ISSN 10637397, Russian Microelectronics, 2011, Vol. 40, No. 7, pp. 486–490. © Pleiades Publishing, Ltd., 2011.
Original Russian Text © D.S. Knysh, V.M. Kureichik, 2009, published in Izvestiya Vysshikh Uchebnykh Zavedenii. Elektronika, 2009, No. 5, pp. 28–34.
There are many algorithms for solving the problem
of routing the switching unit [1, 2]. Genetic algo
rithms (GA)  occupy a leading position among
them. Recently, a new trend has emerged in GA, i.e.,
fuzzy genetic algorithms (FGAs).
The task of routing is one of the most difficult
stages of VLSIC design, so it is broken into two phases,
i.e., global and detailed routing. The detailed routing,
in turn, is divided into routing channels and switching
units. Switching units are formed at the intersection of
channels and contain a large number of circuits, which
leads to excessive heat and overheating of parts of the
VLSIC in the case of lowquality routing in these
areas. In solving this problem, multiple criteria must
be taken into account affecting the quality of the
developed product, including new subtle physical
effects that were not significant earlier. An algorithm
for solving the routing task should consider all the cri
teria. In our opinion the most popular method of cal
culating the objective function using the additive con
volution, does not give a complete picture of the topol
ogy of the switch unit . Therefore, the fuzzy method
has been proposed for assessing the topology of the
switching unit, which is used in the work of genetic
operators and uses the fuzzy operators over fuzzy sets
for the calculation of the final objective function. The
influence of fuzzy techniques on the final solution of
GA has been studied. Namely, the comparison of the
operation of a simple GA and FGA was carried out for
different topologies of a switching unit, which showed
an improvement of solutions from 5 to 30%, even with
a small number of iterations, which indicates the pros
pects of research in this area.
This paper presents a model of the FGA routing of
switching units of circuits of varying widths. The
model is based on the use of genetic operators with the
fuzzy objective function (FOF). The NGATKB pro
gram  has been developed using the C++ program
ming language for Windows XP for routing circuits of
different widths in twolayer switching units. The pro
gram was used to study the work of the FGA and deter
mine the effect of fuzzy methods of calculating the
objective function on the work of genetic operators.
Formulation of the problem.
Routing is the process
of path connection in accordance with imposed
restrictions . The routing of an IC and VLSIC is
divided into global (the mapping of circuits in certain
regions) and detailed (wire routing inside regions)
routing. Detailed routing is divided into the routing of
switching units and routing of channels (Fig. 1) .
In the classical implementation, the switching unit
is an area bounded from four sides by lines of contacts
Genetic Algorithm for Routing Switching Units
D. S. Knysh and V. M. Kureichik
Technological Institute of Southern Federal University, Taganrog, Russia
—A hybrid model is presented of a genetic algorithm for routing switching units using fuzzy logic in
genetic operators. A fuzzy objective function is used to take into account the criteria of the problem. The algo
rithm has been developed with regard to various physical effects during routing, thereby reducing the design
time of devices by eliminating the need for rerouting after the verification stage.
CIRCUIT TECHNOLOGY AND DESIGN
Routing of the
Routing of the channel
Types of detailed routing.