Journal of Computer Science 2 (3): 236-244, 2006

ISSN 1549-3636
© 2006 Science Publications

Empirical Analysis and Mathematical Representation

of the Path Length Complexity in Binary Decision Diagrams
1
A. Assi, 2P.W.C. Prasad, 2B. Mills and 2A. Elchouemi
1
Department of Electrical Engineering, United Arab Emirates University, Al Ain, UAE
2
College of Information Technology, United Arab Emirates University, Al Ain, UAE

Abstract: Information about the distribution of path-lengths in a Binary Decision Diagrams (BDDs)
representing Boolean functions is useful in determining the speed of hardware and software
implementations of the circuit represented by these Boolean functions. This study presents expressions
produced from an empirical analysis of a representative collection of Boolean functions. The Average
Path Length (APL) and the Shortest Path Length (SPL) have simple behavior as function of the number
of variables and the number of terms used in the construction of the Sum of Products (SOPs) in
Boolean expressions. We present a generic expression that is uniformly adaptable to each curve of
path-length versus number of terms over all the empirical data. This expression makes it possible to
estimate the performance characteristics of a circuit without building its BDD. This approach applies to
any number of variables, number of terms, or variable ordering method.

Key words: Binary decision diagram, Boolean function, average path length, shortest path length,
evaluation time

INTRODUCTION general the minimization of the path length in Decision

Diagrams (DDs) is important in database structures,
The use of logic verification and optimization pattern recognition, logic simulation and software
algorithms in VLSI CAD systems requires efficient synthesis[7]. The methods proposed for the minimization
representation and manipulation of Boolean functions[1]. of APL[7-10] reduces the average evaluation time of logic
During the last two decades, BDDs have gained great functions. Most of these methods are based on either
popularity as successful method for the representation Static variable ordering[11,12] or dynamic variable
of Boolean functions[2,3]. The ever-increasing ordering techniques[13]. The minimization of APLs leads
complexity of circuit designs is directly related to the to circuits with smaller depth of paths from the root to
complexity of parameters that describe the Boolean the terminal node of the BDD. The resulting circuit will
function. Over the years, the number of nodes in a BDD be optimized for speed on one hand and on the other
became a major concern since it is proportional to the hand the number of very long paths in the BDD will be
complexity of the Boolean circuit[4]. Over the past two reduced[14]. The minimization of APL is of great
decades most of the problems in the synthesis, design importance in real time operating system
and testing of combinational circuits, have been solved applications[10,15,16]. The minimization of the LPL
using various mathematical methods[5,6]. Researchers in (Longest Path Length) and SPL of a BDD can also
this area are actively involved in developing reduce the evaluation time, which is very important for
mathematical models that predict the number of nodes Pass Transistor Logic (PTL)[7,17,18]. One of the main
in a BDD in order to predict the complexity of the problems with pass transistor networks is the presence
design in terms of the time needed to optimize it and of long paths: the delay of a chain of n pass transistors
verify its logic. is proportional to n2. Inserting buffers can reduce the
Evaluation time is another crucial parameter of the path length, but this increases the silicon area. So the
circuit complexity and it is proportional to the path minimization of the longest evaluation time will
length of a BDD and one can use BDD structures to improve the performance of the circuit[7,18-20]
estimate the evaluation time of the logic function that Analysis of the BDD methods revealed that the
represents a circuit[7,8]. Therefore, minimization of the variable ordering in a given Boolean function plays an
path length can improve the complexity of the circuit important role in minimizing the size of the BDD graph
implementing a Boolean function, which will eventually as well as minimizing the path length[19,10]. One must go
enhance the performance of the final implementation. In through a number of simulations to find the suitable

Corresponding Author: A. Assi, Department of electrical Engineering, United Arab Emirates University, P.O. Box 17555, Al
Ain, U.A.E., Tel: 971-3- 7133609, Fax: 971-3-7626309
236
 J. Computer Sci., 2 (3):236-244, 2006

variable ordering that leads to the minimum size of the Each non-sink node represents the Boolean function
BDD and minimum Path Length. In this approach we corresponding to its edge "1" if v = 1, or the Boolean
need to create the whole BDD representing the Boolean function corresponding to its edge "0" if v = 0.
function with the best possible variable ordering. Definition 2: An Ordered BDD (OBDD) is a BDD in
Building the whole BDD may lead to some complexity
which each variable is encountered no more than once
in the design process in terms of the time required to
implement, verify and test the design. It will be useful to in any path and always in the same order along each
have a kind of estimation of the BDD complexity prior path.
to make decisions on the feasibility of the design[20]. For
any combinational circuit the only available initial Definition 3: A Reduced OBDD (ROBDD) is an
information is the Boolean function that represents this OBDD which no nodes have equivalent behavior.
circuit and the number of its variables. This information
is usually considered to design and verify circuits using Variable ordering: The size of a BDD is largely
well known mathematical methods. affected and its variation can be linear or exponential
There has been a lot of research[21-24] done on the depending on the choice of the variable ordering in
estimation of combinational and sequential circuit
building the BDD. Figure 1 illustrates the effect of the
parameters from the exact Boolean function describing
the circuit. What distinguishes this study and prior variable ordering [R.E. Bryant, 1986] on the size of
work[20,25-27] by the some of the current authors is the BDDs for the Boolean function (1):
use of stochastic technique and estimation of parameters
from only partial information about the Boolean f = x1 ⋅ x2 + x1 ⋅ x2 ⋅ x3 ⋅ x4 + x1 ⋅ x3 ⋅ x4 (1)
function.
It is very hard to perform a comparison without
having an idea about the path length size for a given
number of variables. Therefore, it is important to
develop a mathematical model that predicts the path
length, knowing the number of variables and the
number of product terms of the Boolean function
represented by this BDD.
The main objective of this study is to enhance the
methodologies proposed in[20,25] to estimate the path
length complexity for the Boolean functions represented
by the BDD. First, we present experiments that show
the behaviors of the APL and SPL and then we extract a (a) x1 x2 x3 x4 (b) x1 x3 x2 x4
unique mathematical model for produced experimental
graphs. This study is organized as follows: First is an
Fig. 1: Effect of the variable ordering on the size of
introduction, followed by the necessary terminologies
BDD
and definitions of the BDD and path length. . Later we
review the previous work done on the estimation of the
BDD complexity. The proposed method with the Definition 4: In a BDD, a sequence of edge and nodes
experimental results followed by the mathematical leading from the root node to a terminal node is called
model is given next. Finally the advantages of this Path. The number of non-terminal nodes on the path is
mathematical model followed by an outline of our called the Path Length.
future developments in this research work and
conclusion. Definition 6: The edge traversing probability, denoted
by P(ei 0 ) (or P(ei1 ) ), is the fraction of all 2n
PRELIMINARIES assignments of values to variables whose path includes
ei 0 (or ei1 ), where ei 0 (or ei1 ) denotes edge “0” (or the
Basic definitions for BDDs and path length are edge “1”) directed from away node Vi [7,8]. Since all
given in[1,3,4,7,10]. In the following we review some of paths include the root node, this node is traversed with
these definitions.
probability 1.00. Since all assignments to values of
variables are equally likely, we can use the following
Definition 1: A BDD is a directed acyclic graph
(DAG). The graph has two sink nodes labeled 0 and 1 equation (2) to calculate the P(Vi ) for the rest of the
representing the Boolean functions 0 and 1. Each non- nodes:
sink node is labeled with a Boolean variable v and has P (vi )
two out-edges labeled 1 (or then) and 0 (or else). = P(ei 0 ) = P (ei1 ) (2)
2
237
 J. Computer Sci., 2 (3):236-244, 2006

Definition 5: The APL is equal to the sum of the node Relation between the size of a boolean function and
traversing probabilities of the non-terminal nodes[7,10]. the ROBDD complexity[20]: The complexity of the
Node traversing probability denoted by P(vi ) is the ROBDD mainly depends on the number of nodes
fraction of all 2n assignments of values to the variables represented by the ROBDD. Analysis of the complexity
whose path includes node vi . The APL can be expressed variation in ROBDDs i.e. the relation between the
number of product terms and the number of nodes for
by the following equation (3):
N −1
any number of variables is discussed in these works, the
APL = P (vi ) (3) experimental graph variation reveals that the complexity
i =0 of the ROBDD can be modeled mathematically by
Where, N is the number of non-terminal nodes. equation (4). Figure 3 indicates that the mathematical
model represented by equation (4) provides a very good
Definition 6: The Shortest Path Length (SPL) of a approximation of the ROBDD complexity.
BDD denoted by SPL (BDD), is the Length of the
Shortest Path from the root node to the terminal node. NN = α ⋅ NPT β ⋅ e( − NPT ⋅γ ) + 1 (4)
Where, NN is the number of nodes that represents
Example: Consider the BDD graph shown in Fig. 2. In
the complexity of ROBDD, NPT is the number of non-
this example we will compute the APL and the SPL:
repeating product terms in the Boolean function, α , β
and γ are three constants. Using curve fitting
techniques, the variations of α, β and γ were
mathematically modeled and represented by the
following equations (5), (6) and (7).
1.51
α = 0.9855 ⋅ e(0.063⋅ NV ) (5)
β = 1.031149 ⋅ e( −0.01551933⋅ NV )
(6)
+67.2072 ⋅ e( −1.2985⋅ NV )
γ = 0.962297281 ⋅ e( −0.4187691⋅ NV )
(7)
+41.9723 ⋅ e( −1.5072⋅ NV )
Where, NV is the Number of Variables.

Fig. 2: Node Traversing Probability in a BDD

The root node P(V0 ) is always equal to 1.00.

P(V1 ) = P(e00 ) = 0.50
P (V2 ) = P (e10 ) = 0.50 .
P(V3 ) = P(e2 0 ) = 0.25
P (V4 ) = P (e21 ) = 0.25
P (V5 ) = P (e4 0 ) + P (e11 ) = 0.125 + 0.25 = 0.375
Finally Fig. 3: Experimental/Equation BDD Complexity for
5
10 variables
APL = P (Vi ) = 2.875
I =0
Behavior of XOR/XNOR Min-term
Representations[26]: In this work, the complexity
LPL = Shortest Path Length = x1 → x2 = 2 variation in ROBDD for a specific group of
XOR/XNOR min-terms is analyzed. A graph that
Previous work: Here, we provide a brief description of represents the ROBDD complexity in terms of number
the works done in the area of the estimation of BDD of nodes with respect to the number XOR/XNOR min-
terms of the Boolean function is then plotted and the
complexity prior to explaining the proposed method.
behavior of XOR/XNOR is modeled mathematically by
238
 J. Computer Sci., 2 (3):236-244, 2006

equation (8): Figure 4 show that the mathematical the APL and SPL reaches a maximum
model represented by equation (8) provides a good ( APL ≅ 7.73 , SPL ≅ 5.4 in this case).
approximation of the experimental ROBDD complexity.
0.5
NN = α ⋅ β 2 − ( NXM − β ) 2 +1 (8)
Where, NN is the number of nodes that represents
the complexity of ROBDD, NXM is the number of
XOR/XNOR min-terms in the Boolean function, β is
2n-2 with n the number of input variables and α =
0.605234.

ANALYSIS OF THE COMPLEXITY OF PATH

LENGTH IN BDDS

Proposed method: An experiment was carried out

using Colorado University Decision Diagram (CUDD) Fig. 5: APL size variation for 10 variables using the
Package[28] to analyze the complexity variation of SPL symmetric Sift reordering technique
with the number of product terms for any number of
variables. For each variable count n between 1 and 14
inclusive and for each term count between 1 and 2n-1,
100 Boolean functions were randomly generated and
the APL and SPL average was determined by using
CUDD package for specific variable ordering
technique. This process was repeated until the average
size of the APL and SPL complexities became 1. Then
the experimental graphs for APL and SPL complexities
were plotted against the product term count for each
variable count.

Fig. 6: SPL size variation for 10 variables using the

Symmetric Sift reordering technique

This peak indicates the maximum APL and SPL that

any Boolean function with 10 variables can produce
independently of the number of product terms.
Apart of that the peak also specifies the number of
product terms (critical limit) of a Boolean function that
leads to the maximum number of APL and SPL for any
Boolean function with 10 variables. The number of
. product terms that leads to the maximum APL and SPL
is 66 and 50 respectively. If the number of product
Fig. 4: Experimental/Equation ROBDD Complexity
terms increases above the critical limit, as expected, the
for XOR/XNOR Min-terms
product terms starts to simplify and the BDD will
reduce, which will reduce both the path lengths (APL
Experimental analysis for APL and SPL complexity
and SPL) size.
variations: Figure 5 and 6 illustrates the APL and SPL
The APL and SPL complexity graphs shown in
complexity relation for Boolean functions with product
Fig. 5 and 6 indicate that as the number of product
terms having n=10 variables using the Symmetric Sift
reordering technique of the CUDD tool. terms increases the complexity of the APL and SPL
The graph indicates that the complexity (i.e. size) decreases in a slower rate and ultimately reaches 0.
of the path length in general (APL and SPL) increases Figure 6 illustrates that the falling edge of the SPL
as the number of product terms increases. This is clear graph behaves a bit different than the other complexity
from the rising edge of the curve shown in Fig. 5 and graphs shown in Fig 2 and 3, where the decrease is with
6. At the end of the rising edge in the graph, the size of a roll off, to be independent of the variable count. The

239
 J. Computer Sci., 2 (3):236-244, 2006

APL complexity variation graph is fairly similar to the β 2 = 2.1 for v ≤ 11 and β 2 = 3.5 for v ≥ 12 .
Fig. 2, but the roll of is not steeper as the Fig. 2. Eventually the following equation (11) was used in
The location and height of the peak and the slope of the order to calculate the constant β 2 for APL.
logarithm of the roll off varied. Reduction of the APL
and SPL complexity to 0 implies that all the product 1.8
β 2 = 0.7 3 + (11)
terms simplify to logic 1. A simple algebraic expression [e( v −11.5) ]2 + 1
for these curves was developed, unifying all the cases. The final behavior of the APL and SPL curve can
be found by the following single equation (12):
MATHEMATICAL MODEL FOR THE αi

PATH LENGTH BEHAVIOR log t +1

2 xi
1= ⋅ yi
(log( t x + 1)) βi (12)
i =1
Exponentials of rational polynomials fitted the data
well; but, a theoretical precedent was not apparent. On
the other hand, log(t + 1) (t + 1) not only has the In this mathematical model, the peaks ( xi , yi ) for
same basic behavior, but is also implicated in other both the APL and SPL curves were found by
complexity measures, such as Kolmogorov, Tichner, performing an empirical fit for each time. Figure 8 and
9 depict the experimental results obtained for APL and
Shannon and Lempel-Zif complexity, as well as the
SPL using the CUDD package and the theoretical
density of the prime numbers. The generic SPL graph
results obtained using equation (12). The mathematical
has an initial rise, two peaks and roll off to zero, model represented by equation (12) provides a very
suggesting the sum of two formulas, but with horizontal good estimation for the APL and SPL complexity
and vertical scaling and a little peak shaping. We note behavior, where the experimental and equational results
here that the second peak is not always a peak of the produced a match. Further verification of the
curve, but it is a peak of the difference between the mathematical model was done for Boolean functions
curve and the best affine approximation in that region. with 2 to 15 variables. It can be inferred that the
The generic APL graph has an initial rise which is experimental and mathematical curves are following a
similar to SPL rise, but with only one peak and a roll off similar pattern for any number of variables. Figure 10-
to 0. Analyzing all the above factors for the behavior of 13 illustrates the experimental and mathematical models
the APL and SPL graphs, the complexity behavior was for APL and SPL for variables 8 and 12 respectively.
modeled mathematically by the following equations:
αi
2
log (t + 1)
1= βi (9)
i =1 (log (t + 1))

Where, t is the number of product terms in the

Boolean function. The (mostly) constants α and
β parameters affect the shape of the peak.
For the SPL, the
values α1 = 7, β1 = 1 and α 2 = 10 gave a close fit, but
β2 taking on two distinct values. β2 = 3 for v ≤ 11
and β 2 = 5 for v ≥ 12 . Eventually the following
Fig. 7: Peaks of the SPL complexity behavior
equation (10) was used in order to calculate the
constant β 2 ,

1.8
β2 = 3 + ( v −11.5) 2
(10)
[e ] +1

It can be inferred from Fig. 6 that the curve has two

peaks, which needs four scaling parameters to define
the locations of the peaks (Fig. 7): i.e. ( x1 , y1 ) and
( x2 , y2 ) .
For the APL, the alues α1 = 7, β1 = 0.7 and α 2 = 10 gave
a close fit, but β 2 also taking on two distinct values. Fig. 8: Experimental/Equation SPL complexity
behavior for 10 variables
240
 J. Computer Sci., 2 (3):236-244, 2006

Fig. 9: Experimental/Equation APL Complexity Fig. 12: Experimental/Equation SPL Complexity

behavior for 10 variables behavior for 8 variables

Fig. 10: Experimental/Equation APL Complexity Fig. 13: Experimental/Equation SPL Complexity
behavior for 8 variables behavior for 12 variables

Fig. 11: Experimental/Equation APL Complexity Fig. 14: APL Complexity Estimation Error for 10
behavior for 12 variables variables
241
 J. Computer Sci., 2 (3):236-244, 2006
αi
log t +1
2 xi (13)
1= µ ⋅ ⋅ yi
i =1 ( (
log t + 1
x
))
βi

The amplification factor was calculated and

depicted in Table 1. Figure 16 shows the comparison
graphs of the APL and SPL behaviors for Symmetric
Sift with two of the other CUDD variable ordering
techniques mainly the Genetic Algorithm and
Window2. These two graphs show that the efficiency of
the reordering method has a definite impact on the path
length complexity; an efficient variables ordering leads
to a reduced number of nodes, which leads to reduced
Fig. 15: SPL Complexity Estimation Error for 10 path lengths.
variables
Table 1: Amplification factor (µ)
Variable Reordering Method Amplification Factor (µ)
Random 1.024
Random Pivot 0.998
Sift 1.001
Symmetric Sift 1.000
Symmetric Sift Converge 0.971
Group Sift 1.006
Group Sift Converge 0.963
Window 2 1.085
Window 3 1.045
Window 4 1.018
Window Converge 2 1.058
Window Converge 3 1.025
Window Converge 4 0.989
Annealing 0.945
Genetic Algorithm 0.942
Exact 0.942
Fig. 16: Effect of the reordering methods for SPL and
APL Complexity variations
Advantages: The developed mathematical model
represented by equation (10), provides some useful
Figure 14-15 shows the efficiency of the proposed
information on the following, without the need of
mathematical model, which produces complexity
building the BDD.
estimation error for APL and SPL. It can be inferred
1. The complexity behavior of the APL and SPL,
that the mathematical expression was able to match the
given the number of product terms of the Boolean
experimental curve with minimum error, which is less
function
then ±0.01 for most of the Product terms.
2. The number of product terms for which the
maximum possible depth will occur.
Effect of the reordering methods on path length
3. The maximum complexity of the APL and SPL of
complexity variations: The experiment done earlier
Boolean functions for any number of variables.
using the Symmetric Sift CUDD reordering method was
extended here to understand the relation of Symmetric
CONCLUSION AND FUTURE WORK
Sift APL and SPL graphs with other reordering
techniques. It was observed that the relation between Future work includes minimizing the Complexity
the graphs follows the same pattern and it varies only on estimation error of the match and to develop
the amplitude factor of the curves. experiments to include larger number of variables. We
By analyzing the effect of the reordering methods are in the process of investigating an automated global
on the model, equation (12) can be modified with an fit for any SPL and APL curves in order to find the
additional amplification factor (µ). The amplification complexity for any number of product terms.
factor is 1 for the Symmetric Sift, greater than 1 for Investigating a mathematical model for other BDD
methods with lower efficiency and less than 1 for characteristics (i.e. longest path length and number of
methods with higher efficiency than the Symmetric Sift. paths) is also considered.
Equation (13) represents the mathematical model for the We have discussed the idea of using BDD to study
APL and SPL for any reordering method. and model a relationship between the path lengths and

242
 J. Computer Sci., 2 (3):236-244, 2006

the number of product terms in a Boolean function. 11. Fujita, M., H. Fujisawa and N. Kawato, 1988.
Analyzing the Experimental results, we have introduced Evaluation and improvements of boolean
a single and unique mathematical model, which is based comparison method based on binary decision
on an empirical fit that can predict valuable information diagrams. Proc. Intl. Conf. on Computer Aided
related to the APL and SPL behaviors without building
Design (ICCAD)., pp: 2-5.
the BDD. A great reduction in time complexity for
digital circuits' designs can be achieved and the model 12. Malik, S., A. Wang, R. Brayton and A.S.
can also offer useful information on the design to Vincentelli, 1988. Logic verification using binary
handle the minimization of its evaluation time prior to decision diagrams in a logic synthesis environment.
its implementation. Our experimental results show good Proc. Intl. Conf. on Computer Aided Design
correlation between the experimental results and those (ICCAD)., pp: 6-9.
given by the mathematical model. 13. Rudell, R., 1993. Dynamic variable ordering for
ordered binary decision diagrams. Proc. Intl. Conf.
REFERENCES on Computer Aided Design (ICCAD)., pp: 42-47.
14. Fey, G., J. Shi and R. Drechsler, 2004. BDD circuit
1. Priyank, K., 2000. VLSI Logic Test, Validation
optimization for path delay fault-testability. Proc.
and Verification, Properties & Applications of
of EUROMICRO Symp. on Digital System Design,
Binary Decision Diagrams. Lecture Notes,
pp: 168-172.
Department of Electrical and Computer
15. Balarin, F., M. Chiodo, P. Giusto, H. Hsieh, A.
Engineering University of Utah, Salt Lake City, UT
Jurecska, L. Lavagno, A.S. Vincentelli, E.M.
84112.
Sentovich and K. Suzuki, 1999. Synthesis of
2. Akers, S.B., 1978. Binary decision diagram. IEEE
software programs for embedded control
Trans. Computers, 27: 509-516.
applications. IEEE Trans. CAD., 18: 834-849.
3. Bryant, R.E., 1986. Graph based algorithm for
16. Lindgren, M., H. Hansson and H. Thane, 2000.
boolean function manipulation. IEEE Trans.
Using measurements to derive the worst-case
Computers, 35: 677-691.
execution time. Proc. 7th Intl. Conf. on Real-Time
4. Drechsler, R. and D. Sieling, 2001. Binary
Systems and Applications (RTCSA’00)., pp: 5-22.
Decision Diagrams in Theory and Practice.
17. Shelar, R.S. and S.S. Sapatnekar, 2001. Recursive
Springer-Verlag Trans., pp: 112-136.
bipartitioning of BDD's for performance driven
5. Jain, A., M. Narayan and A.S. Vincentelli, 1997.
synthesis of pass transistor logic. Proc. of
Formal verification of combinational circuits. Proc.
IEEE/ACM ICCAD., pp: 449 - 452.
Intl. Conf. on VLSI Design, pp: 218-225.
18. Bertacco, V., S. Minato, P. Verplaetse, L. Benini
6. Van Eijk, C.A.J., 1997. Formal methods for the
and G.D. Micheli, 1997. Decision diagrams and
verification of digital circuits. Ph.D. Thesis,
pass transistor logic synthesis. Stanford University
Eindhoven University of Technology, Netherlands.
CSL Technical Report, No. CSL-TR-97-748.
7. Nagayama, S. and T. Sasao, 2004. On the 19. Görschwin, F., S. Junhao and R. Drechsler, 2004.
minimization of longest path length for decision BDD circuit optimization for path delay fault-
diagrams. Proc. of Intl. Workshop on Logic and testability. Proc. EUROMICRO Symp. on Digital
Synthesis (IWLS-2004), pp: 28-35. System Design, pp: 168-172.
8. Liu, Y., K.H. Wang, T.T. Hwang, and C.L. Liu, 20. Raseen, M., P.W.C. Prasad and A. Assi, 2005. An
2001. Binary decision diagrams with minimum efficient estimation of the ROBDD's complexity.
expected path length. Proceedings of DATE 01, pp: accepted for Publication in Integration - the VLSI
708-712. Journal, Elsevier Publication.
9. Prasad, P.W.C., M. Raseen, S.M.N.A. Senanayake 21. Nemani, M. and F.N. Najm, 1996. High-level
and A. Assi, 2005. BDD path length minimization power estimation and the area complexity of
based on initial variable ordering. accepted for
boolean functions. Proc. of IEEE Intl. Symp. on
Publication in Journal of Computer Science,
Science Publications, May 2005. Low Power Electronics and Design, pp: 329-334.
10. Ebendt, R., S. Hoehne, W. Guenther and R. 22. Dunne, P.E. and W. van der Hoeke, 2004.
Drechsler, 2004. Minimization of the expected path Representation and complexity in boolean games.
length in BDDs based on local changes. Proc. of Proc. 9th European Conf. on Logics in Artificial
Asia and South Pacific Design Automation Conf. Intelligence, LNCS 3229, Springer-Verlag, pp:
(ASP-DAC’2004), pp: 866-871. 347-35.

243
 J. Computer Sci., 2 (3):236-244, 2006

23. Ramalingam, N. and S. Bhanja, 2005. Causal 26. Prasad, P.W.C., M. Raseen and S.M.N.A.
probabilistic input dependency learning for Senanayake, 2005. XOR/xnor functional behavior
switching model in VLSI circuits. Proc. of ACM on ROBDD representation. Proc. 14th IASTED
Great Lakes Symp. on VLSI, pp: 112-115. Intl. Conf. on Applied Simulation and Modelling
24. Bhanja, S., K. Lingasubramanian and N. (ASM 2005), pp. 115-119, Spain.
Ranganathan, 2005. Estimation of switching 27. Raseen, M., A. Assi, P.W.C. Prasad and A. Harb,
activity in sequential circuits using dynamic 2004. Effect of boolean min-terms on the
bayesian networks. Proc. of VLSI Design, pp: 586- complexity of ROBDDs. Proc. Intl. Conf. on
591. Computational Intelligence (ICCI 2004), pp: 454-
25. Raseen, M., P.W.C. Prasad and A. Assi, 2004. 457.
Mathematical model to predict the number of nodes 28. Somenzi, F., 2003. CUDD: CU Decision Diagram
in an ROBDD. Proc. 47th IEEE Intl. Midwest Package. <ftp://vlsi.colorado.edu/> pub/.
Symp. on Circuit and Systems (MWSCAS), 3: 431-
434.

244