|Rule Placement in Software Defined Networking with Fixed Prefix Lengths
|Ahmed Hadi Hamzah Alghazali
|Computer Engineering, Network
|The separation of the control plane and data plane is the main idea behind software-defined networking. The data plane, or the switches, has only one table and packets are sent to the destination based on information of this table. The controller is also responsible for determining rules and distributing them into switches. One of the problems in SDN is that switches do not have sufficient memory to keep all these rules. In the initial design of these networks, when a switch receives a packet which does not have its rule, the packet is transmitted to the controller so that controller decides about the packet. In most cases, the controller adds a new rule to multiple switches of the path so that subsequent flow packets are identified and routed by the switches. However, this solution is not scalable in large networks since the controller is acting as a critical point if it fails the whole network may collapse. Additionally, if there are many new data streams in the network (such as data centers), the controller cannot respond to a large volume of requests. In this thesis, the proposed approach is based on the distribution of rules into switches so the controller is only used to enter general policies and distribute them into the network. When an unknown packet enters the network, the switches themselves send it to the next switches that are likely to identify the packet according to predefined rules. These methods are mainly different in how the rules are distributed into switches and deciding where to send the unknown packet. We first consider multiple fixed lengths for the prefixes based on the network load and the number of rules, then divide the rules and distribute them into the switches. As the length of the prefix is greater, the number of rules will also increase, and the switch may not have enough storage for saving them. We provide a solution to overcome this problem and determine how rules with fixed prefix are distributed into switches. Finally, we propose a method for determining the precise length of the prefixes which leads to load balancing among the switches. A dynamic algorithm can be provided for networks using this method which dynamically distributes rules between switches based on their current loads
|Investigation of gassing behavior, electric and dielectric properties of different insulating fluids
|Prof. Asghar Akbari Azirani
|High Voltage Engineering
|In the field of the condition monitoring of electrical equipment and systems, the knowledge of aging behavior and investigation of the decomposition products of the insulating materials are essential. Due to that background the following master thesis task is assigned to Mr. Ansari: First of all, dissolved gas analysis method have to be done for some service aged Silicone and Midel 7131 samples that were used in some distribution power transformers. Along with DGA method the AC breakdown voltage at different temperatures ( ϑ =RT, 40oC, 60oC, 90oC and 110oC), the dielectric loss factor, the specific conductivity and the relative permittivity at ϑ =20oC and ϑ =90oC should be performed for mentioned specimens. Furthermore, the surface tension, density and partial discharge inception and extinction voltage at room temperature have to done as complementary diagnostic methods to obtain more information about insulation state. Finally, all of the mentioned measurements should be applied for new Silicone and Midel7131. Then, the obtained results should be compared with aged ones at the same absolute water content. The measured results have to be presented in the usual way and explained in their theoretical background. In addition to the observed changes as a function of the breakdown quantity, water content, temperature, as well as the fluid type, conclusions for the reasons of these changes have to be drawn. The elaboration has to follow the existing guidelines. The submitted copies remain property of the institute.
|Modeling of Inverter for stability analysis
|Prof. Asghar Akbari Azirani
|Using renewable energy sources is growing everywhere in the world. Comparing to conventional power generation, these kinds of sources are distinguished by their clean and sustainable nature, low voltage and low power ratings, being in the vicinity of end users and having higher reliability. Most of these new emerging technologies need to be connected to low voltage grids via interface modules. Interface modules play an important role by making the output of these renewable sources compatible with grid side requirements. Inverter interface sources, where a DC and AC generation is supposed to be connected to the Ac grid, are well-known among other electronic converters. Since in the future distributed grids, named micro grids, inverters are embedded besides other apparatus, so studying the control philosophy of such equipment will be a substantial step toward analyzing the whole grids. Inverters can be modeled with three general schemes, ideal model, average model and exact model, each of which has a specific usage depending on the application. Inverters can be operated under two different control strategies, P-Q control and P-U control but in this work, mainly, the P-Q control strategy of inverters is investigated with taking external active power set-point into account. All mentioned models are tested under P-Q control loop with two different softwares, Matlab and PSCAD, and the result are shown and compared. This Master thesis tries to find and introduce an appropriate model of inverter that can be used instead of exact model in power systems analysis, particularly stability issues. Finding the state space representation of inverter average model, including closed loop P-Q control, is the next step toward modeling of an inverter. Then by using the state space equations, M-file of whole system is written and tested with different inputs. It will be shown that the state space matrix is a kind of nonlinear matrix that can be simplified by a significant assumption, by removing nonlinear part, fast and accurate responses is achieved, compared to nonlinear states equations. At the end, a comparison between different models is presented. This work aims to prove that Inverter Average model can be replaced by Inverter exact model in power system stability studies.
|Design and Implementation of Impedance Control on Dual-user System for Eye Surgery Training
|Prof. Hamid D. Taghirad
|Systems and Control Engineering
|Haptic devices can be used to set the motor skills of a novice surgeons in order to train them on how to perform cataract or vitrectomy eye surgeries. The ARASH:ASiST haptic system facilitates the procedure of surgery training by involving the expert and novice physicians in the process and providing the appropriate haptic feedback. The proposed framework for training is composed of two sections. First, the surgery is directly conducted by the expert surgeon. At the same time, the novice surgeon receives these correct movements through the haptic system. It helps the novice surgeon by increasing his/her level of competence in different tasks by learning the expert’s maneuvers. Second, the novice surgeon performs the surgery with the supervision of the expert. The proposed training framework allows the expert surgeon to intervene immediately into the process as many times as needed in order to avoid undesired complications. This intervention is due to the sudden mistakes happens by the trainee. Due to the interaction between two users and the environment and because of the exerted forces importance, a novel robust impedance scheme is proposed for utilizing in its control system. The stability analysis is presented by considering the closed-loop stability of the nonlinear system. The experimental results are implemented on the ARASH:ASiST haptic device, which is specially designed for vitrectomy surgery. Altogether, the main contribution of this thesis is to design a robust switching gain impedance controller for medical training purposes with a mechanical pedal mechanism to switch the dominant user during the procedure. The simulation and experimental results confirm the effectiveness of the proposed control algorithm in enhancing surgery training quality level.Keywords: Impedance Control, Interaction Control, Dual User Haptic Devices, Surgery Training, Stability Analysis, Vitrectomy, Eye Surgery
|Simulation and numerical solution of Schrodinger equation for anisotropic crystals with relativistic potential
|Dr. Farshid raissi
|Electrical and Computer Engineering
|The ?rst-principle density-functional description of the electronic structures of the high-Tc cuprates has failed to predict the peculiar occurrence of anisotropic band gap as well as insulating behavior at low impurity doping levels for these materials.In this work, we propose a new potential to be incorporated into quantum mechanical description, which leads to a band gap in electronic band structure reminiscent of what is observed in experiments. Also the insulating behavior at low doping levels is readily explained with this potential, which is created due to principles of special relativity. The principles of relativity claim that the potential around a moving charge is different from a stationary charge in that it is no longer spherically symmetric. It increases in the directions perpendicular to the direction of motion while decreasing along the charge’s path of motion.In a solid where electrons are moving and nuclei are stationary, this leads to an uncompensated angular dependent potential that is attractive for electrons in their direction of motion and repulsive in other directions. Aggregate repulsive and attractive forces are created and act on electrons only if they have preferred directions of motion or in other words if the material is anisotropic. Here, we solve Schrodinger equation numerically in real space for a 2D hexagonal lattice, which is the lattice structure of high Tc
materials, incorporating the above mentioned potential. The band structure for 2D hexagonal 7 lattice and simple cubic lattice, have been calculated for classic and relativistic coulomb potentials that are proposed here. The obtained band diagrams, with our proposed potential, can justify experiments quite well. Key words: Schrodinger equations, Band structure, Anisotropic materials, high Tc Cuprates, principles of relativity, coulomb potential, numerical solution
|Functional and structural joint modeling of age-related changes in adult brain cortex using EEG data and MR images
|Sahar Rahimi Malakshan
|Prof. Hamid Abrishami Moghaddam
|Biomedical Engineering, Bio-Electrical Engineering
|Age-related changes in the human brain can be investigated from either structural or functional perspectives. Analysis of structural and functional age-related changes throughout the lifespan may help to understand the normal brain development process and monitor the structural and functional pathology of the brain. This study, combining dedicated electroencephalography (EEG) and magnetic resonance imaging (MRI) approaches in adults (20-78 years), highlights the complex relationship between micro/macrostructural properties and the functional responses to visual stimuli. Here, we aimed to relate age-related changes of the latency of visual evoked potentials (VEPs) to micro/macrostructural indexes and find any correlation between micro/macrostructural features, as well. We studied age-related structural changes in the brain, by using the MRI and diffusion-weighted imaging (DWI) as preferred imaging methods for extracting brain macrostructural parameters such as the cortical thickness, surface area, folding and curvature index, gray matter volume, and microstructural parameters such as mean diffusivity (MD), radial diffusivity (RD) and axial diffusivity (AD) and fractional anisotropy (FA). All the mentioned features except for FA were significantly correlated with age in V1 and V2 regions of the visual cortex. Furthermore, we highlighted, negative correlations between structural features extracted from T1-weighted images and DWI. The latency and amplitude of the three dominants peaks (C1, P1, N1) of the VEP were considered the brain functional features to be examined for correlation with age and structural features of the corresponding age. We observed significant correlations between mean C1 latency and gray matter volume averaged in V1 and V2. In hierarchical models, the structural index not contributed to significant additional variance in the C1 latency after accounting for the variance associated with age. However, the age explained significant additional variance in the model after accounting for the variance associated with the structural feature. Keywords: Age-related changes, functional analysis, structural analysis, joint structural and functional analysis, Visual evoked responses, Magnetic resonance imaging, Diffusion tensor imaging
|Detecting Influential Nodes in Partially Observable Complex Networks using Statistical Methods
|Dr. Seyed Hossein Khasteh
|Artificial Intelligence and Robotics
|Finding the most influential spreaders in a viral marketing campaign is generalized as an optimization framework and named as the problem of Influence Maximization. It has been proven that this problem is an NP-Hard problem under some special circumstances. However, it has also been shown that a simple greedy algorithm can solve it with a reasonable accuracy. Each year a number of new articles have been published using the same influence maximization phrase which the aim of all of them is to find the most critical spreader under some specific diffusion model. In our work we have extended the current methods by making them applicable to partially observable networks. A partially observable network is an induced graph from a phenomenon which we do not have the whole picture of the graph. Some ties which exist in the phenomenon are not present in the model. In other words, we have missing links in our graph-based model. We have combined existing influence maximization methods with a generative network model named Exponential Random Graph Models (ERGM) in order to handle these situations. With these models we construct a probabilistic models of missing network ties. More precisely, we have implemented a link prediction using the aforementioned generative methods prior to the actual run of influence maximization methods. The intuition was that there are important network dynamics in the non-observable sectors of the graph that are ignored in the previous works. Our aim is to replace this total ignorance with a heuristic to add more intuition into Influence Maximization algorithms. Experimental results over two real-world datasets have proved the effectiveness of our method. Keywords: Influence Maximization, Partially Observable Networks, Exponential Random Graph Model, ERGM
|Development, Relationship and Test Charge (SoC) and State Of Health (SoH) Monitoring System for Valve Regulated Lead-Acid (VRLA) Batteries
|Seyyed Kamyar Makinejad
|Dr. Mohammad Tavakoli bina
|Power System Controlled Islanding Considering Transient Stability and Uncertainties
|Electrical Engineering, Systems and Control
|The reliable, secure and stable operation of a power system is a top priority in modern societies. Power systems are exposed to different electric faults. Some severe electric faults can propagate throughout the power system and initiate a cascading outage. Widespread or regional blackouts are caused by propagating and uncontrolled cascading failures. Different emergency controls and system protection schemes (SPS) are utilized to stop the propagation of blackouts or cascading failures. During cascading outages, coherent groups of generators are formed and the harmful dynamics can propagate from one area to another one, due to their weak electrical conditions. In order to mitigate or minimize the effects of blackouts, the intentional splitting or controlled islanding is considered as a last resort.The aim of controlled islanding is to split the whole interconnected power system into smaller electric islands, while the steady state and stability requirements in resulted islands are fulfilled with minimum load and generation changes. However, the mechanism of controlled islanding scheme including the time and location of the network splitting is a major challenge. There are mature and comprehensive islanding algorithms that address the preservation of steady state operational and topological constraints in islanding problems, however considering stability phenomenon in islanding models is still a major gap. The most threatening phenomenon right after the network splitting is the transient instability of synchronous generators. The focus of this thesis is to determine the islanding boundary (i.e. where to island issue) under the transient stability constraint. While the steady-state operational constraints such as power balance in resulted islands are of great importance, a major prerequisite for the success of any controlled islanding scheme is the preservation of transient stability constraint. In this thesis, transient stability constrained network splitting is realized using two different approaches. First, a controlled islanding model, named by Method 1 in this thesis, is proposed to ensure and improve the transient stability of the islanded system. Linear transient stability constraints are derived off-line, based on the extended equal area criterion, to ensure the first swing transient stability of the synchronous machines, right after the controlled line switching. The islanding model with transient stability constraints is first developed as a XII mixed-integer nonlinear programming (MINLP) optimization model. Further, the MINLP model is linearized, resulting in a computationally lighter mixed-integer linear programming (MILP) model. In the second part, based on the transient energy approach, two methods named by Method2 and Method3 in this thesis, are presented. The transient stability constrained islanding models are perfected via two-stage models. First, the conventional controlled islanding problem is formulated as a MILP optimization model with considering operational, coherency and linear AC load flow constraints. The boundary of each island is determined using an optimization model aiming at minimizing the total power imbalance. To consider the transient stability, the network splitting plan obtained from the MIP model is then evaluated in the second stage using a proper transient stability assessment criteria based on the transient energy function method. In the second stage, to satisfy the transient stability constraint of the critical island, a linear constraint is constructed and added to the MILP formulation of a controlled islanding model. Saddle or control unstable equilibrium points (CUEP) are determined using an optimization model. Finally, the proper islanding method is selected and then revised to consider the uncertainty of the controlled islanding problem. All the proposed network splitting models are simulated over the dynamic IEEE 118-bus system. In addition to the discussion given over the simulation results of each islanding method, a comprehensive comparative analysis is proposed to investigate the importance of each model.
|The Influence of Nano Filler TiO2 on Electrical and Dielectric Characteristics of Outdoor Epoxy Resin
|. M. M. Saei Shirazi
|Recent developments in polymeric dielectric nano composites have shown that these novel materials can enhance high voltage components and systems design. Some of the enhancements can be listed as reduction in size, better reliability, high energy density, voltage endurance, and multi-functionality. Nano dielectric systems demonstrated specific improvements that have been published in the literature by different groups working with electrical insulation materials. In this master thesis, the influences of titanium dioxide (TiO2) nano filler on the electrical and dielectric characteristics of an outdoor epoxy resin system are investigated. For this purpose, two different types of specimens with different amounts of nano filler (0 wt.%, 1 wt.%, 3 wt.%, 5 wt.%, and 7 wt.%); needle and needle-plate were applied. In order to produce the specimens, the method which is proposed by the material manufacturing company was inspected and through a try and error trial via changing the temperature and time of casting and mixing, the optimal approach for producing the proper specimens was acquired. Since the proper distribution of the component of specimens such as micro and nano fillers was of high importance in our experiment and would highly affect the consequences, the specimens were examined by a reflection electron microscope (REM). The REM images illustrated that the nano and micro filler were properly distributed. To inspect the effects of nano filler on electrical and dielectric characteristics, the partial discharge, breakdown voltage, D.C. volume resistivity and the loss dissipation factor measurements were applied at different temperatures and also temperature changing condition. The results demonstrated that as the amount of nano filler increased from 0 wt.% up to 7 wt.%, some of the electrical and dielectric characteristics became reinforced, some fainted and some remained unchanged. The D.C. volume resistivity remained unchanged by modifying the amount of nano filler TiO2 between 0 wt.% and 7 wt.%. The breakdown voltage didn't change either. The consequences of the experiment also supported the fact that, the results of partial discharge and tan ? measurements varied for different amounts of nano filler. Although the behaviour of electrical and dielectric characteristics varied for different amounts of nano filler, the total results of 5 wt.% nano filler TiO2 seems to be better than other ones.
|Integration of Augmented Reality and Internet of Things: Possibilities for Helping Patients with Alzheimer’s Disease
|Fatemeh Ghorbani Lohesara
|Dr. Mehdi Delrobaei
|Electrical Engineering (Mechatronics Engineering)
|Independent life of the individuals suffering from Alzheimer’s disease (AD) is compromised due to their memory loss. As a result, they depend on others to help them lead their daily life. In this situation, either the family members or the caregivers offer their help; they attach notes on every single object or take out the contents of a drawer to make those visible when they leave the patient alone. The aim of this thesis is to provide multi-level support and some helping means for AD patients and their family members through the integration of existing science and methods. This study reports results on an intelligent assistive (IA) system, achieved through the integration of Internet of Things (IoT), augmented reality (AR), and adaptive fuzzy decisionmaking methods. The proposed system has four main components; (1) a location and heading data stored in the local fog layer, (2) an AR device to make interactions with the AD patient, (3) a supervisory decision-maker to handle the direct and environmental interactions with the patient, (4) and a user interface for family or caregivers to monitor the patient’s real-time situation and send reminders once required. The system operates in different modes, including automated and semi-automated. The first one helps the user complete the activities in their daily life by showing AR messages or making automatic changes. The second one allows manual changes after the real-time assessment of the user’s cognitive state based on the AR game score. We provide further evidence that the accuracy, reliability and
response time of the IA system are appropriate to be implemented in AD patients’ homes. Moreover, the system response in the semi-automated mode
causes less data loss than the automated mode, as the number of active devices decreases. We have also found that playing an audio message instead of displaying an image message has better performance and less battery consumption. Keywords: Alzheimer’s disease, Augmented reality, Fuzzy decision-making, Intelligent assistive system, Internet of Things.
|Day--Ahead Unit Commitment in Microgrid by Considering Electric Vehicles' Battery Swap Stations
|Professor Masoud Aliakbar Golkar
|Electrical Engineering-Power System
|Due to new environmental politics, electric vehicles (EVs) should catch their popularity in the nearest future. Obviously, the most challenging item beyond Electric Vehicles (EVs) is their refueling process. Low-speed charging is undesirable from users' point of view, on the other hand, high-speed charging is a worrying factor in distribution system sector. It is expected that Battery Swap Station (BSS) will compromise the aforementioned charging strategies and will lead to more feasibility for EVs. BSS presence in electricity distribution system is still a questionable subject. Within the above context, the motivation factors for this thesis supported by issues related to managing BSS nearby other components of distribution network (DN). The manuscript follow this goal in two phases. In phase 1, this research is established based on proposing the stochastic characteristic of battery swap station (BSS) for electric vehicles (EVs). The model can be applied in demand response programs. The historical data on the driving pattern as well as the time of starting and ending the daily conventional journeys and the travelled distance are utilized as input data. For creating more practical synthetic data the correlation between the aforementioned random variables is considered by using multivariate student's t copula function. Afterwards, a Monte Carlo simulation method is provided for extracting the uncertain scenarios of the hourly demand profile and the hourly residual capacity which enters BSS by discharged batteries. The demand profile is presented in discrete quantity regarding the packet format of energy exchange between customers and BSS. Also, the city traffic and station queue concerns are added for leading to more practicality in the model. To reduce the generated scenarios scale without losing their attributes a scenario reduction algorithm based on Kantorovich Distance (KD) method is proposed. The organized stochastic model can be used efficiently as a benchmark for operation problem that contains BSS. Eventually, the model performance is validated via a stochastic demand response program based on time of use pricing over various ix case studies for analyzing the model's key factors. The demand response contains BSS technical constraints in detail which can be used in distribution system operation application. In phase 2, the research concentrates on BSS management nearby other electrical distribution system sector components in microgrid structure. The detailed technical constraints of BSS operation are formulated. The research seeks to evaluate BSS enrolment, aiming at catching the most benefit from Micro-Grid System Operator (MGSO) aspect. A Stochastic Unit Commitment (SUC) program which includes Value of Lost Load (VoLL) caused by the vulnerability to the driving pattern in BSS structure is presented. The VoLL is measured in terms of Conditional Value of Lost Load (CVoLL) definition. CVoLL helps the SUC to handle the associated high cost of VoLL in some extreme low probability scenarios. The thesis evaluates two proposals for BSS by changing its ownership between the private and dependent on MGSO statement. The benefits and challenges of the two proposals will be discussed to demonstrate BSS impacts precisely. The research deals with BSS firstly as a new electricity retailer in the microgrid. Secondly, BSS is a novel unit which belongs to the MGSO for direct controlling in a two-stage hierarchical stochastic process. The results will give a primal idea of having BSS in distribution level of the electrical power system for managing EVs in microgrid structure. Index Terms- Battery Swap Station (BSS), Packet energy, Copula, Traffic and queue, Monte Carlo, Kantorovich Distance (KD), Stochastic demand response, Conditional Value of Lost Load (CVoLL), direct load control, retail electricity market, Stochastic Unit Commitment (SUC), two-stage hierarchical stochastic process.
|Simulation and investigation of online dryer of oil insulation for power transformers using zeolite and paper filter
|Prof. Asghar Akbari Azirani
|Electrical Engineering, Power
|Power transformer is one of the most important equipment of power system. Humidity is one of the most important factors to reduce the lifetime of the transformer. In addition to premature aging of paper and oil insulation in the transformer, this factor causes the formation of water in the oil, bubble production, intensification of the partial discharge phenomenon and mechanical pressure on the transformer windings. Therefore, the moisture in the transformer insulation system should be removed as much as possible. As the advancement of technology, itbecomes possible to build dryer systems to protect expensive equipment such as power transformers, to remove oilsoluble moisture directly from the insulation system and prevent unexpected events. In this dissertation, two online dryer systems with different characteristics such as membrane separation system and molecular sieving system are simulated in two-dimensional spaceusing COMSOL MULTIPHYSICS software. In the membrane system, the effect of parameters such as electric field intensity, temperature and flow rate on drying time and moisture content of oil tank was investigated. In the molecular sieve system, the effect of parameters such as temperature, flow rate, zeolite type, and the dimensions of the chamber were investigated. In the second step of each methods, the obtained results were optimized in COMSOL MULTIPHYSICS software by the optimization module. Finally, optimized parameters are suggested in pertinent section. Keywords: Online drying of transformer, Membrane separation, Molecular sieve, Power transformers.
|Analysis and Simulation of Anomalous Optical Effects in Metasurfaces Based on Dielectric Nanostructures
|Dr. Tavakol Pakizeh
|Electrical Engineering, Field and Wave, Communications
|The study of metasurfaces are attracted considerable interest due to their novel optical functionalities and unique optical phenomena resulting from their interaction withlight. Metasurfaceshaveenabledsomeanomalousopticaleffectswhichwasnot possible before. This thesis discusses some of these concepts and presents a theoretical framework for analysis of anomalous optical effects in optical metasurfaces. Based on current trend of scientific society in this field, we focused on metasurfaces composed of high-refractive-index nanoparticles. However, the theoretical method of this thesis is also applicable in plasmonic metasurfaces. The main contribution of this work is the general method for characterization of a metasurface which composed of nanoparticles that can be modeled with dipole approximation. We will showthatconventionalmethodsoffindingeffectiveparametersmaygiveunphysical results in high-index dielectric metasurfaces. The proposed method can be used to efficiently characterize metasurfaces with simultaneous electric and magneric dipole resonances, illuminated by a plane-wave at an arbitrary angle of incidence. To validate our results, we have also present the results of full-wave simulations and great agreement is observed. We make use of this model and study the anomalous optical effects, and in particular, generalized Brewster’s effect and anomalous Kerker’s effect. Keywords: Metasurfaces, High-index nanoparticles, Generalized Brewster’ effect, Generalized Kerker’ effect, Anomalous optical effects
|Design, Fabrication, and Pattern Optimization of Slotted Leaky-Wave Antennas by Using Modified Slots in Ka-Band
|Dr. Hadi Aliakbarian
|Electrical Engineering, Communications
|Leaky-wave antenna is a type of antenna that can radiate from its guiding structure because of the leakage of a traveling-wave through the slots. This antenna has a narrow beam which is dependent on the antenna’s size. These types of antennas can be used in high-frequency and microwave-frequency applications because of the simple feeding, high gain, and easy fabrication. Also, one of the advantages of these antennas is frequency scanning ability, which means that the beam angle of the antenna radiation pattern can be controlled by frequency changing. Long-slot antenna is a type of leaky-wave antenna that can be used in communication systems, radars, and aerospace applications. The long-slot antennas have some proper characteristics such as high power, low profile, narrow beam, and proper radiation pattern. Nowadays, there is extensive research into the fifthgeneration in the millimeter-wave band. Leaky-wave antennas can be useful for basestation antennas for the next generation, and long-slot leaky-wave antennas are attractive for the millimeter-wave band, which we will be discussed in this thesis. First, we will introduce the leaky-wave antennas, and then the radiation performances of these antennas will be analyzed. Then, we will simulate a new design of a symmetric long-slot antenna in order to improve the radiation pattern, in which the radiation null will be changed to a radiation peak by using a simple idea. As a result, the new antenna has 11.1 dBi realized gain and 1.4 GHz impedance beam-width. Finally, the fabricated antenna will be tested in the laboratory of K. N. Toosi University of Technology. This thesis shows a good agreement between the measured and simulated results. Keywords: Leaky-wave antenna, long-slot antenna, millimeter-wave antenna, and beam shaping.
|A proper topology for scalable software defined networks with hierarchical elastic table distribution in switches
|Mohammed yousif zakariya Alsaadi
|Computer Engineering, Network
|Basic change in network architecture of SDNs is to separate control and data planes. Data plane or switches have only tables based on which transmit the packets to the destination. Control plane also determines rules and distributed them between switch’s tables. One of the problems in software-defined networks is that switches do not have enough memory to store all rules. In the initial design of these networks, when a switch receives a packet which does not have the rules, the packet is transmitted to the controller so that controller decides about this packet. In most cases, the controller adds a new rule to multiple switches of the path so that subsequent flow packets are identified and routed by the switches. But this approach cannot be scaled to large flows because the existence of a controller is the critical point and if the whole network fails, it will be in trouble. Moreover, if there are a large number new data flows in the network (like data centers), a controller would not be able to respond to a large volume of requests. Several approaches have been proposed to solve the scalability problem, among which two of them can be mentioned: Rules are distributed among switches Second approach is to use multi-level networks. In this method, the network is divided into smaller regions and each small region has a controller. Usually, there is a main controller, which distributes the rules among local controllers. The approach proposed in this thesis is a combination of the above methods. In other words, rules are distributed among the switches. Because we believe that controller should only introduce general policies and distribute rules. In order to solve scalability and high latency of packets, switches are divided into smaller regions and a main switch called LS is selected to each region. This switch has all rules of a small region, local switches only have their specified rules. LSs operate as border gateways. In this method, network is divided to smaller areas to balance network load and switches. But since volume of network data flow decrease and increase dynamically in different areas, a second mechanism is also considered for dynamic load distribution. In this way, whenever load of switches increases suddenly, number of switches will also increase and high-level rules are divided among switches. To this end, set of local switches’ rules should also change so that local switches close to one of routers, guide unknown traffic towards the router so that focus is eliminated from one switch. When network load is reduced, number of router switches are also reduced and rules of local switches are adjusted again. This behavior is known as elasticity. Keywords:- Software-defined networking, Leader Switch, Aggregate rules, Multi-level.
|Predicting Depth from Semantic Segmentation using Game Engine Datase
|Mohammad Amin Kashi
|Hamid D. Taghirad
|Depth perception is fundamental for robots to understand the surrounding environment. As the view of cognitive neuroscience, visual depth perception methods are divided into three categories, namely binocular, active, and pictorial. The first two categories have been studied for decades in detail. However, research for the exploration of the third category is still in its infancy and has got momentum by the advent of deep learning methods in recent years. In cognitive neuroscience, it is known that pictorial depth perception mechanisms are dependent on the perception of seen objects. Inspired by this fact, in this thesis, we investigated the relation of perception of objects and depth estimation convolutional neural networks. For this purpose, we developed new network structures based on a simple depth estimation networkthatonlyusedasingleimageatitsinput. Ourproposedstructuresuseboth an image and a semantic label of the image as their input. We used semantic labels astheoutputofobjectperception. Theobtainedresultsofperformancecomparison between the developed network and original network showed that our novel structures can improve the performance of depth estimation by 52% of relative error of distanceintheexaminedcases. Mostoftheexperimentalstudieswerecarriedouton synthetic datasets that were generated by game engines to isolate the performance comparison from the effect of inaccurate depth and semantic labels of non-synthetic datasets. It is shown that particular synthetic datasets may be used for training of depth networks in cases that an appropriate dataset is not available. Furthermore, we showed that in these cases, usage of semantic labels improves the robustness of the network against domain shift from synthetic training data to non-synthetic test data. Keywords: Depthestimation,pictorialdepthcues,semanticsegmentation,synthetic dataset, deep convolutional neural network.
|Tunable Graphene-based Pulse Compressor for Terahertz Application
|Seyed Mohammadreza Razavizadeh
|Prof. Ramezanali Sadeghzadeh
|Optical temporal pulse shortening in communication systems has an important role due to the high capacity bandwidth that fiber can provide. Terahertz short pulses have been extensively employed in many frontier modern applications, where the transmitted signals are usually frequency-chirped or phase-modulated to increase the time-bandwidth product, leading to an increased propagation length. Thus, a key component for communication systems, the pulseshaper, remains elusive outside the lab. Frequently, temporally short pulses are realized with linear dispersion or nonlinear dispersion compression using external dispersive elements that have large chromatic dispersion. This function might be performed by a variety of photonic components such as gratings, prisms, chirped-mirrors and gas-filled hollow-core fibers. Nevertheless, they lack tunability, which is crucial for communication applications, and maybe difficult to integrate with commercial systems. A promising solution may rely on low-loss dispersion-tunable waveguides, wherein graphene, with excellent tunability at THz frequencies is exploited. Such a waveguide can be realized with helically-corrugated circular waveguides which have been successfully used for “gyro” applications, whose helical corrugation is a hybrid metal-graphene ribbon that allows electrical tunability. In this thesis, the possibility of the linear tunable compression of chirped pulses in the positive group velocity dispersion (GVD) region of a dielectric-lined circular waveguide loaded with a helical graphene ribbon has been discussed. We will show that the proposed structure introduces a good tunability of the compression factor via the graphene electrostatic bias. In addition, this thesis also presents that the waveguide dispersion, as the first component of the chromatic dispersion compared to the second component, i.e. material dispersion that introduced by graphene, plays the main role which is caused by the shape the graphene-ribbon helix. It is demonstrated that there is an optimal compression waveguide length over which THz chirped pulses reach the maximum compression. It is shown that by applying an electrostatic controlling gate voltage (Vg) of 0 and 30 V on the helical graphene ribbon, the temporal input pulses of width 8 and 12 ps, propagating through two different lengths (700 m and 1700 m), can be tuned by 5.9% and 8%, respectively, in the frequency range of 2.15–2.28 THz. Another outstanding achievement of this research was to provide a comprehensive system model by incorporating the full-wave time-domain simulations and the numerical transfer function estimation approach. The use of the system transfer function to analyze the structure is preferable to the full-wave simulation because of saving the execution time.