Dissertation Defense Announcements

Aggregate production planning generally deals with configuration of an aggregate plan in advance of 6 to 18 periods (e.g. months) to give the organizations an idea about the amount of invested money, utilized capacity, required inventory and any other procurement activities need to be done before the actual times arrives. Inherent uncertainties faced by the planners (caused by unreliable estimates of demand, cost or production processes) could make the production planning a challenging task. That is, the production planners not only have to deal with the available parameters’ uncertainties (Demand, cost, etc.), but also, new information which become available with the pass of time, sometimes requires several re-planning activities for the future periods. Stochastic and Fuzzy planning are among the popular techniques to deal with the uncertainties in optimization models. While the stochastic/fuzzy programming techniques provide a more realistic representation of future estimations, the production plans need to be also revised from one planning period to another as time rolls and new information become available (a.k.a. rolling horizon planning). However, frequent re-planning activities and changes in the production plans could result in a state of plan instability causing plan related “nervousness” in manufacturing firms, which could undermine manager’s confidence in the system, depriving it of the support needed for successful operations. It could also result in disruptions in the production and delivery systems, which could result in inaccurate personnel scheduling, machine loading, and unnecessary supplier orders.
Frozen horizon along with other solution approaches attempt to provide insights on how to mitigate nervousness, however, most of the existing approaches do not consider the flexibility aspect in production plans. Flexible Requirements Profile (FRP) and bi-objective optimization are alternative stabilizing approaches which are the focus of this research. In FRP, flexible bounds are enforced on production plans to maintain the desired degree of flexibility. Instead of 0% flexibility in the case of a frozen period or 100% flexibility in the case of plan to order, FRP model considers different flexibility levels. For the bi-objective optimization approach, the production planning problem can also be formulated with two objectives, where one trades-off between the traditional cost objective and the plan stability objective.
Our main research objectives are:
1) to compare FRP-APP with Stochastic and Fuzzy APP in terms of both plan cost and stability,
2) to develop and compare new “hybrid” Stochastic and Fuzzy FRP-APP models to combine the strengths of stochastic and fuzzy models, which represent input uncertainties more realistically, and FRP models that have better control over plan variability,
3) to develop and compare new Stochastic and Fuzzy Bi-objective APP models as alternate techniques to trade off the traditional cost objective with the stability objective formally following a multi-objective decision making framework, and
4) to conduct extensive testing of the proposed FRP-based and Bi-objective models under various industry scenarios.

In this study, we used scattering-scanning near-field optical microscopy (s-SNOM) to experimentally characterize several structures relevant to infrared (IR) antenna technology, by measuring the electric near-field strength as a function of position on metallic antennas and transmission lines. Having spatially resolved measurements of electric field amplitude allows assessment of the wave impedance at any location on a standing-wave structure. We improved the usual s-SNOM data-processing method using a principal-components decomposition to allow unambiguous phase retrieval. We demonstrated the efficacy of this technique on IR bow-tie antennas of continuous and discrete designs, allowing comparison of their polarization dependence and spatial response distribution. This phase-retrieval procedure was used throughout our investigations. An IR sensor of particular interest is the antenna-coupled metal-oxide-metal (MOM) diode, which rectifies IR-frequency current waves collected by the antenna to produce an output voltage proportional to the incident irradiance. These sensors are appealing because they have a fast response and do not require cryogenic cooling. IR antennas have a typical impedance in the range of tens of Ohms at resonance, while MOM diodes have impedance in the range of thousands of Ohms. This impedance mismatch is a limiting factor in the detection sensitivity that can be achieved with antenna-coupled MOM diodes. To address this issue, we studied two impedance-matching techniques. The first is based on the fact that a MOM diode under DC bias exhibits a change in its dynamic resistance. We obtained measurements that demonstrate modification of IR-frequency current waves using diodes contained in the antenna structure. The ability to tune the operating point of a MOM diode and thereby modify antenna or transmission-line impedance at IR frequencies offers the possibility of active impedance-matching networks. The second technique we investigated involved tailoring of the feed-point geometry to obtain an antenna with higher impedance that offers better matching. We designed, fabricated and demonstrated several new IR-antenna designs that have impedance in the range of 1000 Ohms. This new class of antennas stands to improve signal-transfer efficiency to high-impedance IR sensors such as MOM diodes.

The proliferation of connected devices creates various use cases and heterogeneous services, e.g., augmented/virtual reality (AR/VR), vehicle-to-everything (V2X), and mobile artificial intelligence. These services and use cases have diverse networking and performance requirements such as throughput and delay, which challenge the "one-fit-all" service architecture in current networks. In this research, an intelligent network management framework in mobile edge computing is explored. The primary challenges lie in the unique characteristics of heterogeneous services and complicated correlations between network management on multiple technical domains and high-dimension performance requirements in the complex mobile networks. This research addresses these challenges with two different management approaches. From the perspective of service providers, multiple mobile systems are designed to allow service adaptation under complex network dynamics, e.g., channel variation and traffic workload, which dynamically and adaptively adjust resource allocations and system configurations by exploiting the unique characteristics of individual services. From the perspective of infrastructure providers, multiple network systems are proposed to enable orchestration intelligence without accurate performance modelings of services, which automatically learn to orchestrate multiple domain network resources for supporting various services by exploiting advanced machine learning techniques.

Freeform optics, or optics with no axis of rotational invariance, provide optical designers more degrees of freedom, flexibility, and opportunity for innovation increasing optical performance and system integration while decreasing the form factor. Advancements in optical fabrication have enabled freeform surface manufacture with greater precision. Metrology instruments and techniques are needed to verify the performance of freeform optical surfaces and systems to keep up with design and manufacture. Freeform optics often have high slopes, no axis of symmetry, and a large departure from spherical, making traditional metrology techniques inadequate. This research was conducted to enable form measurements of freeform mirrors in the 250 mm class for a next generation three mirror anastigmatic (TMA) telescope complete with a statement of the measurement uncertainty to fill the gap in metrology of freeform optics. A flexible metrology instrument that could measure relatively large optics with customizable probe paths and sampling strategies was needed while maintaining the required precision.