International Journal of Engineering Research and Applications www.ijera.

com
ISSN: 2248-9622, pp. 42-45

RESEARCH ARTICLE OPEN ACCESS

Design and Implementation of the NLMS Adaptive Filter for

Error Minimization and Cancellation
Jitendra P Patil*, Mahesh B Dembrani*, and A B Jayaswal*
*(Department of Electronics and Telecommunication, R C Patel Institute of Technology, Shirpur, Maharashtra.

ABSTRACT
Adaptive filtering techniques is one of the main innovations in the area of DSP and has a broad variety of uses in
both research and industry. Adaptive filtering techniques are widely used in the different applications that
includes, adaptive beam creation, echo cancellation, noise cancellation, and adaptive equalization. Noise
cancellation is a communal spectacle in the field of today's telecommunication networks. Signal intervention
triggered by noise disturbs consumers and decreases the efficiency of communication. This paper mainly
concentrates on the use of the NLMS algorithm to reduce this distractions by increasing information efficiency.
Simulink model is designed to cancel the noise signal using the NLMS algorithm and the error signal is also
calculated.
Keywords – Adaptive Filter, Error, NLMS, Simulink.
becomes unreliable and its performance will
I. INTRODUCTION deviate. [6-9].
Widrow and Hoff in 1959 firstly developed The key drawback of LMS based algorithm
the (LMS) Least Mean Square algorithm based on is that, at every other step it has a set process size
the different studies of pattern recognition. After this variable. This includes an interpretation of the input
study the innovation is widely used in the field of signal statistics before beginning an adaptive data
machine learning algorithms. The LMS process is a processing. In real time applications it is very rarely
technique of adaptive filter documented as stochastic feasible and achievable. Since, we have assumed
gradient-based processes since it practices the that signal input to adaptive process is a signal, there
gradient vector of the filter tap weights to assessed are frequently several influences are added, such as
the improved Wiener approach [2-5]. It is efficiently signal input strength and amplitude, which disturb its
suited for the diffeent applications and for the output [10- 13]. The standardized (NLMS)
computational simplicty. Based on the flexibility it is Normalized Least Mean Square algorithm is a
now the standard algorithms using it all the diifernet development of the LMS algorithm which controls
adaptive filtering techniques are tested and the overall issue of computing the full step size
evaluated. value.
Filter tap weights of adpativer filter are The Step size in the adaptive filtering is
updated on each and every execution of the LMS proportional to inverse of the cumulative projected
algorithm, energy to instant values of input trajectory quantities
is x(n). The summation of the predictable output
x(n): is the time delayed input values, input vector and input data is analogous to the weighted sum of
where x(n) and w(n) is given as: the input vector quantities, and detection of interface
vectors auto-correlation matrix, R [14-18]

x(n) and w(n) are representing the coefficients of

the FIR adaptive filter and tap weight vector at time
instant n. The μ variable is defined as the step size
constraint with a important optimistic variable. The
component size parameter regulates the
accomplishment factor. Selection of appropriate NLMS algorithm based recursion equation is given
value for μ is significant for the comparative growth as:
of the LMS based algorithm. If the value of μ is
small than time required for the convergence gor the
optimazation process is increased. If the value of μ
is small than the perfromance of the adaptive filter

Advancement in Electronics & Communication Engineering (AECE-2020) Page | 42

Raj Kumar Goel Institute of Technology (RKGIT), Ghaziabad, UP, India
International Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, pp. 42-45

Adaptive Filters
The aim of every filter is to remove crucial +
data from unwanted piercing data. Whereas in a d e
(k) ∑ (k)
standard generous of filter design, it is programmed
in advance and then it considers the information of
both signal statistics and unwanted noise. The -
adaptive filter constantly responds to an evolving
Filter
context, over the custom of recursive structures. This
is useful since the signal data is not known until it x y
changes immediately over time. (k) (k)
FIR (Finite Impulse Response) and IIR Adaptive
(Infinite Impulse Response) are two different Algorithm
approaches in the digital filtering techniques. IIR
and FIR the main comparable difference is the
performances. The IIR and FIR filters performance Fig.1 Adaptive filter structure
depends only on inputs. The performance in case of
IIR filter depends on input and output. Whereas, the II. IMPLEMENTATION
Output is reliant on the inputs and outputs received In this paper we have presented the
from the prior outputs. Reliability problems Simulink model for the NLMS based filter. The Step
adjoining the IIR filters does not effects the size constraint is selected based on the existing input
performance of the FIR filters parameters. The performance parameters shows the
Adaptive filtering may be defined to be a NLMS algorithm uniform better constancy for
method, determining the constraints that are used for unknown signals. NLMS algorithm based on this
signal processing change according to certain joint robust convergence speed and relative
requirements. Typically, the benchmark is to obtain computational simplicity makes it perfect for instant
the approximate mean square error or correlation. adaptive echo cancelation [1].
Adaptive filters diverge in time constraints, as the NLMS is certainly development of the
criteria is constantly varying in command to meet consistent LMS algorithm. The practical application
the output stipulations. Adaptive filter is observed as of NLMS algorithms is identical adjacent to that of
filter that achieves an on-line estimate stage. The LMS algorithm. These steps are necessary in the
perception of the output benchmark includes the following order for updating of the NLMS
occurrence of a reference signal, generally covered algorithm.
in the estimated step of standard filter structural 1. Calculate, to get the output of the adaptive filter.
implementation. The wide-ranging for an adaptive
filter structure [15,16] is shown in Figure 1.
Whereas, k: indicates the iteration number, x(k): is
input signal, y(k): is output of adaptive filter, d(k): is
2. To estimate the Error signal, that is difference of
the preferred (desired) signal and e(k): indicates the
desired and filtered signal output.
error signal. Error signal is calculated as given by
d(k)-y (k). The output of the error this also updated
the filter coefficients of the adaptive filter signal is 3. To obtain the step size parameter,
used for the updating the adaption algorithm.
Minimizing the target function ensures that, in some
way, the adaptive filter output signal fits the 4. To update filter tap weights in consent to the
intended signal. following reiteration.

For each and every iteration in NLMS

algorithm it needs the 3N+1 arithmetic that is N
times more than the regular LMS algorithm. This is
an acceptable boost, taking into account increases in
stability and error cancellation.
The fig.2 below shows thesimulink model
for NLMS adaptive filter implementation using
Embedded Matlab function to recover the original
signal from noisy signal which is applied as an input
to NLMS filter.

Advancement in Electronics & Communication Engineering (AECE-2020) Page | 43

Raj Kumar Goel Institute of Technology (RKGIT), Ghaziabad, UP, India
International Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, pp. 42-45

Fig 2 Simulink Model for NLMS adaptive filter using Embedded Matlab Function

III. RESULT

Fig 3 Output scope for NLMS Adaptive Filter using Enbedded MatlabFunction
seems to be data-dependent convergence structure.
The fig.3 above shows the scope of The NLMS algorithm implemented provides
simulink model implemeted. The first window efficient error cancellation and minimization as
shows the combined added input signal. The second compared to the existing adaptive filter structures.
window shows the noisy signal appiled as an input NLMS Simulink model presented in this paper can
to matlab embedded function. The NLMS algorithm be widely used for the different medical and
is used to recover the original signal as ahown in communication.
window third. The last wnidow shows the error
signal. REFERENCES
[1] J. Benesty, H. Rey, L. Rey Vega, and S.
IV. CONCLUSION Tressens, “A nonparametric VSS NLMS
The LMS based adaptive algorithm is widely algorithm,” IEEE Signal Process. Lett., vol.
used due to its simple structure and implementation. 13, pp. 581–584, Oct. 2006.
The LMS algorithm struggles with slowness and

Advancement in Electronics & Communication Engineering (AECE-2020) Page | 44

Raj Kumar Goel Institute of Technology (RKGIT), Ghaziabad, UP, India
International Journal of Engineering Research and Applications www.ijera.com
ISSN: 2248-9622, pp. 42-45

[2] Paleologu, C.; Benesty, J.; Grant, S.L.; Conf. (EUSIPCO96), Trieste, Italy, pp. 33–
Osterwise, C.; “Variable step-size NLMS 36, Sept. 1996.
algorithms for echo cancellation” 2009 [13] Dembrani, M. B., K. B. Khanchandani, and
Conference Record of the forty-third Anita Zurani. "FPGA implementation of low
Asilomar Conference on Signals, Systems and power & high speed accurate ECG signal
Computers., pp. 633-637, Nov 2009. detection using DABLMS
[3] Soria, E.; Calpe, J.; Chambers, J.; Martinez, algorithm." Communication and Computing
M.; Camps, G.; Guerrero, J.D.M.; “A novel Systems: Proceedings of the International
approach to introducing adaptive filters based Conference on Communication and
on the LMS algorithm and its variants”, IEEE Computing Systems (ICCCS 2016), Gurgaon,
Transactions, vol. 47, pp. 127-133, Feb 2008. India, 9-11 September, 2016. CRC Press,
[4] Tandon, A.; Ahmad, M.O.; Swamy, M.N.S.; 2017.
“An efficient, low-complexity, normalized [14] J. Shynk, “Frequency-domain and multirate
LMS algorithm for echo cancellation”, IEEE adaptive filtering,” IEEE Signal Processing
workshop on Circuits and Systems, 2004. Mag., vol. 9, pp. 15– 37, Jan. 1992.
NEWCAS 2004, pp. 161-164, June 2004. [15] Ahmed I. Sulyman and Azzedine Zerguine,
[5] Eneman, K.; Moonen, M.; “Iterated "Echo Cancellation Using a Variable Step-
partitioned block frequency-domain adaptive Size NLMS Algorithm", Electrical and
filtering for acoustic echo cancellation,” IEEE Computer Engineering Department Queen's
Transactions on Speech and Audio University.
Processing, vol. 11, pp. 143-158, March 2003. [16] D. L. Duttweiler, “A twelve-channel digital
[6] Krishna, E.H.; Raghuram, M.; Madhav, K.V; echo canceller,” IEEE Trans. Commun., vol.
Reddy, K.A; “Acoustic echo cancellation 26, no. 5, pp. 647–653, May 1978.
using a computationally efficient transform [17] Keshab K. Parhi and Darren N. Pearson,
domain LMS adaptive filter,” 2010 10th “Low-Power FIR Digital Filter
International Conference on Information Architectures,” IEEE international
sciences signal processing and their Symposium on Circuits and Systems, May
applications (ISSPA), pp. 409-412, May 1995.
2010. [18] Fred J Taylor, “An Analysis of the
[7] Dembrani, Mahesh B., K. B. Khanchandani, Distributed Arithmetic Digital Filter,” IEEE
and Anita Zurani. "Comparative Analysis of Transactions On Acoustics, Speech, And
Various Adaptive Filter Structures Using Signal Processing, Vol.ASSP-34, No.5,
Simulink." Proceeding of International October 1986.
Conference on Intelligent Communication,
Control and Devices. Springer, Singapore,
2017.
[8] Lee, K.A.; Gan,W.S; “Improving
convergence of the NLMS algorithm using
constrained subband updates,” Signal
Processing Letters IEEE, vol. 11, pp. 736-
739, Sept. 2004.
[9] S.C. Douglas, “Adaptive Filters Employing
Partial Updates,” IEEE Trans.Circuits SYS.II,
vol. 44, pp. 209-216, Mar 1997.
[10] E. Soria, J. Calpe, J. Guerrero, M. Martínez,
and J. Espí, “An easy demonstration of the
optimum value of the adaptation constant in
the LMS algorithm,” IEEE Trans. Educ., vol.
41, pp. 83, Feb. 1998.
[11] D. Morgan and S. Kratzer, “On a class of
computationally efficient rapidly converging,
generalized NLMS algorithms,” IEEE Signal
Processing Lett., vol. 3, pp. 245–247, Aug.
1996.
[12] G. Egelmeers, P. Sommen, and J. de Boer,
“Realization of an acoustic echo canceller on
a single DSP,” in Proc. Eur. Signal Processing

Advancement in Electronics & Communication Engineering (AECE-2020) Page | 45

Raj Kumar Goel Institute of Technology (RKGIT), Ghaziabad, UP, India