Available online at www.sciencedirect.com
ScienceDirect Procedia Computer Science 57 (2015) 368 – 376
3rd International Conference on Recent Trends in Computing 2015 (ICRTC-2015)
Design of Digital Differentiators Using Interior Search Algorithm Manjeet Kumara,כ, Tarun Kumar Rawata, Aman Jainb, Atul Anshuman Singhb, Aviral Mittalb b
a Division of Electronics and Communication Engineering, Netaji Subhas Institute of Technology, Dwarka, Delhi-110078, India Undergraduate, Division of Electronics and Communication Engineering, Netaji Subhas Institute of Technology, Dwarka, Delhi-110078, India
Abstract This paper presents a novel metaheuristic algorithm called Interior Search Algorithm (ISA), which is applied for digital differentiator design problem. ISA is based on the principles of aesthetic techniques commonly used in interior design and decoration. ISA has a very quick convergence rate and only one control parameter. The approach presented here has alleviated from the problems of premature convergence, stagnation and revisiting of the same solution over and over again, which is common in other optimization techniques. Statistical and simulation results have been compared with already existing differentiator design methods such as segment rule, genetic algorithm (GA), simulated annealing (SA), pole zero method (PZ) and particle swarm optimization (PSO). The results affirm that the proposed method outperforms its counterparts in terms of absolute magnitude error and phase error. © byby Elsevier B.V.B.V. This is an open access article under the CC BY-NC-ND license © 2015 2015The TheAuthors. Authors.Published Published Elsevier (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of organizing committee of the 3rd International Conference on Recent Trends in Computing Peer-review under responsibility of organizing committee of the 3rd International Conference on Recent Trends in Computing 2015 2015 (ICRTC-2015). (ICRTC-2015) Keywords: Interior Search Algorithm; Metahueristic Algorithm; Digital Differentiator; PSO; GA.
1. Introduction Digital differentiators are used in wide range of applications such as digital image processing1, radar engineering2, biomedical3 and control systems4. They are useful systems for determining the time derivative of a signal. The frequency response of an ideal differentiator is given by * Corresponding author. Tel.: +91 8860010669. E-mail address:
[email protected] (M. Kumar)
1877-0509 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of organizing committee of the 3rd International Conference on Recent Trends in Computing 2015 (ICRTC-2015) doi:10.1016/j.procs.2015.07.351
369
Manjeet Kumar et al. / Procedia Computer Science 57 (2015) 368 – 376
ܪௗ ሺ߱ሻ ൌ ݆߱
(1)
where ݆ ൌ ξͳ and ߱ is the angular frequency. The digital differentiator design problem essentially comprises of obtaining a class of digital filter that approximates the frequency response of an ideal differentiator as closely as possible. Digital differentiators can be broadly classified into two types, finite impulse response (FIR), output depends only on present and past inputs, and infinite impulse response (IIR), output depends not only on previous inputs, but also on previous outputs. IIR differentiators have impulse response that is theoretically infinite, and due to their recursive nature, large memory space is required to store previous outputs. Whereas, FIR digital differentiator requires limited memory and has wider range of frequency over which it is stable and has a linear frequency response. Due to these aspects, designing of FIR differentiator is easier and less complex compared to IIR differentiators. Although, in applications where ecient processing of signals is required IIR differentiators are preferred. Many methods of designing digital differentiators already exist in literature. Conventionally, digital differentiators are designed by inverting the transfer functions of integrators with suitable modifications. The integrators are in turn designed using simple linear interpolation between the magnitude responses of diơerent Newton-Cotes integrators such as rectangular, trapezoidal and Simpson integrators5,6,7. A wide band digital differentiator based on wide band third-order trapezoidal integrator designed using Newton-Cotes integration formula has been proposed by Ngo8. Al-Alaoui has proposed segment rule to design digital differentiators based on linear interpolation between diơerent Newton-Cotes integration formulas. Some of the differentiators thus obtained were further optimized using simulated annealing9. Gupta et al. have proposed a class of wide band digital differentiators obtained through linear interpolation of popular digital integration techniques, the SKG (Schneider Kaneshige Groutage), trapezoidal rule, rectangular rule and the optimized 4-segment integrator rule9,10,11,12. Secondorder wide band differentiator obtained through optimization of magnitude response and analysis of pole-zero plot has been presented by Upadhyay13. The differentiator thus obtained has a relative error of less than 2%, making it suitable for real-time application. Another class of differentiator obtained by optimization of the pole-zero locations of existing recursive wide band digital differentiators has been formulated by Upadhyay and Singh 14. The proposed differentiator has a relative error of less than 0.48% over the entire Nyquist band. Pei and Hsu have used fractional delay method to design a first order digital differentiator15. All the above mentioned conventional optimization methods have been found to be eƥcient only in case of unimodal optimization problems. In problems involving multimodal optimization, these methods have demonstrated following shortcomings: (i) requirement of continuous and dierentiable error fitness function, (ii) incapability in searching large search space, (iii) premature convergence or convergence to a local optimal solution, (iv) increased sensitivity to starting points when there is an increase in number of variables and (v) requirement of piecewise linear cost function in case of linear programming method. Hence, it may be concluded that the conventional optimization methods are suitable only for use in unimodal optimization problems. To overcome all the drawbacks of conventional methods, several researchers have proposed many new optimization methods, most of which are based on the natural selection and evolution technique. These methods are formally known as heuristic and metaheuristic evolutionary optimization algorithms. They include, Genetic Algorithm (GA), a search algorithm inspired by Darwin’s ”Survival of the fittest”16, Simulated Annealing (SA), which is based on the thermodynamic eơect17, Particle Swarm Optimization (PSO), which simulates the behavior of bird flocking or fish schooling18, Cat Swarm Optimization (CSO), based on the behaviour of cats for tracking and seeking of an object19, Ant Colony Optimization (ACO), based on the ant food searching behaviour 20, etc. GA has shown the most promising potential to provide answers to the shortcomings of conventional methods. GA is capable of solving multidimensional problem that automatically lead to performance tradeo between design specifications. However, GA exhibits a number of shortcomings21; Because of their slow convergence rate, it shows ineciency in finding global minima, and as a result of this, the solution thus obtained is sub-optimal. Another drawback is the inability to find the optimal solution in computationally tractable time in problems involving large search spaces. Many modifications to the GA have been proposed to overcome the aforementioned diculties21. The standard PSO is a population based stochastic search algorithm. The eciency of PSO lies in the fact that it is simple to implement and requires only a few control parameters. In most of the applications, the standard PSO demonstrates better performance than the conventional methods. However, there are cases where this method results
370
Manjeet Kumar et al. / Procedia Computer Science 57 (2015) 368 – 376
in an inferior solution. This is due to the premature convergence and stagnation, that is, the algorithm converges to an unwanted sub-optimal solution and gets stuck. To overcome this, many modifications have been proposed, such as, Craziness PSO (CRPSO)22, and others23,24. Many heuristic evolutionary algorithms, including some of the above mentioned, have been implemented in designing of digital differentiator. Al-Alaoui has proposed a class of wide band digital dierentiators using GA, SA and Fletcher-Powell25. Gupta et. al26 have demonstrated a class of wide band digital differentiator using a modified form of PSO, GA, two variations of PSO27,28,29,30,31 and PSO-GA hybrid techniques32. Recently, a new algorithm called Interior Search Algorithm33 has been proposed. This algorithm takes into account the aesthetic techniques commonly used for interior design and decoration to solve global optimization problems. Like any other metaheuristic algorithm, it works on the principles of diversification and intensification to search for the global best. ISA provides advantages over conventional optimization methods as it requires tuning of only one parameter. It has a faster convergence rate as compared to other algorithms. It also solves the problems of premature and local convergence, which leads to finding the global minimum much more eƥciently. In this paper, ISA has been utilized to obtain a class of digital differentiator. The proposed differentiators are then compared with the differentiators obtained through SA, GA, modified PSO, pole-zero optimization and segment rule. Comparison of diơerent dierentiators has been done graphically on the basis of magnitude responses, magnitude error responses and the phase responses. Further comparisons based on the values of magnitude error and phase error obtained for various dierentiators have also been done. The remaining of the paper is organized as follows: In Section 2, the digital dierentiator design problem is formulated. Section 3 provides a brief review of ISA and its implementation in digital differentiator design. In Section 4, simulation and statistical results with illustrations are presented. Finally, the paper is concluded in Section 5.
Figure 1. Flow chart for Interior Search Algorithm.
371
Manjeet Kumar et al. / Procedia Computer Science 57 (2015) 368 – 376
2. Problem Formulation The input-output relation of an IIR system is governed by the following dierence equation34,35. ெ ݕሺ݊ሻ σே ୀ ܽ ݕሺ݊ െ ݇ሻ ൌ σୀ ܾ ݔሺ݊ െ ݇ሻ
(2)
where y(n), x(n),ܾ ,ܽ are the filter’s output, input, numerator and denominator coeƥcients, respectively, and N(≥ M) is the differentiator’s order. Assuming that ܽ = 1, the transfer function of the differentiator can be written as
ܪሺ ݖሻ ൌ
షೖ σಾ ೖసబ ೖ ௭ షೖ ଵାσಿ ೖసభ ೖ ௭
(3)
The corresponding frequency response of IIR differentiator becomes: σಾ
ܪሺ߱ሻ ൌ ଵାσೖసబ ಿ
ೖ షೕೖഘ
ೖసభ ೖ
షೕೖഘ
(4)
The values of the coefficients (ܽ ,ܾ ) of the transfer function will determine the type of filter designed (FIR or IIR). In this paper, a novel fitness function is adopted to achieve small magnitude error which is given as: ଵ
ܬൌ ሺȁܪௗ ሺ߱ሻȁ െ ȁ ܪሺ߱ሻȁሻଶ ݀߱
(5)
The cost function is specifically used to minimize the error between the ideal and approximated graphs in the lower and higher frequency ranges. Table 1. Control parameters of ISA Parameter
Value
Population Size
25-50
No. of iteration
50-100
Lower bound (LB)
-2
Upper bound (UB)
2
Alpha (α)
0.2
3. Interior Search Algorithm (ISA) ISA, also known as aesthetic search algorithm, is based on the aesthetic techniques used in interior design and decoration. This algorithm was originally given by Gandomi in 201433. It is a robust and flexible algorithm that can be used to solve wide domains of optimization problems. It is a novel method for solving optimization tasks, has a very quick convergence rate and a capability to handle large search space. This method has been previously used to solve traditional engineering design problem, like welded beam design, pressure vessel design and gear train design33. In this paper ISA has been used to solve the digital differentiator design problem. The flowchart for ISA process is shown in Fig. 1. The implementation steps for digital differentiator design problem using ISA are as follows33. Step 1: Randomly generate a population of elements ݔ between lower bounds (LB) and upper bounds (UB) and find their fitness values, ݂൫ݔ ൯.
372
Manjeet Kumar et al. / Procedia Computer Science 57 (2015) 368 – 376
Step 2: Find the fittest element in the ݆௧ iteration࢞ . For this optimization problem, the element with minimum function value is defined as the global best. Step 3: Randomly divide other elements into two groups, composition group and mirror group, by comparing with a tuned parameter α. If ݎ İ α, the element goes into the mirror group, otherwise it goes into the composition group. This step is important as a balance needs to be maintained between diversification and intensification. Step 4: In the composition group, change each element randomly within the limited search space (LB and UB).
ݔ ൌ ࡸ ሺࢁ െ ࡸ ሻݎଶ
(6)
where ݎଶ is a random value between 0 and 1. Step 5: For the elements in mirror group, place a mirror randomly between each element and the fittest element. The position of the mirror is evaluated by using the formula given below:
ିଵ
࢞ǡ ൌ ݎଷ ࢞
ሺͳ െ ݎଷ ሻ࢞ࢍ࢈
(7)
where ݎଷ is a random value between 0 and 1. The location of the image or virtual location of the element depends on the mirror location and can be formulated as:
ିଵ
࢞ ൌ ʹ࢞ǡ െ ࢞
(8)
Step 6: The location of the global best is slightly changed by using random walk. Random walk works as local search because it searches around global best.
ିଵ
࢞ࢍ࢈ ൌ ࢞ࢍ࢈ ߣ࢘
(9)
where λ is a scale factor which is set according to size of the search space. Here, λ is set to be 0.01 × (UB − LB). Step 7: Calculate the fitness values of new locations of the element and images. If the new value is better, replace it, otherwise keep the initial value. Step 8: Repeat steps 2 to 7 until the stopping criteria (maximum number of iterations) is met. The global minimum fitness value and its corresponding element values are used as coeƥcients in optimal differentiator design. 4. Simulation Results To obtain the proposed differentiators, MATLAB simulation has been extensively performed. The best chosen parameters used for ISA optimization algorithm are reported in Table 1. All the MATLAB simulations have been performed in MATLAB 7.12 version on Intel Core(TM), 3.20 GHz with 4 GB RAM. The best optimal coeƥcients for the designed digital differentiator of second, third and fourth orders have been calculated using ISA, GA, SA, PSO and PZ and are reported in Table 2. In order to show the eƥciency of the proposed differentiator design method, several other differentiators obtained through other methods have been tabulated in Tables 2 and 3, to facilitate comparison. The second, third and fourth order differentiators proposed in this paper have been designed for the normalized frequency range of 0 İ ω İ 1. Table 3 presents a comparison of total absolute magnitude error and maximum phase error of diơerent reported differentiators. They have also been compared graphically in Figs. 2−4. These comparisons have been done on the basis of obtained magnitude response curve, magnitude error curve and phase curve. From Table 3 it can be seen that the proposed differentiators have total absolute magnitude error of 1.609128, 1.323152 and 1.228652 and maximum phase error of 89.9095, 89.9110 and 99.6842, in case of second, third and fourth order differentiators respectively. On the basis of above discussion, it can be concluded that the proposed differentiators are superior to other differentiators tabulated in Table 3. In Figs. 2(a), 3(a), 4(a) it can be observed that the proposed differentiators closely approximates the ideal magnitude response over the entire frequency range.
373
Manjeet Kumar et al. / Procedia Computer Science 57 (2015) 368 – 376
From Figs. 2(b), 3(b), 4(b) it is evident that the deviation of magnitude from ideality varies least in case of proposed differentiators over a wide range of frequency. The phase deviations for diơerent order digital differentiators are depicted in Figs. 2(c), 3(c), 4(c). In view of the above facts, it can be concluded that ISA performs better than the above reported literature.
Table 2. Optimal coefficient of digital differentiator of different algorithms Reference
Method
Order
Numerator coefficients
Denominator coefficients
Al-Alaoui9
3- Segment
3
0.01903 -0.029052
1.0000 0.1846
1.123 -1.181
-0.001748 0.03484
4-Segment
4
1.09677 0.0433
1.0000 1.19769 1.21566
-0.015308 -1.11198
1.197690 0.21580
1.1543 -0.4464
1.0000 0.7945
-0.7079
0.0832
25
Alaoui and Baydoun
GA GA
2 3
1.1533 -0.4432
-0.7060
1.0000 0.7981 0.0884
-0.0041
0 1.0000 0.9050 0.1713
GA
4
1.1553 -0.3170 -0.7560
SA
2
1.1538 -0.5408 -0.613
1.0000 0.7121 0.0607
SA
3
1.1555 -0.358 -0.714
1.0000 0.8662 0.1612
-0.0833
0.0028
SA
4
1.1540 0.2290 -0.8794
1.0000 1.3788 0.623
-0.4486 -0.0549
0.1059 0.0059
0.9269 -0.4628
1.0000 0.9883
-0.0817 Alaoui and Baydoun25
26
Gupta et al.
PSO
13
D.K Upadhyay
Present Study
2
-0.0006
0.0066 0
-0.99996
0.1595 1.0000 1.0000 0.2759
PSO
3
0.3237 1.0000 -0.7133 -0.6124
0.1595
PSO
4
0.1806 1.0000 -0.7133
1.0000 0.55207 -0.1578
0.4568 0.2611
-0.0708 -0.01
1.161 -0.79856
1.0000 0.6844
-0.55728
0.0676
1.144341 -0.574021
1.0000 0.696004
PZ ISA
2 2
-0.599883
0.063291
ISA
3
1.155408 -0.359594
1.0000 0.865227
-713884 -0.079311
0.158882 0.003219
ISA
4
1.156659 0.637118
1.0000 1.730208
-0.904285 -0.751576
1.016742 0.222981
-0.136173
0.011178
374
Manjeet Kumar et al. / Procedia Computer Science 57 (2015) 368 – 376 Table 3. Comparison of performance criteria attained by already existing literature Reference
Method
Order
Al-Alaoui9
3- Segment 4-Segment 25
Alaoui and Baydoun
Alaoui and Baydoun25
Gupta et al.26
Total Absolute Magnitude Error
Maximum Phase Error
3
5.503161
269.739995
4
13.040399
172.601481
GA
2
2.084897
89.900552
GA
3
1.858099
89.902310
GA
4
1.318132
89.784931
SA
2
1.596657
89.909167
SA
3
1.511931
89.912323
SA
4
1.381094
89.601356
PSO
2
36.687205
89.998382
PSO
3
91.464599
89.715870
PSO
4
3.371737
179.770729
D.K Upadhyay13
PZ
2
4.793251
89.910229
Present Study
ISA
2
1.609128
89.909579
ISA
3
1.323152
89.911009
ISA
4
1.228652
99.684216
5. Conclusions The paper proposes a novel class of digital differentiators which have been approximated using the ISA algorithm. The ISA algorithm is simple as it has only one tuning parameter and thus has a faster convergence rate. The algorithm was simulated for the design of second, third and fourth order differentiators and the optimal coeƥcients of the transfer functions were derived. The magnitude response of the proposed differentiators accurately approximates the ideal differentiator over the entire Nyquist frequency range. Hence, the proposed differentiators are either comparable or better than the existing differentiators. Therefore, the proposed digital differentiators are viable alternatives over the existing digital differentiators. This work can be further extended for the design of digital integrator, two-dimensional differentiator, Hilbert transformer and many more attractive real time signal processing applications.
Magnitude
2.5
Absolute Magnitude Error
Ideal ISA SA GA PSO P-Z
3
2
1.5 2.965 2.96
0.05
100
0.045
80
0.04
60
0.035
40
0.03
20 Phase (deg)
3.5
ISA SA PSO GA P-Z
0.025
0.02
0 47
-20
46
Ideal ISA SA PSO GA P-Z
2.955
1
2.95
0.015
-40
0.01
-60
0.005
-80
0
-100
2.945 2.94 2.935 2.93
0.5
2.925 2.92 2.915 0.922
0
0
0.1
0.2
0.3
0.924
0.926
0.4 0.5 0.6 Normalized Frequency
(a)
0.928
0.93
0.7
0.932
0.934
0.8
45
44
43
42
41 0.635
0.64
0.645
0.65
0.655
0.66
0.665
0.936
0.9
1
0
0.1
0.2
0.3
0.4 0.5 0.6 Normalized Frequency
(b)
0.7
0.8
0.9
1
0
0.1
0.2
0.3
0.4 0.5 0.6 Normalized Frequency
0.7
0.8
0.9
1
(c)
Figure 2. Comparison of second order digital differentiator using ISA and other reported work (a) magnitude response (b) absolute magnitude error (c) phase response.
375
Manjeet Kumar et al. / Procedia Computer Science 57 (2015) 368 – 376
3.5
0.12 Ideal ISA SA PSO GA 3 Segment
2
1.5 2.96
2.955
1
0.08
80 60 40 20
Phase (deg)
Magnitude Response
2.5
0.1
Absolute Magnitude Error
3
100 ISA SA PSO GA 3 Segment
0.06
0 37.8
-20
37.78 37.76
0.04
Ideal ISA SA PSO GA 3 Segment
-40
2.95
-60
2.945
0.02
0.5
2.94
-80
2.935 0.935
0 0
0.1
0.2
0.3
0.936
0.937
0.938
0.939
0.4 0.5 0.6 0.7 Normalized Frequency
0.94
0.941
0.8
0.9
0
1
0
0.1
0.2
0.3
0.4 0.5 0.6 0.7 Normalized Frequency
(a)
0.8
0.9
-100
1
37.74
0
0.1
0.2
37.72 37.7 37.68 37.66 37.64 37.62 0.6961 0.6962 0.6963 0.6964 0.6965 0.6966 0.6967 0.6968 0.6969
0.3
0.4 0.5 0.6 0.7 Normalized Frequency
(b)
0.8
0.9
1
(c)
Figure 3. Comparison of third order digital differentiator using ISA and other reported work (a) magnitude response (b) absolute magnitude error (c) phase response.
3
Magnitude
2.5
Absolute Magnitude Error
Ideal ISA PSO GA SA 4 segment
2
1.5 2.885 2.88 2.875
1
2.87 2.865
0.05
100
0.045
80
0.04
60
0.035
40
ISA SA PSO GA 4 segment
0.03 0.025 0.02
Phase (deg)
3.5
20 0 55.18
-20
0.015
-40
0.01
-60
55.16
Ideal ISA PSO GA SA 4 segment
2.86 2.855 2.85
0.5
2.845
-80
2.84
0.005
2.835 0.9
0 0
0.1
0.2
0.3
0.905
0.4 0.5 0.6 0.7 Normalized Frequency
(a)
0.91
0.8
0.915
0.9
1
0
-100
0
0.1
0.2
0.3
0.4 0.5 0.6 0.7 Normalized Frequency
(b)
0.8
0.9
1
0
0.1
0.2
0.3
55.14 55.12 55.1 55.08 55.06 55.04 55.02 55 0.4904 0.4905 0.4906 0.4907 0.4908 0.4909 0.491 0.4911 0.4912 0.4913
0.4 0.5 0.6 0.7 Normalized Frequency
0.8
0.9
1
(c)
Figure 4. Comparison of fourth order digital differentiator using ISA and other reported work (a) magnitude response (b) absolute magnitude error (c) phase response.
Acknowledgements Authors are thankful to Netaji Subhas Institute of Technology for providing excellent research facilities and environment in the Digital Signal Processing Lab at Electronics and Communication Engineering department. References 1. Gonzalez, R. C. and Wintz, P. Digital image processing. 2nd edn. Addison Wesley Publishing Company, Reading, Mass; 1992. 2. Skolnik, M. I. Introduction to radar systems. McGraw-Hill, New York; 1980. 3. Tompkins, W. J. and Webster, J. G. Design of microcomputer-based medical instrumentation. Prentice Hall Professional Technical Reference; 1980. 4. Franklin, G. F., Powell, I. D. and Workman, M. L. Digital control of dynamic systems. 2nd edn. Addison Wesley, Reading, Mass.; 1990. 5. Al-Alaoui, M. A. Novel approach to designing digital differentiators. Electronics Letters, 1992: 28-15, 1376-1378. 6. Bihan, J. L. Novel class of digital integrators and differentiators. Electronics Letters, 1993: 29-11, 971-973 7. Al-Alaoui, M. A. Novel IIR differentiator from the Simpson integration rule. Circuits and Systems I: Fundamental Theory and Applications, IEEE Transactions, 1994: 41-2, 186-187.
376
Manjeet Kumar et al. / Procedia Computer Science 57 (2015) 368 – 376 8. Ngo, N. Q. A new approach for the design of wideband digital integrator and differentiator. Circuits and Systems II: Express Briefs, IEEE Transactions, 2006: 53-9, 936-940. 9. Al-Aloui, M. A. Novel class of digital integrators and differentiators. IET Signal Processing, 2011: 5-2, 251-260. 10. Gupta, M., Jain, M. and Kumar, B. Recursive wideband digital integrator and differentiator. International Journal of Circuit Theory and Applications, 2011: 39-7, 775-782. 11. Gupta, M., Jain, M. and Kumar, B. Novel class of stable wideband recursive digital integrators and differentiators. IET signal processing, 2010: 4-5, 560-566. 12. Schneider, A. M., Kaneshige, J. T. and Groutage, F. Higher order s-to-z mapping functions and their application in digitizing continuous-time filters. Proceedings of the IEEE, 1991: 79-11, 1661-1674. 13. Upadhyay, D. K. Recursive wideband digital differentiators. Electronics letters, 2010: 46-25, 1661-1662. 14. Upadhyay, D. K. and Singh, R. K. Recursive wideband digital differentiator and integrator. Electronics Letters, 2011: 47-11, 647-648. 15. Pei, S. C. and Hsu, H. J. Fractional bilinear transform for analog-to-digital conversion. Signal Processing, IEEE Transactions, 2008: 56-5, 2122-2127. 16. Goldberg, D. B. Genetic Algorithms in Search Optimization and Machine Learning. Addison-Wesley, San Francisco; 1989. 17. Davidson, R. and Harel, D. Drawing graphs nicely using simulated annealing. ACM Transactions on Graph, 1996: 15-4, 301-331. 18. Kennedy J and Eberhart R. Particle Swarm Optimization. Proceedings of the IEEE International Conference on Neural Network, 1995: 4, 1942-1948 19. Chu, S. C., Tsai, P. W. and Pan, J. S. Cat swarm optimization. PRICAI 2006: Trends in Artificial Intelligence. Springer Berlin Heidelberg , 2006: 854-858. 20. Karaboga, N., Kalinli, A. and Karaboga, D. Designing digital IIR filters using ant colony optimisation algorithm. Engineering Applications of Artificial Intelligence, 2004: 17-3, 301-309. 21. Boudjelaba, K., Ros, F. and Chikouche, D. An eƥcient hybrid genetic algorithm to design finite impulse response filters. Expert Systems with Applications, 2014: 41-13, 5917-5937. 22. Mandal, S., Ghoshal, S. P., Kar, R. and Mandal, D. Optimal linear phase FIR band pass filter design using craziness based particle swarm optimization algorithm. Journal of Shanghai Jiaotong University (Science), Springer, 2011: 16-6, 696-703. 23. Valle, Y. D., Kumar, G., Mohagheghi, S. and Hernandez, J. C. Particle swarm optimization: basic concepts, variants and applications in power systems, IEEE Transactions on Evolutionary Computation, 2008: 12-2, 171-195. 24. Eberhart, R. C. and Shi, Y. Particle swarm optimization: developments, applications and resources. Evolutionary Computation, 2001. Proceedings of the 2001 Congress, IEEE, 2001: 1, 81-86. 25. Al-Alaoui, M. A. and Baydoun, M. Novel wide band digital differentiators and integrators using diơerent optimization techniques. Signals, Circuits and Systems (ISSCS), 2013 International Symposium, 2013: 1-4. 26. Gupta, M., Relan, B., Yadav, R. and Aggarwal, V. Wideband digital integrators and differentiators designed using particle swarm optimisation, IET Signal Processing, 2014: 8-6, 668-679. 27. Tripathi, P. K., Bandyopadhyay, S. and Pal, S. K. Multi-objective particle swarm optimization with time variant inertia and acceleration coeƥcients. Information Sciences, 2007: 177-22, 5033-5049. 28. Eberhart, R. C. and Shi, Y. Comparing inertia weights and constriction factors in particle swarm optimization. Evolutionary Compu tation, 2000. Proceedings of the 2000 Congress, IEEE, 2000: 1, 84-88. 29. Li, Y. Dynamic particle swarm optimization algorithm for resolution of overlapping chromatograms. Natural Computation, 2009. ICNC’09. Fifth International Conference, IEEE, 2009: 3, 246-250. 30. Clerc, M. and Kennedy, J. The particle swarm-explosion, stability, and convergence in a multidimensional complex space. Evolutionary Computation, IEEE Transactions, 2002: 6-1, 58-73. 31. Rini, D. P., Shamsuddin, S. M. and Yuhaniz, S. S. Particle swarm optimization: technique, system and challenges. International Journal of Computer Applications, 2011: 14-1, 19-26. 32. Kao, Y. T. and Zahara, E. A hybrid genetic algorithm and particle swarm optimization for multimodal functions. Applied Soft Co mputing, 2008: 8-2, 849-857. 33. Gandomi, A. H. Interior search algorithm (ISA): A novel approach for global optimization. ISA transactions, 2014: 53-4, 11681183. 34. Proakis J. G., Manolakis D. G. Digital Signal Processing : Principles, Algorithms and Applications, 3rd edn. Prentice Hall of India; 2005. 35. Vaidyanathan P.P. Design and Implementation of Digital FIR Filters. Handbook of Digital Signal Processing, edited by Douglas F. Elliott, 1987: 55-172.