE, Vol.63(12), pp.1692-1716.
[6] Ramli, R.M., Ali O. Noor, A.O. and Samad, S.A. (2012), “A Review of Adaptive Line Enhancers for
Noise Cancellation”, Australian Journal of Basic and Applied Sciences, 6(6): 337-352.
[7] Widrow, B. and Hoff, M.E. (1960), “Adaptive switching circuits”, Proceedings of WESCON
Convention Record, part 4, pp.96-104.
[8] Robbins, H. and Monro, S. (1951), “A stochastic approximation method”, The Annals of Mathematical
Statistics, Vol.22(3), pp.400-407.
[9] Godard, D. (1974), “Channel equalization using a Kalman filter for fast data transmission”, IBM
journal of Research and Development, Vol.18(3), pp.267-273.
[10] North, R.C., Zeidler, J.R., Albert, T.R. and Ku W.H. (1992), "Comparison of Adaptive Lattice Filters
to LMS Transversal filters for Sinusoidal Cancellation", IEEE International Conference on Acoustics,
Speech, and Signal Processing, Vol. 4, pp. 33-36.
[11] Lau, Y.S. Hossain, Z.M. and Harris, R. (2003), "Performance of Adaptive Filtering Algorithms: A
Comparative Study", Proceedings of the Australian Telecommunications, Networks and Applications
Conference (ATNAC), Melbourne.
[12] Shin, H.C., Sayed, A.H. and Song, W.J. (2004), "Mean Square Performance of Adaptive Filters using
Averaging Theory", IEEE Conference on Signals, Systems and Computers, Vol. 1, pp.229-234.

58
[13] Stella, M., Begusic, D. and Russo, M. (2006), “Adaptive noise cancellation based on neural network”,
International Conference on Software in Telecommunications and Computer Networks, pp.306-309.
[14] Golpayegani, G.N. and Jafari, A.H. (2009), “Improved adaptive neural fuzzy filter and its application
in noise cancellation”, 3rd International Conference on Bioinformatics and Biomedical Engineering,
pp.1-7.
[15] Hadei, S.A. and Lotfizad, M. (2010), “A family of adaptive filter algorithms in noise cancellation for
speech enhancement”, International Journal of Computer and Electrical Engineering, Vol.2, No.2.
[16] Tavares Rayes, J.A., Escamilla Hernandez, E. and Sanchez Garcia, J.C. (2012), “DSP-based
oversampling adaptive noise canceller for background noise reduction for mobile phones”, 22nd
International Conference on Electrical Communications and Computers, pp.327-332.
[17] Mohammed, J.R. (2014), “Low complexity adaptive noise canceller for mobile phones based remote
health monitoring”, International Journal of Electrical and Computer Engineering, Vol.4, No. 3,
pp.422-432.
[18] Siddappaji and Sudha, K.L. (2015), “Performance analysis of new time varying LMS (NTVLMS)
adaptive filtering algorithm in noise cancellation system”, International Conference on
Communication, Information & Computing Technology, pp. 1-6.
[19] Yadev, C.S., Krishna, B.A. (2014), “Study of Different Adaptive Filter Algorithms for Noise
Cancellation in Real-Time Environment”, International Journal of Computer Applications, Volume
96– No.10.
[20] Proakis, J.G. and Manolakis, D.G. (1992), Digital Signal Processing: Principles, Algorithms, and
Applications, New York: Macmillan; Toronto: Maxwell Macmillan Canada.
[21] Sazzad, Z.M.P., Alam, M. S., Moslehuddin, Huq, A. and Ahmed, F. (1999), “Design aspects of
window-based digital FIR filters”, Dhaka University Journal of Science, 47(1): 5-13.
[22] Farhang-Boroujeny, B. (2013), Adaptive Filters: Theory and Applications, John Wiley & Sons.
[23] Perez-Meana, H. (2007), Advances in Audio and Speech Signal Processing: Technologies and
Applications, Idea Group Inc. (IGI).
[24] Rubak, P. Johansen, L.G. & Hermansen, K. (1996), “Adaptive Noise Canceling in Headsets”,
Proceedings of Nordic Acoustical Meeting, NAM’96, Helsinki, Finland, pp. 99-104.
[25] Huq, A., Zhiqiang, M.A. and Kenji, N. (1993), A Method for determining the Optimum Tap
Combination in Pole Zero Adaptive Filters, Technical Report of IECE.

P ERFORMANCE A NALYSIS O F A DAPTIVE N OISE C ANCELLER E MPLOYING N LMS A LGORITHM

More Related Content

What's hot (20)

Viewers also liked (16)

Similar to P ERFORMANCE A NALYSIS O F A DAPTIVE N OISE C ANCELLER E MPLOYING N LMS A LGORITHM (20)

Recently uploaded (20)

P ERFORMANCE A NALYSIS O F A DAPTIVE N OISE C ANCELLER E MPLOYING N LMS A LGORITHM