Democritus University of Thrace (DUTH)
Ανάπτυξη μεθοδολογίας για την βελτιστοποίηση τύπων δεδομένων δυναμικών εφαρμογών σε ενσωματωμένα συστήματα
Abstract
dc:descriptionThe next generation of embedded computing systems will be dominated by mobile, smart devices capable of delivering a wide range of communication and multimedia services anywhere. This set of devices includes PDAs, smartphones and video consoles. On the other hand, as technology progresses, users demand more and more complex and dynamic applications. Cases like voice transmission VoIP, high definition video rendering, internet browsing and 3D applications share as common characteristic the increased need in data storage and access. In order to provide a flowing performance in this wide range of applications, the optimization of the embedded systems is necessary in both the hardware and the software level. In dynamic applications, data is stored in entities called data structures, dynamic data types or simply containers, like arrays, lists or trees, which can adapt dynamically to its needs. These containers are realized at the software architecture level and are responsible for keeping and organizing the data in the memory and also servicing the application’s requests at run-time. The latter include abstract data type operators for storing, retrieving or altering any data value, which is not tied to a specific container implementation and shares a common interface. The implementation of these operators depends on the chosen container and each one of them can favour specific data access and storage patterns. Moreover, modern applications behave differently based on the various inputs that stimulate them. This leads to the execution of different paths in the source code, which in turn translates to a dynamic and most importantly unpredictable use of data. Since data are stored on data structures, the implementation of these structures should be customized to the particular data usage pattern. Therefore, it is necessary to carefully design the data structures inside a modern application. Unfortunately, in today’s development cycles for embedded systems, often there is not time for such an optimization. Developers tend to target the fast completion and debugging of the source code in order to quickly enter the market. As a result, they use existing solutions either in the form of template libraries or own implementations developed for other applications. These solutions are not customized for the particular application, which can lead to suboptimal performance. This thesis addresses the optimization of dynamic data structures in dynamic applications, focusing on the customization of these structures to meet the exact needs of the application. The proposed approach is automatic and tested in a variety of application domains.
Degree
thesis:*- Grantor dc:publisher
- Democritus University of Thrace (DUTH)
- Year dc:date
- 2011
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Baloukas, Christos
- Μπαλούκας, Χρήστος
Subjects
dc:subject × 14- Δομές δεδομένων
- Δυναμικές εφαρμογές
- Ενσωματωμένα συστήματα
- Βελτιστοποίηση C/C++
- Pareto βελτιστοποίηση
- Data structures
- Dynamic applications
- Embedded systems
- C/C++ optimization
- Pareto optimization
- Natural Sciences
- Computer and Information Sciences
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/26792