KFUPM ePrints

Fuzzy genetic algorithm for VLSI floorplan design.

Adiche, Hakim Salah. (1997) Fuzzy genetic algorithm for VLSI floorplan design. Masters thesis, King Fahd University of Petroleum and Minerals.

[img]
Preview
PDF
3880Kb

Arabic Abstract

التخطيط السطحي من الخطوات الهامة في تصميم الدوائر المتكاملة ذات النطاق الواسع جداً . وهدف التخطيط السطحي هو البحث عن حلول مقبولة لمسألة صعبة ذات تصاميم مختلفة متعلقة بتصغير المساحة السطحية وشكل الوحدات أو الخلايات وفي نفس الوقت تصغير طول التوصيلات بين الوحدات وتحسين سرعتها . ويعتبر المنطق الغير الواضح (Fuzzy Logic) وسيلة طبيعية لحل هذا النوع من المسائل الصعبة من خلال تصميم قواعد وقيم غير واضحة مرتبطة بمسألة معينة . في هذه الرسالة تقوم بتنفيذ خوارزمية جينية غير واضحة ، وتقوم التجارب باستعمال الخوارزمية الجينية الغير الواضحة لحل مسألة التخطيط السطحي على شكل شرائح تم مقارنة النتائج مع تلك التي تحصل عليها باستعمال (Simulated Anncaling) و (Tabu Search) الغير الواضحين .

English Abstract

Floorplanning is an essential step when a building block VLSI design methodology is used. Floorplanning is executed to seek answers to several difficult design problems such as sizes and shapes of blocks, location of pins and pads, etc., while attempting to minimize several objectives. Examples of objectives are minimization of overall area of floorplan, minimization of delays and minimization of total wire length. Floorplanning falls into the class of hard combinational optimization problems with multiple objectives and constraints. Iterative algorithms have been found effective search heuristics for such class of problems. However, there has not been agreement on how best one can evaluate the cost of individual solutions in the case of multi-objective optimization. Fuzzy logic is a natural vehicle for solving this problem. Fuzzy logic allows the mapping of different objectives into the interval [0, 1] through a carefully designed set of fuzzy logic values and fuzzy logic rules. This fuzzy mapping transforms the optimization of a vector-valued function into the optimization of a scalar function, resulting from the activation of a nuber of fuzzy logic rules. In this work, fuzzy Genetic, simulated annealing, and tabu search algorithms (FGA, FSA, and FTS respectively) are implemented. The evaluation of the cost of an individual solution is the result of the application of the fuzzy logic rules which combine fuzzy linguistic values defined on the problem domain. The FGA, FSA, and FTS were tested on the floorplanning problem with the same fuzzy cost functions.



Item Type:Thesis (Masters)
Date:November 1997
Date Type:Completion
Subjects:Computer
Divisions:College Of Computer Sciences and Engineering > Computer Engineering Dept
Creators:Adiche, Hakim Salah.
Committee Advisor:Youssef, Habib
Committee Members:Sait, Sadiq M. and El-Barr, Mostafa Abd and Jamoussi, Mohammed
ID Code:9593
Deposited By:KFUPM ePrints Admin
Deposited On:22 Jun 2008 16:44
Last Modified:25 Apr 2011 08:52

Repository Staff Only: item control page