Please use this identifier to cite or link to this item: https://dspace.univ-ouargla.dz/jspui/handle/123456789/36979
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorMohamed Ben Bezziane-
dc.contributor.advisorBasmaHamrouni-
dc.contributor.authorBoussaid, Hadjer-
dc.contributor.authorDjoudi, Dalale-
dc.contributor.authorReguig, Nada-
dc.date.accessioned2024-10-01T08:55:48Z-
dc.date.available2024-10-01T08:55:48Z-
dc.date.issued2024-
dc.identifier.citationFaculté des Nouvelles Technologies de l’Information et de la Communicationen_US
dc.identifier.urihttps://dspace.univ-ouargla.dz/jspui/handle/123456789/36979-
dc.descriptionIndustrial Computer Scienceen_US
dc.description.abstractThis thesis explores the application of Long Short-Term Memory (LSTM) models to predict vehicle trajectories, aiming to optimize urban transportation systems and enhance the deployment of Mobile Edge Computing (MEC) units in smart city environments. The research utilizes a comprehensive dataset comprising GPS coordinates of taxis in San Francisco, processed into time series data for predictive modeling. Through meticulous experimentation, the thesis demonstrates the superior performance of LSTM models over traditional Simple Recurrent Neural Networks (RNNs) in predicting vehicle positions with high accuracy. These predictions are crucial for determining optimal locations for MEC units, which are strategically hosted on or near 5G antennas to improve data throughput and reduce communication latency. By clustering the predicted positions, the study provides a framework for efficient MEC placement, contributing to the reduction of network congestion and enhancement of urban infrastructure management. This work not only highlights the capabilities of LSTM models in real-time data analysis but also underscores their potential in supporting the infrastructure for next-generation wireless networks in urban settings.en_US
dc.description.abstract‫تستكشف‬‫هذه‬ ‫األطروحة‬ ‫تطبيق‬ ‫نماذج‬ ‫الذاكرة‬ ‫قصيرة‬ ‫األمد‬ ‫طويلة‬ ‫األمد‬ ‫(‬ ‫‪LSTM‬‬‫)‬‫للتن‬ ‫بؤ‬ ‫بمسارات‬‫المركبات‪،‬‬ ‫بهدف‬ ‫تحسين‬ ‫أنظمة‬ ‫النقل‬ ‫الحضري‬ ‫وتعزيز‬ ‫نشر‬ ‫وحدات‬ ‫الحوسبة‬ ‫المتنق‬ ‫لة‬‫عل‬ ‫ى‬ ‫الحافة‬‫(‬ ‫‪MEC‬‬‫)‬ ‫في‬ ‫بيئات‬ ‫المدينة‬ ‫الذكية‪.‬‬ ‫تستخدم‬ ‫البحث‬ ‫مجموعة‬ ‫بيانات‬ ‫شاملة‬ ‫تتضمن‬ ‫إح‬ ‫داثيات‬‫‪GPS‬‬ ‫لسيارات‬‫األجرة‬ ‫في‬ ‫سان‬ ‫فرانسيسكو‪،‬‬ ‫تم‬ ‫معالجتها‬ ‫إلى‬ ‫بيانات‬ ‫متسلسلة‬ ‫زمنيًا‬ ‫للنمذجة‬ ‫ا‬ ‫لتنبؤ‬‫ية‪.‬‬‫من‬ ‫خ‬ ‫الل‬ ‫التجارب‬‫المدققة‪،‬‬ ‫تُظهر‬ ‫األطروحة‬ ‫أدا‬ ‫ً‬ ‫ء‬‫متفوقًا‬ ‫لنماذج‬ ‫‪LSTM‬‬ ‫على‬ ‫الشبكات‬ ‫العصبية‬ ‫ا‬ ‫ل‬‫تكرارية‬ ‫البسيطة‬‫(‬ ‫‪RNNs‬‬‫)‬ ‫التقليدية‬ ‫في‬ ‫التنبؤ‬ ‫بمواقع‬ ‫المركبات‬ ‫بدقة‬ ‫عالية‪.‬‬ ‫هذه‬ ‫التنبؤات‬ ‫ضرو‬ ‫رية‬‫لتح‬ ‫ديد‬ ‫المواقع‬‫المثلى‬ ‫لوحدات‬ ‫‪MEC‬‬ ‫‪،‬‬‫التي‬ ‫يتم‬ ‫استضافتها‬ ‫على‬ ‫أو‬ ‫بالقرب‬ ‫من‬ ‫هوائيات‬ ‫الجيل‬ ‫ال‬ ‫خامس‬‫لتحسين‬ ‫معدالت‬‫نقل‬ ‫البيانات‬ ‫وتقليل‬ ‫تأخيرات‬ ‫االتصال‪.‬‬ ‫من‬ ‫خالل‬ ‫تجميع‬ ‫المواقع‬ ‫ا‬ ‫لمتوقعة‪،‬‬‫تقد‬ ‫م‬‫الد‬ ‫راسة‬‫إطا‬ ‫ً‬ ‫ر‬‫ا‬ ‫لوضع‬‫‪MEC‬‬ ‫بكفاءة‪،‬‬ ‫مما‬ ‫يسهم‬ ‫في‬ ‫تقليل‬ ‫ازدحام‬ ‫الشبكة‬ ‫وتعزيز‬ ‫إدارة‬ ‫البنية‬ ‫التحتية‬ ‫ال‬ ‫حضري‬‫ة‪.‬‬‫ال‬ ‫تس‬ ‫لط‬ ‫ض‬‫ا‬ ‫هذه‬‫العمل‬ ‫الضوء‬ ‫فقط‬ ‫على‬ ‫قدرات‬ ‫نماذج‬ ‫‪LSTM‬‬ ‫في‬ ‫تحليل‬ ‫البيانات‬ ‫الزمنية‬ ‫الحقيقية‬ ‫ولك‬ ‫نها‬‫تؤكد‬ ‫أي‬ ‫ً‬ ‫على‬‫إمكانياتها‬ ‫في‬ ‫دعم‬ ‫البنية‬ ‫التحتية‬ ‫لش‬ ‫بكات‬‫الجيل‬ ‫الالحق‬ ‫الالسلكية‬ ‫في‬ ‫البيئات‬ ‫ا‬ ‫لحضري‬‫ة‪.‬‬-
dc.description.sponsorshipDepartment of Computer Science and Information Technologiesen_US
dc.language.isoenen_US
dc.publisherUNIVERSITE KASDI MERBAH OUARGLAen_US
dc.subjectLong Short-Term Memory (LSTM)en_US
dc.subjectMobile Edge Computing (MEC)en_US
dc.subjectdataseten_US
dc.subjectSimple Recurrent Neural Networks (RNNs)en_US
dc.titleOptimizing MEC Deployment Using Vehicle Density: A Deep Learning Approachen_US
dc.typeThesisen_US
Appears in Collections:Département d'informatique et technologie de l'information - Master

Files in This Item:
File Description SizeFormat 
BOUSSAID_DJOUDI_REGUIG.pdfIndustrial Computer Science4,42 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.