Theses

Permanent URI for this collection

Browse

Recent Submissions

Now showing 1 - 5 of 34745
  • Publication
    MAXIMAL-LENGTH SEQUENCE CODE CLASSIFICATION OPTIMIZATION PROCEDURE UTILIZING DEEP LEARNING NEURAL NETWORKS
    (Monterey, CA; Naval Postgraduate School, 2022-09) Reeder, Christina N.; Romero, Ric; Electrical and Computer Engineering (ECE); Cristi, Roberto
    Direct sequence spread spectrum techniques are often utilized in encoding communications signals because they can decrease signal spectrum lower than the thermal noise floor of a receiver, making them harder to detect. Accurate and timely classification of spreading codes for message decoding has become an area of interest. In this work, we evaluate the difference in classification performance between a traditional matched filter bank method and trained neural networks. We demonstrate that trained neural networks may out perform matched filters specifically in the medium SNR range. Additionally, we explore performance of a neural network trained to detect and classify direct sequence coded signals along with a null alternative by adding a “noise only” signal classification option. We find that there is a probability of false alarm (𝑃𝑓a) associated with a neural network trained to detect signals with a “noise only” classification option. We conclude that trained neural networks offer an increase in both percentage of classification (𝑃𝑐) and time-to-classify performance. However, we also conclude that more work is required to optimize the neural network for the decoding of preamble codes of different lengths and types. This work demonstrates the feasibility of using trained neural networks for use in decoding direct sequence coded signals.
  • Publication
    ACTIVE BAYESIAN DEEP LEARNING WITH AN ACOUSTIC VECTOR SENSOR
    (Monterey, CA; Naval Postgraduate School, 2021-09) Atchley, Sabrina L.; Orescanin, Marko; Computer Science (CS); Monaco, Vinnie
    Traditional passive monitoring of the ocean’s acoustic signals is conducted with an omnidirectional hydrophone, which only measures acoustic pressure. Vector sensors, unlike hydrophones, respond to both the acoustic pressure and the vector motion of water, providing additional information. This thesis focuses on utilizing vector sensor data as input to a neural network and studies the advantage of utilizing all four channels over single-channel data from the acoustic pressure sensor. A Bayesian deep learning approach is used to build multi-class classification models that provide estimates of uncertainty. The best model had an F1 score of .798 using single-channel data and .81 when using four-channel data from the vector sensor. However, the addition of information from the four-channel signal significantly reduces predictive uncertainty, demonstrating the advantage of utilizing all four channels for passive sonar classification. Next, active learning is examined, an algorithm that typically depends on uncertainty estimates to select the best training data. This is likely the first study on active learning with Bayesian deep learning models in passive sonar classification. With active learning using 23% of the training data, we trained within two percent of the F1 score compared to the entire training data. Additionally, the active learning experiments demonstrated that uncertainty-based acquisition functions increased performance using four channels over single-channel data.
  • Publication
    WAIVING THE STANDARDS: THE EFFECT OF RECRUITMENT WAIVERS IN THE ROYAL AUSTRALIAN AIR FORCE
    (Monterey, CA; Naval Postgraduate School, 2024-03) Woodside, Jennifer K.; Tick, Simona L.; Ahn, Sae Young; Department of Defense Management (DDM); Department of Defense Management (DDM)
    The Australian Defence Force faces a recruitment challenge, prompting all services to increasingly rely on recruitment waivers to temporarily match enlistment standards with personnel requirements. This study assesses the growing utilization of waivers by the Royal Australian Air Force (RAAF), evaluating how they may affect RAAF’s ability to both achieve its recruiting requirements and not incur losses during time spent under training. Using linear probability models to analyze historic enlistment and waiver issuance data from 2016 through 2021, this study reveals a statistically significant yet negligible effect of recruitment waivers on the successful service of past aviator cohorts. An examination of waiver categories, characteristics and military occupation groups uncover further positive and negative effects. Medical waivers exhibit a significant negative impact on service, while waivers related to driver’s license, physical fitness, security background and criminal history can positively influence select occupation groups. The findings serve as positive indicators for the effectiveness of the current recruitment waiver policy. Based on the observed impact from prior usage, the continued strategic deployment of waivers is recommended ensuring their use aligns with RAAF’s risk tolerance and recruitment needs. It is essential for RAAF to recognize that waivers are not a solution for the recruitment challenge but rather a tool to enhance the strategy into the future.
  • Publication
    UNINTENDED CONSEQUENCES: AN ANALYSIS OF THE IMPACT OF INCREASED TIME-IN-SERVICE PROMOTION REQUIREMENTS ON NCO RETENTION AND PERFORMANCE IN THE USMC
    (Monterey, CA; Naval Postgraduate School, 2024-03) Young, Mark W.; Ahn, Sae Young; Department of Defense Management (DDM); Bacolod, Marigee
    In 2019, the Marine Corps announced that the minimum required time-in-service and time-in-grade for promotion to sergeant and staff sergeant would increase in 2020. Since that policy was enacted, a 2,700-sergeant deficit has been identified. This study confirms that deficit is linked to the promotion policy change by estimating the impact of the increased promotion requirements on the retention and job performance of corporals and sergeants. To estimate the policy impact, I mimic an experimental research design and employ a difference-in-differences framework, comparing Marines in jobs where the average time to promote increased the most against Marines in jobs where promotion timing stayed the same or changed minimally. The results show that corporals in the treatment group were significantly more likely to separate after the new policy was enacted, while sergeants in the treatment group were less likely to separate. Additionally, corporals in the treatment group were more likely to be meritoriously promoted to sergeant after the new policy was in effect, though the effect of the policy on the performance of treated corporals was negligible. Based on these results, I recommend that the Marine Corps focus retention incentives and lateral entry initiatives towards military occupational specialties that have been most affected by this policy, as well as further evaluate meritorious promotion management to enhance its effectiveness in selecting individuals for early advancement.
  • Publication
    THE SEXTORTION PHENOMENON: CREATING A LEGAL FRAMEWORK TO EFFECTIVELY SAFEGUARD MINOR VICTIMS
    (Monterey, CA; Naval Postgraduate School, 2024-03) Vera, Brittany J.; Matei, Cristiana; Brannan, David W.; National Security Affairs (CHDS)
    This thesis studies the prevalent and complex issue of sextortion, focusing on its detrimental effects on minors. It critically analyzes how this form of exploitation, which involves blackmailing minors for sexual content, sexual activity, or money, unfolds across the digital world. The current legal framework in California, Penal Code 518(b), fails to differentiate between adult and minor victims sufficiently and does not classify sextortion as a sex crime. To test the adequacy of current laws to prosecute sextortion in California, this thesis compared two case studies involving sex offender registration/community notification and human trafficking and conducted interviews with district attorneys, defense attorneys, and law enforcement officers across California. It also explored legal frameworks from Utah, South Carolina, and New Jersey, which have enacted more targeted sextortion laws for minors. This thesis underscores the need for laws that are adaptable to technological advances, especially artificial intelligence, and that can more effectively protect vulnerable young children. The thesis concludes that a combination of legislative reform and educational initiatives is crucial for combating sextortion. This approach addresses the legal gaps and raises awareness and prevention, aiming to offer comprehensive protection and justice for minors affected by this heinous crime.