LEE- Bilgi ve Haberleşme Mühendisliği Lisansüstü Programı

Bu topluluk için Kalıcı Uri

http://hdl.handle.net/11527/19200

Gözat

Başlık ile LEE- Bilgi ve Haberleşme Mühendisliği Lisansüstü Programı'a göz atma
  • Öge
    Efficient super-resolution and MR image reconstruction networks
    (Graduate School, 2023-01-30) Ekinci, Dursun Ali ; Töreyin, Behçet Uğur ; 708191011 ; Information and Communication Engineering
    The process of generating a high-resolution image from a low-resolution image is known as image super-resolution. From a low-resolution image, many high-resolution images can be produced. Therefore, super-resolution is a difficult problem with ill-posed nature. In recent years, many deep neural networks were suggested to retrieve missing high-frequency features. Models started to get deeper to enhance performance. But, using these models in devices with limited resources is challenging due to their high computational complexity and memory requirement. Therefore, in this thesis, different knowledge distillation schemes are suggested to compress super-resolution networks. Offline- and self-distillation frameworks are used to decrease the number of repeating residual blocks in SRResNet and EDSR networks. Test results are obtained on benchmark datasets for scale factor four. After many experiments, results show that previous layers learn to produce similar outputs to following layers. However, when redundant layers are removed, performance of the compressed networks are less than their vanilla trained versions. Therefore, further study on this subject is required to prevent performance decrease. Magnetic resonance imaging is a valuable tool in medicine to identify diseases. To obtain a high quality image, enough $k$-space data is needed. This increases the necessary scan time. As in other image processing fields, deep neural networks are also used in magnetic resonance image reconstruction task from undersampled data. Since super-resolution also aims to restore missing information, using some concepts from super-resolution can help improve reconstruction performance. In this study, Iterative Up and Down Network is proposed to solve this problem. Network benefits from iterative up and down sampling framework and multiple scale factors. Training of the network is done using fastMRI dataset. Test results are obtained on two datasets which are fastMRI and IXI. Proposed network has processing units and test results show that increasing the number of these units improve the performance of the network. Also, using multiple scale factors further increased the performance. Quantitative results show that suggested approach is superior than some well-known state-of-the-art networks. When qualitatively compared to other methods, suggested model performs favorably.
  • Öge
    Group authentication and its application
    (Graduate School, 2023-03-17) Alimadadnezhad, Sevda ; Özdemir, Enver ; 708201006 ; Information and Communications Engineering
    This Thesis is about Group Authentication and Its Application. One of the most important subjects of secure communication is authentication. The data exchange process among communicating parties starts only after handling identification of each party. By increasing the usage of digital communication and private data transmission, authenticating a big amount of parties at the same time has an important role to have a secure and confident connection. The development of new technology in several fields forces various new devices to use public Internet. For example, autonomous vehicles are expected to compromise large part of communication system in the near future. The unmanned drones are already a part of surveillance and delivery systems. Most importantly, the internet service providers exploit the satellite technology to provide infrastructure for its users to access Internet while absence of base stations. In all these examples, the communicating parties do not necessarily perform identity confirmation. Group authentication provides this opportunity to authenticate a big part of a group at the same time. In this thesis it has been analyzed and evaluated the previous researches in group authentication field and the methods used for this purpose. Group membership authentication is one of the common method in group authentication which confirms the membership of a subset of a predetermined group by a single operation at the same time to figure out if all are members of the group or there is any nonmember. During the group authentication all legal members are allowed to establish a common key which is secret to the group. A trusted authority called group manager is responsible to generate tokens and distribute to each user. In computational side of these researches mathematical theorems and methods has been used; The basic theorem is Newton interpolation theorem which has brought many theorems and corollaries behind like Lagrange interpolation theorem. In addition, Elliptic curves method has been used in order to make the key transferring between the users secure. Even though the first group authentication method appeared in 2014, some other applications have been implicitly employed group authentication like methods as their security measurements. For example, a group like authentication scheme in autonomous vehicle realm is being called a batch authentication and in federated learning applications some researchers call it secure multi-party computation and it is called group authentication in wireless networks in order to authenticate the group users at the same time, but finally all mentioned terms are discussed as group authentication in this thesis. As an improving technologies and increasing the users, malicious usage of these technologies are growing as well. impersonating or fake data transferring through a large group of users may cause several serious problems in peoples life and also in the system which is used; On the other hand, authenticating a large group of users one by one is practically useless. Group authentication is going to become a necessary method for systems with large membership because of the ability of authenticating a group at the same time and using much more less storage, energy and time than the previous methods.
  • Öge
    Kesirli kalkülüs ve derin değerlendirme yaklaşımı ile G-8 ülkeleri ve Türkiye için ekonomik verilerin modellenmesi, etki faktörlerinin analizi ve öngörü çalışması
    (Lisansüstü Eğitim Enstitüsü, 2022-09-21) Önal Tuğrul, Nisa Özge ; Karaçuha, Ertuğrul ; 708182004 ; Bilgi ve Haberleşme Mühendisliği
    Temelleri 17. yüzyılda atılmış olan kesirli analiz, türev ve integral operatörlerinin gerçel veya karmaşık sayı kuvvetlerini tanımlayan bir matematiksel analiz dalıdır. Kesirli türevin tanımından kaynaklanan bazı eklemeler sayesinde fiziksel gerçeklere daha uygun olduğu tespit edilmiştir. Kesirli kalkülüsün hafıza özelliği ile ekonomik veriler üzerinde diğer yöntemlere göre daha anlamlı sonuçlar elde edileceği düşünülmektedir. Böylece bu tez çalışmasında, kesirli kalkülüs ve en küçük kareler metodu kullanılarak derin değerlendirme metodolojisi ile geliştirilmiş tek fonksiyon girdili ve çok fonksiyon girdili olmak üzere iki matematiksel yaklaşım önerilmektedir. Bu matematiksel yaklaşımların kodlanması ve uygulaması Matlab üzerinde yapılıp, ülkelerin ekonomik verileri kullanılarak modelleme, etki faktörü analizi ve öngörü çalışması yapılmıştır. Tek fonksiyon girdili derin değerlendirme yaklaşımında girdi faktörünün geçmiş değerleri dikkate alınırken, çok fonksiyon girdili derin değerlendirme metodolojisi ile bir faktörü etkileyen diğer faktörler ve geçmiş verileri de sisteme katılmaktadır. Bu yaklaşımlar ile modelleme, etki analizi ve öngörü metodolojileri geliştirilmiştir. Tek fonksiyon girdili türevli derin değerlendirme yaklaşımı (DAM) ile Amerika, Brezilya, Çin, Hindistan, İtalya, İspanya, İngiltere, Japonya, Türkiye ve Avrupa Birliği için 1960-2018 yıllarına ait Kişi Başına Gayrisafi Yurtiçi Hasıla (GSYİH) verileri ile modelleme ve öngörü çalışması yapılmıştır. Uygulama sonuçlarını inceleyecek olursak en düşük modelleme hatası % 0.81 ile Amerika için elde edilirken, en yüksek modelleme hatası % 7.26 olarak Brezilya'da elde edilmiştir. En düşük test hatası % 0.056 ile İspanya'da görülmüş olup, en yüksek test hatası İngiltere'de %0.91 olarak gerçekleşmiştir. Çizelge 4.3'te ise 2019 için Kişi başına GSYİH değerleri için öngörü yapılmıştır. Çok fonksiyon girdili derin değerlendirme yaklaşımı (M-DAM) ile G-8 ülkeleri ve Türkiye'nin 2000-2019 yıllarına ait ekonomik faktör verileri (Cari Hesap Dengesi (% GSYİH), İhracat (% GSYİH), GSYİH Büyüme Oranı (yıllık %), Tasarruflar (% GSYİH), Yatırımlar (% GSYİH), İthalat (% GSYİH), Enflasyon (% yıllık), Faiz Oranı (Gecelik) ve İşsizlik Oranı) kullanılarak modelleme, etki faktörü analizi ve öngörü çalışması yapılmıştır. İlk olarak modelleme sonuçlarını değerlendirecek olursak, ekonomik faktörlerin modellenmesine yönelik 81 faktör için yapılan modellemede maksimum hata Fransa'nın Enflasyon faktöründe %0.0003112 olarak bulunmuştur. Öngörüye ilişkin olarak da toplam 162 adet öngörü için hatanın %10 dan küçük olduğu oran %75.4 hatanın %10 dan büyük olduğu oran da %24.6 bulunmuştur. Ayrıca 2018 ve 2019 yılları için öngörülebilirlik sınıflamasında, en öngörülebilir ülke 1. sırada Almanya 2. sırada Kanada, Fransa, UK, USA, 3. sırada İtalya ve Japonya, 4. sırada Rusya ve en son sırada (5.) Türkiye'nin olduğu tespit edilmiştir. Faktörlere diğer faktörlerin ağırlıklı ortalama etkisi bakımından 2018-2019 yılları ve geçmişe yönelik 3,5 ve 10 yıl için açıklamış olduğumuz varsayım çerçevesinde güçlü pozitif ve güçlü negatif etkiler değerlendirilmiştir. Ülkelerin ekonomik faktörlerinin etkilerinden bazı öne çıkan ortak veya farklı etki durumlardan bahsedecek olursak; Yatırımların etki ettiği faktörleri incelediğimizde; Kanada, Fransa, Almanya, İtalya, Türkiye ve UK'de ihracatı güçlü pozitif olarak etkilemiştir. Japonya'da orta derecede pozitif, Rusya ve USA'da ise zayıf pozitif olarak etki etmiştir. İthalat; Fransa, Almanya, İtalya, Japonya, Rusya, UK ve USA olmak üzere 7 ülkede ihracata güçlü pozitif, Kanada'da orta derecede pozitif olarak etki etmiştir. Bunun aksine Türkiye'de ihracatı zayıf negatif olarak etkilemiştir. Faiz; Almanya ve Rusya'da ihracatı güçlü pozitif olarak etkilemiştir. Türkiye, UK ve USA'da tasarruflar faizi güçlü pozitif olarak etki etmiştir. Enflasyon ihracat ilişkisi incelendiğinde sadece Türkiye'de enflasyon ihracata güçlü pozitif etki ederken, Rusya'da etkisi görülmemiştir ve diğer yedi ülkede negatif etkilerde bulunmuştur. Tasarruflar Kanada'da faiz oranlarına güçlü negatif olarak etki ederken, Fransa, İtalya, Türkiye, UK ve USA'da güçlü pozitif, Japonya'da orta derecede pozitif, Almanya ve Rusya'da zayıf pozitif olarak etki etmiştir. Gelecek çalışmalarda geliştirdiğimiz çok girişli değerlendirme sistemi (M-DAM) modelimize fonksiyon türevleri de katılarak geliştirilmesi hedeflenmektedir.
  • Öge
    New RF energy harvesting models for next-generation wireless communication systems
    (Graduate School, 2022-11-09) Babaei, Mohammadreza ; Durak Ata, Lütfiye ; 708182003 ; Information and Communications Engineering
    Deployment of massive sensor and Internet of Things (IoT) devices in next-generation wireless communication systems reveals energy limitations as one of the challenges. Typically, IoT devices are powered by a battery, which restricts their capacity and working time. Energy harvesting (EH) has been regarded as a promising approach which can increase the life-time of a wireless communication system. In EH, energy is obtained by wind, solar, vibration, etc. Thus, the harvested energy is transformed into electricity and can be used by the desired nodes. However, the aforementioned traditional EH methods are not always available. Additionally, radio frequency (RF) EH has emerged as a key promising technique that enables wireless systems to harvest energy from the incoming signals in the environment. This energy is available as dedicated or ambient energy and can be harvested throughout the whole day. Hence, RF EH can be an effective alternative for empowering battery-free IoT devices, resulting in increased operational time. As RF EH methods for power-constrained nodes, simultaneous wireless information and power transfer (SWIPT) and wireless-powered communication (WPC) schemes are studied in the literature. The SWIPT scheme employs two EH receiver structures: power-splitting (PS) and time-switching (TS). The power-constraint node harvests power from the incoming signal energy in PS EH mode, with one portion of the signal power used for harvesting energy and the remainder for information processing. On the other hand, in the TS EH mode, EH and information processing (IP) are allocated to two non-overlapping time intervals, respectively. Furthermore, the amount of harvested energy is regarded as a linear or nonlinear (NL) function of the input power of the energy receiving (ER) node. The input and output powers of the EH circuit are directly proportional in the linear EH model. For the region with low input power, the NL EH model provides the same amount of energy as the linear model. However, the amount of harvested power saturates to a predefined threshold power at the high input power of the EH circuit. In this thesis, three different RF EH system models are investigated and the closed-form analytical expressions for their performances are derived. Moreover, the performance of each considered system is investigated in terms of bit error probability (BEP) and outage probability. Besides, theoretical derivation results are provided for different system parameters and supported by the Monte-Carlo simulation results. Comprehensive insights into the studied systems are provided, which broaden the view of engineers toward the EH system designs. The first section investigates the bit error rate (BER) performance of a full-duplex (FD) overlay cognitive radio (CR) network with linear/NL EH capability. The studied overlay CR network comprises of a primary transmitter/receiver (PT/PR) pair, a secondary transmitter (ST), and a secondary receiver (SR) operating in FD mode, whereby ST harvests energy from both PT and SR during the first communication time slot. In the second time slot, SR receives its signal from ST while PT sends its signal to PR. The BEP expressions for the primary/secondary users (PU/SU) are obtained analytically and verified through Monte-Carlo simulations using both linear and NL EH models at ST. Additionally, to determine the trade-off between EH and IP, corresponding system performances are evaluated with regard to a power allocation coefficient at SR. The results demonstrate that, in contrast to the non-cooperative (direct transmission) case, the proposed FD-CR system with NL EH improves the PU BEP performance. Besides, SU benefits from the licensed spectrum of PU with significant BER performance. In the second part of the thesis, the performance of a wireless-powered, two-hop, amplify-and-forward relaying system is studied when there is no direct link between the source and the destination. The power-constrained source and relay receive energy from a dedicated power beacon (PB) that broadcasts an energy-bearing signal. For both linear and NL energy harvesting models, theoretical derivations of BEP, outage probability, and throughput expressions are performed. Additionally, Monte Carlo simulations are performed to verify the theoretical results that are presented for various system parameters. The results show how the realistic NL EH model is different from the traditional linear EH model, which overestimates the performance of the system when a large amount of energy is harvested. This results in a misunderstanding of the actual performance of EH systems. However, both models operate similarly and provide appropriate results at low levels of harvested energy. In the last part of this dissertation, the performance of the proposed two novel NL EH models is analyzed in terms of average harvested power, throughput, and BEP. The system comprises a single multi-antenna power-constraint source that transmits its signal to a destination with multiple antennas while harvesting power from a dedicated PB. For a comprehensive analysis of the system, closed-form expressions are derived for Nakagami-$m$ fading channels and the special case of Rayleigh channels. In addition, for existing NL EH models, the simulation results are obtained using the Monte-Carlo method. The results provide a broader picture of EH systems and comprehensively compare the proposed NL EH models to linear, piece-wise linear, and NL EH models available in the literature. As a result, these provide better perspectives on the analysis and design of EH systems.