Uni-Logo
Deutsch      
Computer Architecture - Team Bernd Becker
        Startseite         |         Institut für Informatik         |         Technische Fakultät
 

Binary Decision Diagrams (BDDs)

| project staff | project description | project structure | publications |


project staff

Chair of Computer Architecture
Rolf Drechsler, Prof. Dr.
Wolfgang Günther, Dr.


project description

For the verification of digital circuits binary decision diagrams (BDDs) were established in 1986. These don't only allow a compact and clear representation of boolean functions, but there are also efficient algorithms for manipulation. For example the AND of two functions can be achieved in time proportional to the product of both BDDs.
An example: for the boolean function f(x1,x2,x3) = x1 x2 + x3 the BDD is given by:



project structure

Optimizing the order of variables



The size of BDDs depends on the based order of the variables. Therefore their optimization is an important problem which has proved itself to be difficult (NP-hard). Due to this many heuristic methods for minimizing BDDs have been suggested up to now. To accelerate an exact method and also heuristics, lower bounds were used, which mainly were used in theoretical observations so far. Furthermore the heuristic minimization also analyses the best form of minimization depending on the previous setup-process.

Linerarly transformated BDDs (LTBDDs)



Since there are functions of exponential size under all orders of variables it is necessary to generalize BDDs so that the displaysize can be decreased but that there are still efficient BDDs for manipulation. Such an approach are linearly transformed BDDs (LTBDDs), which transform the inputvariables with the help of EXOR-links. Here an exponential difference in size could be shown, ther are good heuristics for minimizing display and synthesisoperations can still be processed efficiently.

Free BDDs (FBDDs)



Another generalization of BDDs are Free BDDs (FBDDs). Here an exponential difference is also possible in theory. However there aren’t very many efficient minimizationtechnics up till now. Here an exact method was presented and also a heuristic, which provides FBDDs which are smaller than the best known BDDs for the first time.



publications
Wolfgang Günther, Dr., Rolf Drechsler, Prof. Dr.
Linear Transformations and Exact Minimization of BDDs
Great Lakes Symp. VLSI, 1998
Rolf Drechsler, Prof. Dr., Nicole Drechsler, Dr., Wolfgang Günther, Dr.
Fast Exact Minimization of BDDs
IEEE Design Automation Conference, 1998
Wolfgang Günther, Dr., Rolf Drechsler, Prof. Dr.
BDD Minimization by Linear Transformations
Advanced Computer Systems, 1998
Wolfgang Günther, Dr., Rolf Drechsler, Prof. Dr.
Minimization of Free BDDs
ASP Design Automation Conf., 1999
Wolfgang Günther, Dr., Rolf Drechsler, Prof. Dr.
Minimization of BDDs using Linear Transformations based on Evolutionary Techniques
IEEE International Symposium on Circuits and Systems, 1999
Rolf Drechsler, Prof. Dr., Wolfgang Günther, Dr.
Using Lower Bounds During Dynamic BDD Minimization
IEEE Design Automation Conference, 1999
Wolfgang Günther, Dr., Rolf Drechsler, Prof. Dr.
Efficient Manipulation Algorithms for Linearly Transformed BDDs
International Conference on Computer Aided Design, 1999
Rolf Drechsler, Prof. Dr., Wolfgang Günther, Dr.
History-based Dynamic Minimization during BDD Construction
X IFIP International Conference on VLSI, 1999
Rolf Drechsler, Prof. Dr., Wolfgang Günther, Dr.
Optimization of Sequential Verification by History-Based Dynamic Minimization of BDDs
IEEE International Symposium on Circuits and Systems, 2000
Wolfgang Günther, Dr.
Minimization of Free BDDs using Evolutionary Techniques
Int'l Workshop on Logic Synthesis, 2000
Wolfgang Günther, Dr., Rolf Drechsler, Prof. Dr.
On the Computational Power of Linearly Transformed BDDs
Information Processing Letters, 75(3):119-125, 2000