electronics

Article
Passivity Preserving Model Order Reduction Using
the Reduce Norm Method
Namra Akram 1 , Mehboob Alam 2 , Rashida Hussain 1 , Asghar Ali 1 , Shah Muhammad 3 ,
Rahila Malik 1 and Anwar Ul Haq 4, *
1 Department of Mathematics, Mirpur University of Science and Technology (MUST), Mirpur 10250, Pakistan;
[email protected] (N.A.); [email protected] (R.H.); [email protected] (A.A.);
[email protected] (R.M.)
2 Department of Electrical Engineering, Mirpur University of Science and Technology(MUST), Mirpur 10250,
Pakistan; [email protected]
3 Department of Mathematics, King Saud University, Riyadh 11451, Saudi Arabia; [email protected]
4 Department of Informatics (I-13), Bolzmanstr. 3, Technical University Munich, 85748 Garching, Germany
* Correspondence: [email protected]




Received: 17 April 2020; Accepted: 4 June 2020; Published: 9 June 2020


Abstract: Modeling and design of on-chip interconnect, the interconnection between the components
is becoming the fundamental roadblock in achieving high-speed integrated circuits. The scaling
of interconnect in nanometer regime had shifted the paradime from device-dominated to
interconnect-dominated design methodology. Driven by the expanding complexity of on-chip
interconnects, a passivity preserving model order reduction (MOR) is essential for designing and
estimating the performance for reliable operation of the integrated circuit. In this work, we developed
a new frequency selective reduce norm spectral zero (RNSZ) projection method, which dynamically
selects interpolation points using spectral zeros of the system. The proposed reduce-norm scheme can
guarantee stability and passivity, while creating the reduced models, which are fairly accurate across
selected narrow range of frequencies. The reduced order results indicate preservation of passivity
and greater accuracy than the other model order reduction methods.

Keywords: model order reduction; passivity; spectral-zeros; krylov subspace; integrated circuit;
reduce norm; VLSI; interconnect

1. Introduction
In large-scale complex systems, it is often desirable to use a set of mathematical equations to
represent the behavior of the systems [1,2]. Over the last few decades, the continuous scaling of
on-chip devices and interconnect resulted in exponential increase in the complexity of modeling and
simulation [3,4]. The automated design of on-chip interconnect is important in meeting the increasing
demand of modern high-speed electronics [5,6]. Over the past decade, the rapid advancement in both
device and interconnect design automation have presented a great challenge. The complex and large
design space of on-chip interconnect necessitates efficient modeling techniques [7–9]. Model order
reduction (MOR) is essential to overcome the exceptional computational resource requirements of full
simulations of the integrated circuit (IC) [10]. The motivation is the desire to develop simplified models,
which reduce the simulation time and also capture the main properties of the complex original system.
For large scale on-chip interconnect problems, our work aims at developing efficient reduced order
model (ROM), which will accelerate simulation and provide rapid design optimization of the system.
MOR methods, which are widely used and successful for large-scale linear systems, fall into two
main types, i.e., projection and non-projection based methods [11–13]. Krylov-subspace and singular

Electronics 2020, 9, 964; doi:10.3390/electronics9060964 www.mdpi.com/journal/electronics

Electronics 2020, 9, 964 2 of 21

value decomposition (SVD) based methods are the well-known projection methods. In non-projection
based methods, Hankle-norm approximation is a popular example of MOR method [14]. In developing
ROM, there are multiple requirements to be satisfied, while approximating the original system. The
reduced number of state variables and accurate representation of the original system by correctly
following the terminal behavior are two such important factors [15,16]. In this work, we will only
consider stable and passive interconnect systems, as a result making it desirable that reduced models
preserve the important properties of stability and passivity. It is observed that an original stable but
not passive system, once coupled with other passive components, can make the system unstable [17].
Passivity preserving model order reduction for the on-chip interconnect systems mainly uses two main
methods. The first uses SVD-based balance approximation and the other estimates ROMs via Krylov
subspace methods [18–20]. It is observed here that for large systems, the approximation using the
Krylov subspace projection gives a good estimation of ROM compared to SVD-based methods [21].
Passivity preservation MOR via Krylov subspace method with congruence transformation
includes passive reduced-order interconnect macromodeling algorithm (PRIMA) [19] and its variant
SPRIM [22]. Generally, projection methods use reachability subspaces, constructed using sequence
of vectors. These techniques are basically moment matching of the systems and gives an, iterative
approximation of the eigenvalues. There are various algorithms based on Krylov subspace projection
method [23–25]. Computationally, these methods are less expensive and their, iterative nature makes
them attractive as compared to balanced approximations [26,27]. However, these Krylov based
approximation methods match one moment per iteration and sometime may result in low-accuracy [28].
In addition, unlike balanced approximation methods, the stability is not always assured and the system
approximation is also without an upper error bound. Passivity preserving methods based on balanced
approximations such as PMTBR [29,30], PRTBR [31] and FABT [32,33] are SVD based techniques. These
techniques require dense computations of the order of n2 and n3 and may not be suitable for ROM
of large-scale systems [34]. Another approach of passivity preservation model order reduction uses
spectral zeros (SZ) as positive real interpolation points to produce ROM with roots as its SZs [35,36].
An efficient scheme to implement this techniques was later proposed, which uses certain block matrix
with SZs as its spectrum [37]. The selection criteria of SZs for optimal matching in these schemes is
still not obvious. A selection criteria based on dominant spectral zeros was later suggested, which uses
the magnitude of the associated residues of the system dominant poles [38]. However, the selection of
SZs using dominant pole is computationally expensive.
In this work, we focus on solving complex problems in the design of modern processors, which
operate on dual frequencies to account for its low power as well as turbo mode operations. Note
that the proposed scheme is for high-speed ICs, where chip operates at GHz frequency range with
no more than 1–3 GHz difference between their dual mode frequencies. In these high-speed ICs,
the scaling of features significantly influence the performance [39], especially with increasing processor
frequencies. This results in design automation focused research efforts in the MOR [19,40,41]. In this
paper, we propose a frequency selective reduce norm spectral zero (RNSZ) projection method, which
uses a simple reduce-norm criteria to select interpolation points using SZs of the system. In the
proposed method, the frequency of interest is divided into regular intervals and interpolation points
are the selected SZ, which provides the best estimation in the minimum-norm sense. The result
is the preservation of passivity and stability properties in the ROM. It is important to note that
selected interpolation points are part of the stable SZ and therefore guarantee passivity of the reduced
system. The simulation results of ROM using proposed method show more accurate approximations
with less absolute error for the selected narrow band of frequencies compared to other model order
reduction methods.
In next section, we discuss MOR and important passivity preserving reduction techniques.
The proposed reduced-norm frequency selective technique is explained in Section 3. Section 4 discusses
the simulation and analysis of the results and the paper is finally concluded in Section 5.
Electronics 2020, 9, 964 3 of 21

2. Model Order Reduction of On-Chip Interconnects

Recent advances in the IC fabrication ultimately lead to the transformation of the computer-aided
design (CAD) industry [10]. The scaling of technology over a period of time has shifted the design
paradigm to interconnect dominated methodology. In the CAD tools, MOR reduces the design
complexity of the numerical simulation [42]. In a RLC (resistor, inductor, capacitor) network model
of an on-chip interconnect, preserving of the passivity inherent in the system is essential [10]. It is
important to note that all passive systems are stable but on the contrary all stable systems are not
guaranteed to be passive [17]. This section discusses the preliminaries of model order reduction using
projection with special emphasis to passivity preserving techniques for on-chip interconnects.

2.1. State Description and Mor Using Projection

In the state space model of the on-chip interconnect, the original system of order n in
descriptor-form is given as

Eẋ = Ax(t) + Bu(t),

y(t) = Cx(t) + Du(t), (1)

where x(t) ∈ Rn represents the internal states of the system, u(t)∈ Rm and y(t)∈ R p are the system
input and output respectively. Correspondingly, E, A ∈ Rn×n , (A, E) are Generalized EigenValues
(GEV), B ∈ Rn×m , C ∈ R p×n , D ∈ R p×m are the matrices defining linear maps between inputs, output
and internal states. It is important to note that the extracted model in the state space form shown in (1)
accounts for all the resistance, inductance and capacitance effects of the on-chip interconnect. The block
matrices of resistance (R), inductance (L) and capacitance (C) will appear in the state space as
" # " # " #
C 0 R
ẋ (t) = x (t) + u ( t ),
0 L 0
| {z } | {z } | {z }
E A B
h i h i
y(t) = x (t) + u ( t ),
| {z } |{z}
B T =C D

ˆ (Ê, Â, B̂, Ĉ, D) for (1) of order k is given as

The associated ROM ∑

Êx̂˙ = Âx̂(t) + B̂u(t),

ŷ(t) = Ĉx̂(t) + Du(t), (2)

where x̂ ∈ Rk , Ê ∈ Rk×k , Â ∈ Rk×k , B̂ ∈ Rk×m , Ĉ ∈ R p×k , D ∈ R p×m with k << n. As noted earlier,
the design automation tools require that the reduced models should preserve important properties of
stability and passivity of the original system. The theory of stability plays an important role in the
MOR. In dynamical system, the concept is explained in terms of internal and external stability of the
system. Similarly, for the passivity, a passive ROM does not generate any energy. Passivity in ROM is
sometime ensured by positive real transfer function [15].
In projection based MOR methods consisting of the projection matrices V and W, the ROM is
obtained such that V, W ∈ Rn×k and WT V = Ik . A careful selection of V and W by properly deciding
projection subspace ensures that ROM not only preserves the stability and passivity, but also retains the
dynamical characteristics of the original response. The reduced system matrices using these projectors
are represented as:

Ê = WT EV, Â = WT AV, B̂ = WT B, Ĉ = CV. (3)

Electronics 2020, 9, 964 4 of 21

The accuracy of ROM is completely determined by the choices of the projection matrices V and
W. These projection martices control the way system properties are preserved by the reduced system
and also, identifies the accuracy of the ROM. It is important to note that in the applications of circuit
simulation, both invertible and non-invertible matrix E arises in the system simulation. A singular
matrix E give rise to new set of challenges as some of the system poles and SZs lies at infinity. In MOR,
invertible and non-invertible E are handled differently [43,44]. However in this work, E is an invertible
non-singular matrice, as it appears in the state representation of on-chip interconnects.

2.2. Passivity Preserving Mor

Passivity of the LTI system is preserved if the transfer function G(s) is positive real [15]. This
means G(s) satisfies the following conditions

• G(s) is analytic for Real(s) > 0;

• G(s̄) = G(s) for all s ∈ C;
• G(s) + G(−s)T ≥ 0 for Real(s) > 0.

Therefore, for real systems, we always have G(s̄) = G(s) and the last condition shows the presence
of stable matrix function V(s) (inverse exists) such that G(s) + GT (−s) = V(s)VT (−s) is the spectral
factorization of G, where V is the spectral factor of G. The SZs of G are obtained from the zeros of V,
i.e., Λi , for i = 1, · · · , n, such that det V(Λi ) = 0.
This section defined the passivity preserving in the MOR and existence of stable rational function,
which helps in obtaining SZs of the system. In the next section, we have proposed a scheme, which
selects a subset of these SZs as interpolation points to give best estimation of system approximation
in the minimum-norm sense. The result is a ROM, which preserves the passivity and stability of
the system.

3. Passivity Preserving MOR Using RNSZ

This section defines the procedure of MOR using reduce-norm spectral zero (RNSZ) method. It is
a frequency selective reduction method, which uses subset of SZ as interpolation point using simple
reduce-norm criteria to construct projection matrices. The result is an optimally reduced system, which
also preserves the essential properties of the original system.

3.1. Passivity Preserving Projection

The RNSZ method establishes relationship of the selected SZ with the system spectral zeros.
However, it is desirable first to discuss some basic parameters of the system and its properties, which
are used in the derivation. Given the descriptor system (1), the Laplace transform of the transfer
function ∑ is given as

G(s) = C(sE − A)−1 B + D. (4)

Note that the poles of the system ∑ are the GEVs of the pair (A, E). Let the matrix (A − ΛE)
be non-singular throughout for some Λ ∈ C [45], which is represented as a matrix pencil (A, E) in
mathematical terms. Then the system ∑ is said to be stable if and only if all finite poles are lying in the
left half of the s plane, i.e., Re(Λ) < 0 with |Λ| 6= ∞. In this connection, it turns out that the SZs of ∑
are all s ∈ C, such that det [G(s) + GT (−s)] = 0, where

GT (−s) = BT (−sET − AT )−1 CT + DT . (5)

This definition shows that each SZ has the reflection in the s plane. Note that the Real SZs si are in
pairs (si ,−si ), whereas complex exist in quadruples (si , siT ,−si , −siT ). Alternately, these definitions also
explain the positive realness of the transfer function and if SZs are used to drive projection matrices,
the system preserves the passivity of the ROM.
Electronics 2020, 9, 964 5 of 21

Essentially, this is the motivation of our proposed frequency selective passivity preserving RNSZ
method. The method divides the frequency of interest in to regular intervals and selects the SZs, which
minimizes the norm of the distance. The selected SZs become the interpolation point for constructing
projection matrices (V and W), which simultaneously assure stability and passivity of the system,
while creating the frequency-selective ROM. In the proposed RNSZ method, we first constructed the
following matrix Ḡ and Ē from the state representation of system given by (1).
 
A B
Ḡ :=  −AT −CT  , (6)
 
C BT D + DT

and
 
I
Ē :=  I . (7)
 
0

These matrices are used for extracting the finite eigenvalues of a GEV problem. In this
formulation, the finite SZs (subset of eigenvalues) of Ḡ are all complex numbers Λ, such that Rank
(Ḡ − ΛĒ) < 2n + p, i.e., it is equivalent to solving the finite GEVs Λ(Ḡ,Ē). The pair, therefore, provides
the desired passivity preserving projection vectors V and W to maintain the passivity of the reduced
model. Note that solving the generalized eigenvalue problem generates eigenvalues and eigenvectors
necessary for finding the SZs.

3.2. Spectral Zero Selection and Iterative Algorithm for RNSZ

The preceding section explained the passivity preserving projection followed by the important
task of constructing matrices V and W. The projection matrices are determine by the selected SZs,
which are the subset of the n stable SZs of the original system. It is observed here that the proposed
method for the selection of SZs is associated to the minimum norm value of the SZs. The selected
SZs characterize the system’s output and their use as interpolation points determine the accuracy
of the ROM system. The pseudocode giving details of the proposed RNSZ method in selecting the
spectral zeros is given in Algorithm 1. Using examples from [37], the partial fraction expansion of
G(s) can be written as rational function G(s) + GT (−s) and is used to compute the stable spectral
zeros. In the first step, [Q, R] factorization are taken into account for calculating the eigenvalues and
eigenvector matrices of the system. Note that SZs of ∑ are the GEVs of the following structured
eigenvalue problem.
( Ḡ − R Ē) Q = 0, (8)

where Ḡ and Ē are given by (6) and (7) respectively. In Equation (8), Q and R represent the matrices
of the eigenvectors and eigenvalues, whereas SZs are the diagonal elements of R. Note that the
formulation of (Ḡ, Ē) pair, ensures that the eigenvalues, which are the SZs of the system, lay on the
left half plane. In our case D + DT is singular, which is equivalent to solving a generalized eigenvalue
problem for the computation of SZs. Therefore, the GEVs of Λ(Ḡ, Ē) are the SZs of the original system.
It is important to note that the RNSZ method is a frequency selective approximation of large-scale
dynamical system. Considering SZs of ∑ be the poles of G(s), we suggest a criterion to choose
SZs on the basis of dominance of each SZs in the frequency of interest. Therefore, the proposed
method introduces a reduce-norm criteria for the selection of SZs of the system transfer function G(s).
Motivation is to solve complex problem of model order reduction of on-chip interconnects of a modern
processor. We take the example of a Intel Core i7-2920XM processor (Intel, Santa Clara, CA, USA),
which operates on dual frequencies ( f 1 and f 2 ) to account for its low power as well as turbo mode
Electronics 2020, 9, 964 6 of 21

operations [46]. The frequency selection is performed by uniformly distributing the frequency of
interest. The kth frequency interval between the low and high frequency can then be defined as

f2 − f1
 
fk = f1 + k , (9)
N

where N is the order of the ROM, with f 1 and f 2 defining selected frequency range with f 2 > f 1 .
Therefore, in the total n stable SZs, the kth selected SZ is chosen based on following reduce norm criteria

SZk = | || f k || − ||imag(SZ )|| |min . (10)

Note that the selected interpolation points are SZs, which reduce the minimum norm distance
between the selected frequency interval and the distributed SZs in the complex spectral domain. In this
reduction, the N selected SZs form the basis of developing ROM, with corresponding eigenvectors of
the selected SZ given by [Qx , S, Qy ] = svd(XT Y). The eigenvectors of the selected SZs are then used in
constructing the projection vectors V and W of the system as

V = XQx S−1 , and W = YQy S−1 . (11)

Note that these projection matrices not only dynamically generate ROM but also satisfy the
orthogonality condition. This is a generalization of the approximation by projection using our proposed
RNSZ method. We can now directly state the fundamental results of RNSZ by the following theorem.

Algorithm 1 Pseudocode of RNSZ Passivity Preserving MOR Algorithm

INPUT = (A, B, C, D, E),

OUTPUT = (Â,B̂, Ĉ,D̂,Ê),
Construct Ḡ and Ē from (A, B, C, D, E)
Perform [Q,R] = qr(Ḡ,Ē); QR factorization of R
SZ = diag(R); Spectral Zeros of the system
Define f1 and f2; Limits of minimum and maximum frequency
N: Order of reduced system
Divide [f2 to f1] into N interval
Select SZ
for k = 1,2,...,N
SZk = | || f k || − ||imag(SZ )|| |min
end
Generate S and Q using selected SZs (i.e., SZk ) and corresponding eigenvectors,
[Qx , S, Qy ] = svd(XT Y)
Construct V, W: Projection Matrices
V = XQx S−1
W = YQy S−1
Generate Reduce order Model
Â= WT AV; B̂ = WT B; Ĉ = CV; D̂ = D; Ê = WT EV.

Theorem 1. Given a descriptor system defined by (1), the transfer function of ∑ (A,B,C,D,E) is
G(s) = C(sE−A)−1 B + D is positive real. The system SZs are the required interpolation points
 using
 criteria
f2 − f1
SZk = | || f k || − ||imag(SZ )|| |min for k = 1,2,3,...,N in the frequency interval f k = f 1 + k N , where N

is the order of ROM. The reduced norm system ∑ ˆ (A,B,C,D,E) is defined by  = WT AV, B̂ = WT B, Ĉ = CV,
D̂ = D, Ê = W T EV and Ĝ(s) = Ĉ(sÊ − Â)−1 B̂ + D̂ are both passive and stable.
Electronics 2020, 9, 964 7 of 21

Next, we explain the implementation of RNSZ method by using a basic example of RLC ladder
network.
Example 3.1: Consider Figure 1 RLC Network [37]. Let us define x1 , x3 and x5 be the voltage
across C1 , C2 and C3 respectively, whereas the current across L1 and L2 are represented by x2 and x3
respectively. The input voltage u and output current y are also defined in (1). The capacitors and
inductors have unity value, while R1 = 12 and R2 = 15 . The rest of the example using RNSZ method is
solved as follows.

Figure 1. RLC ladder circuit consisting of resistors, inductors and capacitors forming a 5th order system.

The transfer function G(s) of the RLC circuit given by Figure 1 can be written as

s5 + 3s4 + 6s3 + 9s2 + 7s + 3

G (s) = . (12)
s5 + 7s4 + 14s3 + 21s2 + 23s + 7

The partial fraction expansion in the form of rational function is given as

2(s10 − s8 − 12s6 + 5s4 + 35s2 − 21)

G (s) + G T (−s) = . (13)
(s5 + 7s4 + 14s3 + 21s2 + 23s + 7)(s5 − 7s4 + 14s3 − 21s2 + 23s − 7)

Minimal realization of the state matrices of G (s) (12) is given as

   
−2 1 0 0 0 0
 −1 0 1 0 0  0
  

A =  0 −1 0 1 0  , B = 0 ,
   
0 −1 0
   
 0 1  0
0 0 −0 −1 −5 2
 
1 0 0 0 0
0 1 0 0 0
h i  
C= 0 0 0 0 −2 , E = 0 0 1 0 0 , D = 1.
 
 
0 0 0 1 0
0 0 0 0 1
Using reduce norm selection criteria SZk = | || f n (k)|| − ||imag(SZ )|| |min , we have selected three
spectral zeros, i.e., [0.1833 + 1.5430i, 0.1833 − 1.5430i, −0.1833 + 1.5430i] to construct projection
matrices V and W using (11). Using these projectors, we can generate following ROM.
   
2.8629 −3.9654 4.0350 0.9925
An = −1.0999 2.2405 −3.5414 , Bn = −0.6084 ,
   
0.2158 −0.2190 1.5778 0.1270
h i
Cn = 2.3915 −2.3852 2.3852 ,
Electronics 2020, 9, 964 8 of 21

 
1 0 0
En = 0 1 0 , Dn = 1.
 
0 0 1

4. Simulation and Analysis

This section discuses the accuracy and efficiency of our proposed RNSZ method. We take examples
of on-chip interconnect structures including interconnect bus and a RLC network. The models are
created using the field solvers [47,48] for capacitance and inductance extraction. A combination of
modified nodal analysis is then used to create state space representation as described in (1). For
the purpose of checking the accuracy of our proposed RNSZ method, we generated examples of an
order 79 and 1079 RLC network and interconnect bus respectively. In addition, we also take a look
at series of examples of on-chip spiral inductor with transfer function ranging from few kHz to GHz.
The examples covers the frequency ranges of 3 kHz to 3 GHz, 1 MHz to 3 GHz, 10 MHz to 3 GHz and
100 MHz to 3 GHz to discuss some limitations of the proposed RNSZ method. The results discussed at
the end of this section highlight that RNSZ is suitable for application specific solution and its use may
be restricted only for narrow-band high-speed ICs applications.

20

-20
Magnitude (dB)

-40

-60

-80

-100
108 109 1010

Frequency (Hz)

Figure 2. Original system frequency response of a on-chip 79 order RLC network.

Firstly, we took the example of a low order 79 RLC network and compared the reduce order
modeling result of PRIMA [19] and interpolation via SZ [37] with RNSZ method. The original system
of RLC network is plotted in Figure 2. The comparison results in Figures 3 and 4 (projection showing
selected frequency of interest from 2.5 GHz to 3.5 GHz) shows a close match of RNSZ, with an
overlapping peaks and response compared to the original system. However, a significant deviation is
observed in case of SZ method using interpolation at selected spectral zeros and PRIMA. A comparison
of the absolute error in dB is plotted in Figure 5. The error plot shows that our proposed method of
RNSZ method has considerably reduce error relative to the PRIMA and interpolation via SZ.
Electronics 2020, 9, 964 9 of 21

20

RNSZ
Original
0 Interested area

-20
Magnitude (dB)

-40

-60

SZ
-80
PRIMA

-100
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
Frequency (GHz)

Figure 3. Frequency response of a on-chip RLC network with original system order of n = 79. The
reduce models of using reduce norm spectral zero (RNSZ), PRIMA and spectral zeros (SZ) methods are
of order N = 10. The reduce model generated by RNSZ closely matches the fine details of the original
system compared to other techniques.

-10 SZ

-20
Original

-30
Magnitude (dB)

-40

-50

-60

-70 PRIMA

RNSZ
-80

-90
2.5 2.6 2.7 2.8 2.9 3 3.1 3.2 3.3 3.4 3.5

Frequency (GHz)

Figure 4. Magnitude response of the frequency of interest, i.e., operating frequency range of i7-2920XM
from 2.5 GHz to 3.5 GHz. The proposed RNSZ method closely matches with the original system.
Electronics 2020, 9, 964 10 of 21

In order to compare complexity for large span of frequency, we also plotted the same RLC
network example by increasing the span from 2 GHz to 3.5 GHz in Figure 6. In this case, note that
the number of states required by RNSZ method doubled from N = 10 to 20. It is observed that the
complexity of the amplitude response and the desired minimum error will determine the order of
ROM, which considerably varies from example to example. The presence of resistor, inductor and
capacitor in on-chip interconnect makes the response oscillatory and unique to each examples. The
number of oscillation peaks and bottoms in a particular span of frequency is directly proportional to
the complexity of the ROM of system. A higher order example of spiral inductor covering a large span
of frequency, i.e., from 1.5 GHz to 3.5 GHz is discussed at the end of this section.
The second example is a state space model of an interconnect bus. The initial system response
of the bus is plotted in Figure 7. The example is taken, keeping in mind the state of art the processor.
The extraction tools creates this example keeping in mind Intel Core i7-2920XM modern processor [46].
Note that i7-2920XM works on two frequencies, i.e., 2.5 GHz and 3.5 GHz, which actually defines
the frequency of interest in this simulation. Figure 8 displays the magnitude response of original n =
1076 order interconnect bus, with reduced order N = 100 response using PRIMA [19], SZ [37] and
proposed RNSZ method.

50

SZ

PRIMA
0
Absolute Error (dB)

-50

-100

RNSZ

-150
2.5 2.6 2.7 2.8 2.9 3 3.1 3.2 3.3 3.4 3.5
Frequency (GHz)

Figure 5. The absolute error plot shows minimum error for the RNSZ method. The calculated mean
absolute error for the RLC network in the frequency of interest (2.5 to 3.5 GHz) is −71.64 dB, −35.89
dB and −44.85 dB for RNSZ, SZ and PRIMA methods respectively.
Electronics 2020, 9, 964 11 of 21

10

0
SZ
-10

-20

-30
Magnitude (dB)

-40

-50

-60

-70 PRIMA
Original

-80 RNSZ

-90

-100
2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6
Frequency (GHz)

Figure 6. Magnitude response of RLC network for a large span of frequency (2 GHz to 3.5 GHz).
The complexity of the reduce order model increased from N = 10 to 20.

-25

-30
Magnitude (dB)

-35

-40

-45
108 109 1010
Frequency (Hz)

Figure 7. Original system frequency response generated by a interconnect bus of order n = 1076.
Electronics 2020, 9, 964 12 of 21

-25

Interested area
Original PRIMA

-30
Magnitude (dB)

-35

-40

SZ
-45 RNSZ

-50
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
Frequency (GHz)

Figure 8. System frequency response of a interconnect bus with original system order n = 1076. The
reduced order model (ROM) of order N = 100 for the other methods shows the response using RNSZ
completely matches the original system.

The result of the comparison in Figures 8 and 9 shows that a reduce order 100 using proposed
RNSZ method correctly captures the behaviour of the system. However, the PRIMA implementation
misses most of the resonance peaks and does not accurately characterize the original system response. It
is observed that the Arnoldi based PRIMA matches the low frequency response first and a reduced order
100 is not sufficient to accurately cover the selected frequency range. The error plot shown in Figure 10
highlights the strength of our proposed method (RNSZ) showing the minimum error over the entire
range of frequency, i.e., from 2.5 GHz to 3.5 GHz. Note that the reduce error is due to the dynamical
selection of SZs based on reduce-norm criterion proposed in Section 3. The selection of SZ methods
based on interpolation at the selected spectral zeros [37] performs better than PRIMA. However,
Figure 10 shows that the calculated mean absolute error for the interconnect bus in the frequency of
interest is −180.52 dB, −112.93 dB and −42.61 dB for the RNSZ, SZ and PRIMA methods respectively.
In order to show that the proposed RNSZ also works well with large span of frequency, we
selected an on-chip spiral inductor example. The original system of order n = 1008 along with ROM
of order N = 50 produced by RNSZ and PRIMA methods are plotted in Figure 11. The results
clearly show a close match of RNSZ method. However, the PRIMA method show deviations of the
magnitude plot compared to original system and missed out most of the peaks of the system transfer
function. The simulation example clearly show validity of the proposed RNSZ method for large span
of frequency. Note that the order and hence the number of states required to represent the ROM will
increase with large frequency of interest and a more complex example.
Electronics 2020, 9, 964 13 of 21

-25

RNSZ
-30

Original
Magnitude (dB)

-35

-40

SZ

-45
PRIMA

-50
2.5 2.6 2.7 2.8 2.9 3 3.1 3.2 3.3 3.4 3.5
Frequency (GHz)

Figure 9. Magnitude response for the frequency of interest (2.5 GHz to 3.5 GHz) for an Intel Core
i7-2920XM processor interconnect bus. The proposed method (RNSZ) closely matches the fine details
of the original on-chip interconnect bus.

-50

-100
Absolute Error (dB)

PRIMA

-150

-200 SZ

RNSZ
-250

-300
2.5 2.6 2.7 2.8 2.9 3 3.1 3.2 3.3 3.4 3.5
Frequency (GHz)

Figure 10. The absolute error plot shows minimum error for the RNSZ method. The calculated mean
absolute error for the interconnect bus example in the frequency of interest is −180.52 dB, −112.93 dB
and −42.61 dB for RNSZ, interpolation by SZ and PRIMA methods respectively.
Electronics 2020, 9, 964 14 of 21

20

10

0 Original
PRIMA

-10
Magnitude (dB)

-20

-30

-40

-50
RNSZ

-60

-70

-80
1.5 2 2.5 3 3.5
Frequency (GHz)

Figure 11. System frequency response of a spiral inductor with original system order of n = 1008.
The selected large span of frequency now cover over 2 GHz of frequency range, i.e., from 1.5 GHz
to 3.5 GHz. A minimum ROM of order N = 50 is required by RNSZ method to match the response.
However, with similar ROM (N = 50), the PRIMA method missed out most of the response and peaks
of the original system.

A comparison of the mean absolute error of the ROM generated by the proposed RNSZ, PRIMA
and SZ methods are tabulated in Table 1. The calculated mean absolute errors for the RLC network
and Interconnect bus shows that the proposed RNSZ correctly captures the behaviour of the system
with almost negligible error.

Table 1. Mean absolute error comparison for various examples.

Examples PRIMA [19] SZ [37] Proposed RNSZ

RLC network −44.85 dB −35.89 dB −71.64 dB
Interconnect bus −42.61 dB −112.93 dB −180.52 dB
Spiral inductor −37.97 dB - −177.28 dB

In the end, we take series of examples generated from an on-chip spiral inductor with transfer
function spanning from few kHz to GHz range as shown in Figure 12. In first example, we take the
frequency range from 100 MHz to 3 GHz and plot the original and ROM function (order n = 38) in
Figure 13. In this simulation, we selected lower and upper range of frequencies to be 100 MHz and
3 GHz respectively. The absolute error plot in Figure 14 shows a close match with a mean absolute error
of −129.98 dB in the selected range of frequencies. The second example is the same spiral inductor,
however this time, we selected the lower and upper range of frequencies to be 10 MHz and 3 GHz
respectively. The simulation results are plotted in Figure 15, which shows a close match with ROM of
order n = 50 computed using RNSZ method. The absolute error is shown in Figure 16 with a mean
absolute error of −100.97 dB in the selected range of frequencies. It is important to note that these two
examples cover narrow band of frequencies and their simulation results shows overlapping peaks
with their mean absolute error less than −100 dB.
In order to gain good inside of the problem related with large band of frequencies, the next two
examples are spread over a large span of frequencies, i.e., 1 MHz to 3 GHz and 3 kHz to 3 GHz. In third
Electronics 2020, 9, 964 15 of 21

example we selected lower and upper range of frequencies to be 1 MHz and 3 GHz respectively.
The simulation results are plotted in Figure 17 with a reduce order of n = 180 matching the original
frequency response. It is worth mentioning here that ROM order in this case is quite high, almost
comparable to the order of the original system. Therefore, the example highlights that the proposed
RNSZ techniques is not suitable for large span of frequencies. The absolute error plot in Figure 18
shows a good match with a mean absolute error of −73.98 dB in the selected range of frequencies.
The fourth and last example is for a frequency range of 3 kHz to 3 GHz. The simulation results are
plotted in Figure 19 and again a very high reduce order of n = 186 is required to match the original
frequency response. The absolute error plot in Figure 18 shows a good match with a mean absolute
error of −76.29 dB in the selected range of frequencies.

20

-20
Magnitude (dB)

-40 100 MHz to 3 GHz

10 MHz to 3 GHz
-60

1 MHz to 3 GHz
-80

3 kHz to 3 GHz
-100

104 105 106 107 108 109 1010

Frequency (Hz)

Figure 12. System frequency response of a spiral inductor with original system order of n = 191. The
selected spans of frequency now cover different frequency ranges from 3 kHz to 3 GHz.

20

0 Original
RNSZ

-20
Magnitude (dB)

-40
100 MHz to 3 GHz

-60

-80

-100
104 105 106 107 108 109 1010
Frequency (Hz)

Figure 13. Original system frequency response of a spiral inductor (order n = 191) compared with a
ROM generated using RNSZ of order N = 38 covering 100 MHz to 3 GHz frequencies.
Electronics 2020, 9, 964 16 of 21

50

-50
Absoulute Error (dB)

-100

-150

-200

-250

-300 100 MHz to 3 GHz

-350
104 105 106 107 108 109 1010
Frequency (Hz)

Figure 14. Absolute error of ROM generated using RNSZ of order N = 38 compared to original system
frequency response of a spiral inductor (order n = 191). The mean absolute error is −129.98 dB from
100 MHz to 3 GHz frequencies.

20

Original
0 RNSZ

-20
Magnitude (dB)

-40

-60
10 MHz to 3 GHz

-80

-100
104 105 106 107 108 109 1010
Frequency (Hz)

Figure 15. Original system frequency response of a spiral inductor (order n = 191) compared with a
ROM generated using RNSZ of order N = 50 covering 10 MHz to 3 GHz frequencies.

20

-20
10 MHz to 3 GHz
-40
Absolute Error (dB)

-60

-80

-100

-120

-140

-160

-180

-200
104 105 106 107 108 109 1010
Frequency (Hz)

Figure 16. Absolute error of ROM generated using RNSZ of order N = 50 compared to original system
frequency response of a spiral inductor (order n = 191). The mean absolute error is −100.97 dB from
10 MHz to 3 GHz frequencies.
Electronics 2020, 9, 964 17 of 21

20

0 Original
RNSZ

-20
Magnitude (dB)

-40

-60

1 MHz to 3 GHz
-80

-100
104 105 106 107 108 109 1010
Frequency (Hz)

Figure 17. Original system frequency response of a spiral inductor (order n = 191) compared with a
ROM generated using a high order (N = 180) RNSZ method covering 1 MHz to 3 GHz frequencies.

-20

-40
Absolute Error (dB)

-60

-80

-100

-120 1 MHz to 3 GHz

-140
104 105 106 107 108 109 1010
Frequency (Hz)

Figure 18. Absolute error of ROM generated using RNSZ of order N = 180 compared to original system
frequency response of a spiral inductor (order n = 191). The mean absolute error is −73.98 dB from 1
MHz to 3 GHz frequencies.

20
Original
RNSZ
0

-20
Magnitude (dB)

-40

-60

-80 3 kHz to 3 GHz

-100
104 105 106 107 108 109 1010
Frequency (Hz)

Figure 19. Original system frequency response of a spiral inductor (order n = 191) compared with a
ROM generated using a high order (N = 186) RNSZ method covering 3 kHz to 3 GHz frequencies.
Electronics 2020, 9, 964 18 of 21

In the end, Table 2 summarizes the results to compare accuracy of RNSZ simulation over large
band of frequencies. The on-chip spiral inductor example, with transfer function plotted in Figure 12,
spans frequencies from few kHz to GHz. Its simulation results are plotted from Figures 13–20 and
are included in Table 2. The table compares ROM accuracy and complexity for different range
of frequencies.

20

-20
Absolute Error (dB)

-40

-60

-80

-100

-120 3 kHz to 3 GHz

-140
104 105 106 107 108 109 1010
Frequency (Hz)

Figure 20. Absolute error of ROM generated using RNSZ of order N = 186 compared to original system
frequency response of a spiral inductor (order n = 191). The mean absolute error is −76.29 dB from
3 kHz to 3 GHz frequencies.

Table 2. Comparison of spiral inductor example (Figure 12 with original system order n = 191).

Frequency Range Reduce Order Mean Absolute Error Remarks

3 kHz–3 GHz 186 −76.29 dB Matches original function with high order ROM
1 MHz–3 GHz 180 −73.98 dB Matches original function with high order ROM
10 MHz–3 GHz 50 −100.97 dB Closely matches original function
100 MHz–3 GHz 38 −129.98 dB Closely matches original function

It is important to note that for large span of frequencies (row 1 and 2 of Table 2), the order of the
ROMs are quite high, which signifies no real advantage gain by model order reduction of the original
system. However, for small span of frequencies (row 3 and 4 of Table 2), the results of RNSZ method
gives greater accuracy with fairly small ROMs in addition to guaranteeing stability and passivity of
systems. The results discussed in this section and Table 2 clearly highlight the fact that RNSZ is very
efficient for application specific solutions, i.e., for narrow band of frequency with span ranging in few
GHz and its use may be restricted only for high-speed applications.

5. Conclusions
Over the last few decades, scaling in deep submicron had shifted the paradime from
device-dominated to interconnect-dominated design methodology. In this work, we developed a new
reduce-norm frequency selective projection method by using interpolation point based on SZs of the
system. The proposed reduce-norm scheme dynamically selects spectral zeros, which can guarantee
stability and passivity, while creating the ROMs accurate across selected narrow range of frequencies
for application specific MOR.

Author Contributions: Conceptualization, M.A.; Formal analysis, N.A., M.A. and R.M.; Funding acquisition,
R.H., A.A., S.M., N.A., R.M. and A.U.H.; Investigation, N.A. and M.A.; Methodology, N.A. and M.A.; Project
administration, A.U.H.; Software, N.A., and M.A.; Supervision, M.A.; Validation, N.A., M.A. and R.M.;
Electronics 2020, 9, 964 19 of 21

Writing—original draft, M.A. and N.A.; Writing—review & editing, M.A., N.A., R.H., A.A. and S.M. All authors
have read and agreed to the published version of the manuscript.
Funding: The authors extend their appreciation to the Technical University of Munich (TUM), Mirpur University
of Science and Technology (MUST), Mirpur 10250, Pakistan and Deanship of Scientific Research at King Saud
University for funding this work through research group no.RG-1441-351.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations
The following abbreviations are used in this manuscript:

R The set of real numbers

C The set of complex numbers, i.e., s-plane
Λ Eigenvalues
SZ Spectral Zeros
MOR Model Order System
ROM Reduced Order Model
RNSZ Reduced Norm Spectral Zero
LTIs Linear Time Invariant system

References
1. Ciuprina, G.; Ioan, D. Scientifc Computing in Electrical Engineering; SCEE ser. Mathematics in Industry;
Springer: Berlin, Germany, 2007.
2. Wan, F.Y. Mathematical Models and Their Analysis; SIAM: Philadelphia, PA, USA, 2018; p. 79.
3. Kundu, S.; Chattopadhyay, S. Network-On-Chip: The Next Generation of System-on-Chip Integration; CRC Press:
Boca Raton, FL, USA, 2018.
4. Gugulothu, B.; Naik, B.R.; Boodidha, S. Modeling of Capacitive Coupled Interconnects for Crosstalk Analysis
in High Speed VLSI Circuits. In Proceedings of the 2019 International Conference on Communication and
Signal Processing (ICCSP), Chennai, India, 4–6 April 2019.
5. Mitra, P.; Sharma, B.; Chandna, V.K.; Rathore, V.S. Design and Performance Evaluation of Hybrid
Wired-Wireless Network on Chip Interconnect Architectures. In Third International Congress on Information
and Communication Technology; Springer: Singapore, 2019; pp. 191–199.
6. Ayala, J.L. (Ed.) Communication Architectures for Systems-On-Chip; CRC Press: Boca Raton, FL, USA, 2018.
7. Alam, M.; Nieuwoudt, A.; Massoud, Y. Wavelet-Based Interpolation Point Selection for Multi-Shifted Arnoldi.
In Proceedings of the 2007 IEEE International Symposium on Circuits and Systems, New Orleans, LA, USA,
27–30 May 2007; pp. 653–656.
8. Silveira, L.M.; Kamon, M.; White, J. Efficient reduced-order modeling of frequency-dependent coupling
inductances associated with 3-D interconnect structures. IEEE Trans. Components Packag. Manuf. Technol. Part
B 1996, 19, 283–288.
9. Alam, M.; Nieuwoudt, A.; Massoud, Y. Wavelet-based passivity preserving model order reduction for
wideband interconnect characterization. In Proceedings of the 8th International Symposium on Quality
Electronic Design (ISQED’07), San Jose, CA, USA, 26–28 March 2007; pp. 432–437.
10. Wang, X.; Yu, M.; Wang, C. Structure-Preserving-Based Model-Order Reduction of Parameterized
Interconnect Systems. Circuits, Syst. Signal Process. 2018, 37, 19–48.
11. Gugercin, S. Projection Methods for Model Reduction of Large-Scale Dynamical Systems. Ph.D. Thesis, Rice
University, Houston, TX, USA, 2003.
12. Wang, X.L.; Jiang, Y.L. Two-sided projection methods for model reduction of MIMO bilinear systems. Math.
Comput. Model. Dyn. Syst. 2013, 19, 575–592.
13. Alam, M. Compact Models for nanophotonic Structures and On-Chip Interconnects. Ph.D. Thesis, Rice
University, Houston, TX, USA, 2007.
14. Chaudhary, S.K.; Kumar, A. Hankel norm approximation of a stable non-minimal system. In Proceedings of
the 1st International Conference on Electronics, Materials Engineering and Nano-Technology, Science City,
Kolkata, 28–29 April 2017; pp. 1–4.
15. Antoulas, A.C. Approximation of Large-Scale Dynamical Systems; SIAM: Philadelphia, PA, USA, 2005.
Electronics 2020, 9, 964 20 of 21

16. Alam M.; Nieuwoudt A.; Massoud Y. Reduced-Order Wide-Band Interconnect Model Realization using
Filter-Based Spline Interpolation. In Proceedings of the 2007 Asia and South Pacific Design Automation
Conference, Yokohama, Japan, 23–26 January 2007; pp. 373–378.
17. Zulfiqar, U.; Tariq, W.; Li, L.; Liaquat M. A Passivity-Preserving Frequency-Weighted Model Order Reduction
Technique. IEEE Trans. Circuits Syst. II 2017, 64, 1327–1331.
18. Sheldon, L.H.; Tan, X.D. Advanced Model Order Reduction Techniques in VLSI Design; Cambridge University
Press: Cambridge, UK, 2007.
19. Odabasioglu, A.; Celik, M.; Pileggi, L.T. PRIMA: Passive Reduced-Order InterconnectMacromodeling 194
Algorithm. IEEE Trans. CAD 1998, 17, 645–654.
20. Alam, M.; Nieuwoudt, A.; Massoud, Y. Model order reduction using spline-based dynamic multi-point
rational interpolation for passive circuits. Analog. Integr. Circuits Signal Process. 2007, 50, 273–277.
21. Freund, R.W. Krylov-subspace methods for reduced-order modeling in circuit simulation. J. Comput. Appl.
Math. 2000, 123, 395–421.
22. Freund, R. Sprim: Structure-preserving reduced-order interconnect macromodeling. In Proceedings of the
IEEE/ACM International Conference on Computer Aided Design, San Jose, CA, USA, 7–11 November 2004;
pp. 80–87.
23. Alam, M.; Nieuwoudt, A.; Massoud, Y. Efficient multi-shifted Arnoldi projection using wavelet transform.
J. Circuits, Syst. Comput. 2007, 16, 699–709.
24. Alam, M.; Nieuwoudt, A.; Massoud, Y. Dynamic Multi-Point Rational Interpolation for Frequency-Selective
Model Order Reduction. In Proceedings of the 2006 IEEE Dallas/CAS Workshop on Design, Applications,
Integration and Software, Richardson, TX, USA , 29–30 October 2006; pp. 95–98.
25. Alam, M.; Nieuwoudt, A.; Massoud, Y. Provably passive second order model order reduction for package
parasitics using spectral zeros. In Proceedings of the 2007 IEEE Northeast Workshop on Circuits and Systems,
Montreal, QC, Canada, 5–8 August 2007; pp. 791–794.
26. Silveira, L.M.; Kamon, M.; Elfadel, I.; White, J. A Coordinate Transformed Arnoldi Algorithm for Generating
Guaranteed Stable Reduced Order Models of RLC Circuits. In Proceedings of the International Conference
on Computer Aided Design, San Jose, CA, USA, 10–14 November 1996; pp. 288–294.
27. Zheng, H.; Pileggi, L.T. Robust and Passive Model Order Reduction for Circuits Containing Susceptance
Elements. In Proceedings of the IEEE/ACM International Conference on Computer Aided Design, San Jose,
CA, USA, 10–14 November 2002; pp. 761–766.
28. Schilders, W.H.; Van der Vorst, H.A.; Rommes, J. Model Order Reduction, Theory, Research Aspects and
Applications; Springer: Berlin, Germany, 2008.
29. Phillips, J.; Silveira, L.M. Poor Man’s TBR: A Simple Model Reduction Scheme. IEEE Trans. Comput. Aided
Des. Integr. Circuits Syst. 2005, 24, 43–55.
30. Silveira, L.M.; Phillips, J. Exploiting Input Information in a Model Reduction Algorithm for Massively
Coupled Parasitic Networks. In Proceedings of the 41st Design Automation Conference, San Diego, CA,
USA, 7–11 July 2004; pp. 385–388.
31. Phillips, J.; Daniel, L.; Silveira, M. Guaranteed Passive Balancing Transformations for Model Order Reduction.
In Proceedings of the 2002 Design Automation Conference, New Orleans, LA, USA, 10–14 June 2002;
pp. 52–57.
32. Su, Q.; Balakrishnan, V.; Koh, C.K. Efficient Approximate Balanced Truncation of General Large-Scale RLC
Systems via Krylov Methods. In Proceedings of the 15th International Conference of Asia and South Pacific
Design Automation on VLSI Design, Bangalore, India, 7–11 January 2002; pp. 311–316.
33. Ismail, Y. Evaluating Noise Pulses in RC Networks due to Capacitive Coupling. In Proceedings of the 2002
IEEE International Symposium on Circuits and Systems, Phoenix-Scottsdale, AZ, USA, 26–29 May 2002;
pp. 653–656.
34. Vogel, C.R.; Wade, J.G. Iterative SVD-based methods for ill-posed problems. SIAM J. Sci. Comput. 1994, 15,
736–754.
35. Antoulas, A.C. A new result on passivity preserving model reduction. Syst. Control. Lett. 2005, 54, 361–374.
36. Alam M.; Nieuwoudt A.; Massoud Y. Frequency selective model order reduction via spectral zero projection.
In Proceedings of the 2007 Asia and South Pacific Design Automation Conference, Yokohama, Japan, 23–26
January 2007; pp. 379–383.
Electronics 2020, 9, 964 21 of 21

37. Sorensen, D. Passivity preserving model reduction via interpolation of spectral zeros. Syst. Control Lett.
2005, 54, 347–360.
38. Ionutiu, R.; Rommes, J.; Antoulas, A.C. Passivity-preserving model reduction using dominant spectral-zero
interpolation. IEEE Trans. Comput. Aided Des. Integr. Circuits Syst. 2008, 27, 2250–2263.
39. Mondal M.; Massoud, Y. Reducing Pessimism in RLC Delay Estimation Using an Accurate Analytical
Frequency Dependent Model for Inductance. In Proceedings of the IEEE/ACM International Conference on
Computer-Aided Design, San Jose, CA, USA, 6–10 November 2005; pp. 690–695.
40. Wang, J.M.; Chu, C.-C.; Yu,Q.; Kuh, E.S. On Projection-Based Algorithms for Model-Order Reduction of
Interconnects. IEEE Trans. Circuits Syst.- I 2002, 49, 1563–1585.
41. Kamon, M.; Wang, F.; White, J. Generating Nearly Optimally Compact Models from Krylov-Subspace Based
Reduced-Order Models. IEEE Trans. Circuits Syst.- II 2000, 47, 239–248.
42. Benini, L.; DeMicheli, G. Dynamic Power Management: Design Techniques and CAD Tools; Springer Science and
Business Media: Berlin, Germany, 2012.
43. Ionutiu, R.; Lefteriu, S.; Antoulas A. Comparison of model reduction methods with applications to circuit
simulation. In Scientific Computing in Electrical Engineering; Springer: Berlin, Germany, 2007; Volume 11.
44. Ionutiu, R. Model Reduction for Descriptor Systems with Applications to Circuit Simulation; William Marsh Rice
University: Houston, TX, USA, 2007; Rep. TREE07-07.
45. Mehrmann, V.; Stykel, T.; Benner, P.; Sorensen, D. Balanced Truncation Model Reduction for Large-Scale Systems
in Descriptor Form, in Dimension Reduction of Large-Scale Systems; Ser. LNCSE; Springer: Berlin, Germany,
2005; p. 45.
46. Charles, J.; Jassi, P.; Ananth, N.S.; Sadat, A.; Fedorova, A. Evaluation of the Intel® Core™ i7 turbo boost
feature. In Proceedings of the IEEE International Symposium on Workload Characterization, Austin, TX,
USA, 4–6 October 2009; pp. 188–197.
47. Nabors, K.; White, J. FastCap: A multipole accelerated 3-D capacitance extraction program. IEEE Trans.
Comput.-Aided Des. 1991, 10, 1447–1459.
48. Kamon, M.; Tsuk, M.J.; White, K. FASTHENRY: A multipole-accelerated 3-D inductance extraction program.
IEEE Trans. Microwave Theory Tech. 1994, 42, 1750–1758.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).