Aristotle University Of Thessaloniki (AUTH)
Χρονοδρομολόγηση εργασιών σε κατανεμημένα συστήματα - εκτίμηση της απόδοσης
Abstract
dc:descriptionIn the field of modern distributed systems, where a lot of applications run at the same time, the accurate resource management technique that ensures end to end Quality of Service (QoS) is perhaps one of the most important issues for the end user. New applications are continuously being born, particularly applied for pervasive environments that enable high rates connectivity between mobile devices and at the same time claiming end to end reliability and integrity. Moreover, the software that is intended for pervasive services in ubiquitous computing environment requires a new design form of distributed network-centric applications. Taking all these issues into account the creation of efficient strategies becomes imperative need for the localised monitoring of mobile users characteristics. These characteristics are in terms of energy conservation and energy consumption of nodes, synchronization, scheduling and co-scheduling of all internal communicational processes. Such problems can be faced using significances that imitate behaviours of biological entities, or mimic the characteristics and behaviours of these entities. These entities co-operate to iteratively construct potential observation schedules. By combining cognitive science and anticipatory systems of biological intelligence via artificial biological entities and behaviours, these techniques can be applied to distributed systems offering significant throughput optimization. In this research, several case studies have been used to experimentally assess the performance of the proposed approaches in terms of quality solutions and computational effort. A variety of disciplines have recently advocated the use of self-adaptive and auto-configuration methods, including biodynamics, cybernetics and computer modelling. Of these methods, one which exhibits numerous powerful features that are desirable in communication systems is adaptive swarm based intelligence. Swarm-based self-configuration does not require the need of external help, supervision or control. An approach is modeled and presented called bandwidth clustering scheme suited for network resource allocation problem. Bandwidth clustering is used in a swarm-based active network environment where active packets continuously communicate with active nodes by using the Split Agent Routing Technique (SART). This mechanism enables the adaptation of the system to new conditions (bandwidth reservation/capacity allocation), as well as the passing of additional information to neighbouring nodes for which the information is held in transmitted packets. The performance, reliability and adaptivity degree of the proposed scheme is thoroughly examined through simulation, for different traffic measures, as well as the corresponding QoS offered (in terms of the end-to-end delay, available bandwidth and probability of packet loss). An approach for the stabilisation of reliability in mobile environment is also presented, in terms of communication maintenance and efficient utilization of battery power. Unlike wired networks in mobile networks, mobility induces frequent route changes. Energy conservation is an important issue that has to be taken into account for mobile devices. Traffic plays a major role for energy consumption because of its unpredictable incoming flow nature. This part of work proposes an adaptive traffic-based control method for energy conservation, which bounds an asynchronous operation where each node evaluates dissimilar sleepwake schedules/states based on each node’s incoming sleep-history traffic. Simulation study is carried out for throughput, traffic characterization against performance and energy conservation evaluation of the proposed model taking into account a number of metrics and estimation of the effects of incrementing the sleep time duration to conserve energy. The proposed method could be applied to infrastructureless networks with any underlying routing protocol to provide independency, portability, as well as “fair” collaboration among energy conservation mechanism and routing protocol. Due to the spread of different file sharing platforms, peer-to-peer applications have become highly popular in today's pervasive environments. These applications could host multimedia on-demand applications in a fully distributed dynamically changing topology environment. In such a multiclient environment, the asymmetry in communication causes a degradation of reliability. A novel mimicking approach is proposed based on the advantages of epidemic selective resource placement through mobile Infostations. Epidemic placement policy combines the strengths of both proactive multicast group establishment and hybrid Infostation concept. With epidemic selective placement we face the flooding problem locally (in geographic region/landscape) and enable end to end reliability by forwarding requested packets to epidemically ‘selected’ mobile users in the network on a recursive basis. The sharing policy is based on a mathematical model of Markov chain rules. Markov chain model was chosen to evaluate extra storage requirements, for modelling of infectious diseases to determine the file sharing termination criteria. The selection of users is performed based on their remaining capacity, weakness of their signal and other mobility limitation characteristics. Experimental examination through simulation is performed for the response and reliability offered by epidemic placement policy which reveals the robustness and reliability in file sharing among mobile peers.
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2005
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Mavromoustakis, Constandinos
- Μαυρομουστάκης, Κωνσταντίνος
Subjects
dc:subject × 18- Κινητές λογισμικές μονάδες πρακτόρων
- Διαμοιρασμός δεδομένων
- Κατανεμημένα συστήματα
- Υπολογιστική βιολογική νοημοσύνη
- Τεχνητές σημασιολογικές επιδημικές μεθόδους
- Δυναμική δρομολόγηση
- Συντήρηση ενέργειας
- Mobile software agents
- Data dissemination
- Distributed systems
- Biologically inspired intelligence
- Epidemic conceptual methods
- Dynamic itinerary
- Energy conservation
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Natural Sciences
- Computer and Information Sciences
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/14997