<p> In this thesis, an Overview of the LTE Downlink Scheduling Algorithms is presented. LTE provides high data rates and is expected to be used by many users in future since it has good coverage and has access to internet through their devices. The main mechanism of this LTE network in handling the user traffic is the scheduler. A scheduler is a key functionality in which the base-station decides which user should be given resources to receive or send the information. We focused on LTE Downlink schedulers and their functionality by studying their algorithms. The users are assigned with the shared resources by the network scheduler. We considered two scheduling algorithms, Round Robin and the Best CQI. The impact of each scheduler in terms of throughput and fairness and their abilities are known. Throughput and fairness are the two performance measures in the scheduler. By studying these two schedulers, a new scheduler is proposed which is a compromise between the throughput and the fairness.</p>

Low frequency noise measurements have been widely used to investigate the nature of defects in semiconductor devices. Characterization of low frequency noise performance at different gate bias along with the temperature is very useful to study and identify the defects in the devices. In this work, the frequency, gate bias and temperature dependence of low frequency noise of three differently processed AlGaN/GaN high electron mobility transistors (HEMTs) have been evaluated. Both the frequency-dependence and gate-voltage dependence of the low frequency noise of AlGaN/GaN HEMTs are difficult to understand within the context of the popular Hooge mobility fluctuation model. Instead, the noise is consistent with a carrier-number fluctuation model that includes a non-constant defect-energy distribution Dt(Ef ). A strongly varying Dt(Ef ) in these devices is confirmed by measurements of the temperature dependence of the noise. Estimates of the effective border-trap density before and after 1.8 MeV proton irradiation are provided for both commercial and research-grade devices using a number-fluctuation model. The input-referred noise magnitude for AlGaN/GaN HEMTs decreases at biases that are significantly more positive than the threshold voltage because the gated region of the HEMT comprises a relatively small portion of the channel, and the noise is attenuated by the voltage divider formed by the gated and ungated regions of the channel.

With the development of robotic industry, subhuman robots are paid more attention. In order to meet peopleâs requirements, robots need to grasp humanâs behaviors and service human beings. Skill transfer is the core for this procedure. In this thesis, we supply a process to transfer behaviors from humans to robots or from robots to robots. The technical contributions of this procedure include: (1) two approaches to capture humanâs behavior trajectories; (2) building model to solve robotic kinematics problems; (3) applying Dynamic Movement Primitives (DMP) to achieve targets of reproducing trajectory; (4) combining DMP with Sparse Autoencoder to increase efficient of procedure. As the result, trajectories are transferred successfully from human to robot and from robot to robot. Meanwhile, performances of inverse kinematics and DMP are proved.

Surgical interventions frequently involve the use of instruments that apply heat to tissue, e.g. to seal blood vessels. Controlling the temperature elevation created by these instruments is particularly important when operating in proximity to delicate anatomy, where the buildup of temperature can cause accidental injury and lead to permanent impairment or death. This thesis explores the use of infrared (IR) camera technology to provide thermal monitoring for two specific medical applications, i.e. bone drilling and blood vessel sealing. The technical contributions of this work include: (1) the creation of a miniature thermal stereo camera system intended to provide thermal monitoring during minimally-invasive surgical procedures; (2) the development of software to acquire and visualize the video stream produced by IR thermal cameras, based on the GigE Vision interface standard.

<p> SRAM (Static Random Access Memory) and DRAM (Dynamic Random Access Memory) are two widely used memories in microprocessor as &lsquo;cache&rsquo;. The principle of storing data is different for both. However, they belong to a common family of memories called RAM (Random Access Memory). The use of a specific type of RAM depends on the requirement. The memory size, delay, power consumption, and area are the deciding factors for choosing a type of RAM.</p><p> This project report explores the design of SRAM and DRAM. The tools used include Microwind, and P-Spice A/D. The performances of SRAM and DRAM are calculated and compared on the metrics of delay, power, and area. The calculation of the parameters provides an insight of SRAM and DRAM, and it also suggests where the performances can be improved.</p>

Recently, by increasing the productivity of industrial manufacture, industrial robots have played a key role in many fields of industry (e.g. automobile production, food production, etc.) However, there are two problems rarely mentioned in this field. First, compared with automatization in other fields, industrial robots are programmed manually by a human operator. Second, because of the physical difference between robots and difference of operating platform, there doesnât exist a general method to define the skill (motion records) of robots and make it possible to reuse the skills between robots. In this work, we are trying to propose a skill definition of transfer system which combine the strengths of traditional DMP algorithm and deep learning method. Specifically, in our method, a set of motion primitive bases are generated from motion records in different robots. Skills are re-defined by the linear coefficient of the primitive bases and transferred based on motion primitive bases translation between different platforms. Experiment shows that our method can successfully transfer skills between different models with less space requirement.

The optic nerve is a vital bundle of axons which carries all visual information from the retina posterior to the brain for higher order processing. The optic nerve and eye orbit are affected by many devastating diseases including optic neuritis, glaucoma and anterior ischemic optic neuritis. This dissertation addresses the use of magnetic resonance imaging for investigating anatomical and microstructural changes in the optic nerve in healthy controls and disease cohorts. We propose a fully automated pipeline for segmentation of the optic nerve and other eye-orbit structures. This pipeline is applied to large-scale disease cohort to search for correlations between morphological changes and functional visual measures. We introduce a clinically viable advanced MRI sequence for accurate visualization of the optic nerve and sub-arachnoid cerebrospinal fluid. We develop and improve upon an algorithm to automatically estimate optic nerve and surrounding cerebrospinal fluid radius along the length of the optic nerve. We perform a short- and long-term reproducibility study on young healthy controls for algorithm evaluation and publicly release this data for the standardized comparison of future proposed algorithms. We apply this validated automatic radius estimation algorithm to a clinical population of patients with multiple sclerosis to detect differences in patientsâ eyes with and without a history of optic neuritis. Finally, we utilize a simulation framework to numerically optimize quantitative magnetization transfer imaging sampling patterns to move towards reducing scan times and increasing clinical viability of quantitative magnetization transfer imaging for microstructural characterization of tissue.

Alpha, heavy-ion, neutron, and proton experimental results from 130-nm to 28-nm technology nodes are establish single-event upset cross section trends in soft and hardened flip-flop designs. Trends show that at any LET value soft flip-flops show a decreasing single-event upset cross section with decreasing feature size. Hardened redundant storage node flip-flops show similar cross sections across technologies if the redundant storage node transistor spacing is held constant. Technology computer aided design (TCAD) simulations are used to show there are many competing mechanisms that influence flip-flip single-event upset cross sections as technology feature sizes decrease.

<p> This thesis demonstrates the benefits of using Graph Signal Processing (GSP) techniques for an intelligent taxis transportation system. Graph Signal Processing, an application arising to handle multiple source signals on a graph, has developed into an active field of research during the last several years due to its ability to analyze enormous datasets or dynamic data that usually pose a challenge to researchers. One of the most significant operations of Graph Signal Processing that arises in many areas is noise reduction. </p><p> This thesis introduces a possible method of using Graph Signal Processing and its operations to analyze signals in a network of taxi stand locations. Two examples are given using real data of taxis' and stands' locations in San Francisco where the number of taxis around these stands is the detected signal. The results showed the effectiveness of using Graph Fourier Transform to detect the anomalies in the signals which represent unusual transportation activities or driver distributions within the taxi network. </p><p> Signal denoising is addressed by using four techniques which are often based on the signal filtering methods. The first technique used the low pass filter, followed by a harmonic decline filter, and then standard and modified Kalman filters, including the case for uncertain observation or process noise between the standard and adaptive Kalman filters. The results are compared with the other filters.</p>

<p> This project report presents a fast-acquisition all-digital delay-locked loop (ADDLL) using a starting-bit prediction algorithm for the successive-approximation register (SBP-SAR). The SBP effectively eliminates the harmonic lock and the false lock. The ADDLL design allows wide clock frequency range operation. The algorithm and the digital circuit are digitally simulated and the performance and the advantage over the Conventional-SAR are shown. </p>