Search Results (1 - 4 of 4 Results)

Sort By  
Sort Dir
 
Results per page  

Haddd, Rami J.Feed-Forward Bandwidth Indication: An Accurate Approach to Multimedia Bandwidth Forecasting and its Application in Ethernet Passive Optical Networks
Doctor of Philosophy, University of Akron, 2011, Electrical Engineering
In this dissertation, we address important issues related to forecasting the bandwidth of stored and live video over optical access networks. The nature of the communication network usually governs the video transport standard used; nevertheless, the nature of the transmitted traffic imposes some challenges on the transport standards that should also be considered. As far as video traffic is concerned, two major issues should be taken into consideration. First, compressed video has a variable bit rate which requires a large transmission bandwidth to ensure a certain Quality of Service (QoS). Therefore, video transport standards must be able to utilize the network bandwidth efficiently. Second, the timing requirement for video is stringent and must be met to ensure smooth playback at the receiving end. In this dissertation, we also propose an accurate video bandwidth forecasting approach called Feed-Forward Bandwidth Indication (FFBI). FFBI will assist the dynamic bandwidth allocation to help utilize the multimedia network resources more efficiently and comply with the timing requirements of video transmission. FFBI provides 100% accurate bandwidth forecast that comes for free for pre-recorded video and at the expense of some introduced delay for live video. With video transport over access networks projected to supplant other transport mechanisms in the next few years, we conducted a performance analysis of FFBI within Ethernet Passive Optical Networks (EPONs). We found that the use of FFBI can provide a 50% reduction in queueing delay compared to the use of no forecasting and a 35% reduction in queueing delay compared to other forecasting methods. In addition, we also found that the use of FFBI can provide a 60% reduction in delay jitter compared to the use of no forecasting and a 88% to 92% reduction in delay jitter compared to other forecasting methods.

Committee:

Subramaniya Hariharan, Dr. (Advisor); Joan Carletta, Dr. (Committee Member); Hamid Bahrami, Dr. (Committee Member); Timothy O'Neil, Dr. (Committee Member); Craig Menzemer, Dr. (Committee Member)

Subjects:

Electrical Engineering

Keywords:

bandwidth forecasting; feedforward bandwidth indication; queueing delay; video bandwidth; bandwidth allocation; feedforward; multimedia communication

Franke, Timothy JosephIdentification and Cancellation of Harmonic Disturbances in Radio Telescopes
Doctor of Philosophy, Case Western Reserve University, 2015, EECS - System and Control Engineering
A new class of algorithms for the identification and cancellation of harmonic disturbances on rotary dynamic systems is proposed and demonstrated with applications on the Green Bank Telescope (GBT). The approach is a model-based iterative algorithm that exploits the structure of the problem to significantly reduce the number of tests needed to perform identification. During each such test, the system is in steady-state periodic operation. The crucial trick involves constructing a correspondence between the coefficients of disturbance terms and their time-periodic harmonics. This transformation enables the modeling and calibration of the system in a compact, harmonic representation. This approach has numerous advantages. The harmonic model fully captures the behavior of arbitrarily complex linear systems. It is not a feedback approach and therefore will not destabilize existing controllers. The algorithm displays rapid convergence and requires a minimal number of tests to construct a model. Finally, a previously identified model may be reused to quickly update a calibration. All of these properties make it ideal for the calibration of feedforward compensators on a wide range of systems. The nature of cogging on the GBT motors is rigorously studied. Various system identification tests are performed to characterize the behavior of the cogging with respect to operating conditions. A single motor calibration routine is developed and deployed on telescope hardware. The performance of the GBT with individually calibrated motors is tested. Additionally, the algorithm is extended to handle multiple interacting motors. A solution method is presented that yields reliable, physically reasonable solutions to the multiple motor problem. The calibration method is updated to compensate for the GBT encoder measurement error. The behavior of interpolation error was studied on two different encoders. Global variation of encoder calibrations is studied over the range of the telescope's elevation axis. Finally, generalization of the algorithm to non-constant rate, but still periodic trajectories, is explored. These tests probe the limits of the algorithm's underlying ideas.

Committee:

Mario Garcia-Sanz, Dr. (Advisor); M. Cenk Cavusoglu, Dr. (Committee Member); Vira Chankong, Dr. (Committee Member); Roger Quinn, Dr. (Committee Member)

Subjects:

Engineering

Keywords:

control theory; feedforward compensation; parameter identification; motor cogging; encoder interpolation; telescope control

Howard, Shaun MichaelDeep Learning for Sensor Fusion
Master of Sciences (Engineering), Case Western Reserve University, 2017, EECS - Computer and Information Sciences
The use of multiple sensors in modern day vehicular applications is necessary to provide a complete outlook of surroundings for advanced driver assistance systems (ADAS) and automated driving. The fusion of these sensors provides increased certainty in the recognition, localization and prediction of surroundings. A deep learning-based sensor fusion system is proposed to fuse two independent, multi-modal sensor sources. This system is shown to successfully learn the complex capabilities of an existing state-of-the-art sensor fusion system and generalize well to new sensor fusion datasets. It has high precision and recall with minimal confusion after training on several million examples of labeled multi-modal sensor data. It is robust, has a sustainable training time, and has real-time response capabilities on a deep learning PC with a single NVIDIA GeForce GTX 980Ti graphical processing unit (GPU).

Committee:

Wyatt Newman, Dr (Committee Chair); M. Cenk Cavusoglu, Dr (Committee Member); Michael Lewicki, Dr (Committee Member)

Subjects:

Artificial Intelligence; Computer Science

Keywords:

deep learning; sensor fusion; deep neural networks; advanced driver assistance systems; automated driving; multi-stream neural networks; feedforward; multilayer perceptron; recurrent; gated recurrent unit; long-short term memory; camera; radar;

Strang, Adam JeffreyFATIGUE-INDUCED EARLY ONSET OF ANTICIPATORY POSTURAL ADJUSTMENTS: HOW IS EARLY ONSET FUNCTIONAL?
Master of Science, Miami University, 2005, Physical Education, Health, and Sport Studies
To examine how fatigue-induced early onset of anticipatory postural adjustments (APAs) are functional, APAs of 30 participants were recorded before and after conditions of either rest (control group, n = 15) or a dead-lift task performed to exhaustion (fatigue group, n = 15). APAs were induced using a unilateral arm-raising maneuver, and were analyzed in the following muscles; lumbar paraspinal, thoracic paraspinal, and the hamstring group. Postural stability was assessed using a force plate. Fatigue had no effect on postural stability, and yet caused earlier APA onsets in 3 of the 6 postural muscles evaluated. Nevertheless, just 1 of the 3 muscles exhibiting early APA onset registered a greater APA EMG integral after the fatiguing task. The findings are consistent with the hypothesis that fatigue-induced early APA onset functions by affording the generation of a muscular impulse similar to that which would be generated in a non-fatigued state.

Committee:

William Berg (Advisor)

Subjects:

Psychology, Experimental

Keywords:

Postural Control; Feedforward; Anticipatory Postural Adjustment