Systems Architecture Commons^™

Federated Learning Systems For Mobile Sensing Data, Xiaopeng Jiang

Dissertations

Federated Learning (FL) has emerged as a new distributed Deep Learning (DL) paradigm that enables privacy-aware training and inference on mobile devices with help from the cloud. This dissertation presents a comprehensive exploration of FL with mobile sensing data, covering systems, applications, and optimizations.

First, a mobile-cloud FL system, FLSys, is designed to balance model performance with resource consumption, tolerate communication failures, and achieve scalability. In FLSys, different DL models with different FL aggregation methods can be trained and accessed concurrently by different apps. In addition, FLSys provides advanced privacy-preserving mechanisms and a common API for third-party app developers to …

A Holistic And Collaborative Behavioral Health Detection Framework Using Sensitive Police Narratives, Martin Keagan Wynne Brown

Dissertations

Identifying behavioral health is paramount for law enforcement officers to provide appropriate follow-up community care. In the current practice, law enforcement offices manually identify these behavioral health cases to allow the designation of the relevant follow-up resources. Police reports generated by officers' response to 911 calls remain an untapped resource for identifying such incidents. Therefore, we advocate for the incorporation of manual annotations from experts, natural language processing (NLP), active learning, advanced machine learning, and ensemble techniques to detect behavioral health cases within police reports. In this dissertation, we develop tools and frameworks to automatically detect behavioral health cases from …

A Practical Approach To Automated Software Correctness Enhancement, Aleksandr Zakharchenko

Dissertations

To repair an incorrect program does not mean to make it correct; it only means to make it more-correct, in some sense, than it is. In the absence of a concept of relative correctness, i.e. the property of a program to be more-correct than another with respect to a specification, the discipline of program repair has resorted to various approximations of absolute (traditional) correctness, with varying degrees of success. This shortcoming is concealed by the fact that most program repair tools are tested on basic cases, whence making them absolutely correct is not clearly distinguishable from making them relatively more-correct. …

Resource Optimization In Support Of Iot Applications, Ihab Ahmed Mohammed

Dissertations

With the rise of the Internet of Things (IoT) and smart communities, managing computation and communication resources required by billions of smart devices becomes a concern. To tackle this problem, we develop algorithms for resource management to ensure better Quality of Service (QoS), safety, and performance. We focus our efforts on three problems.

In the first problem, we studied the strict QoS requirements of applications and differentiated service requirements in different situations of vehicular networks. We propose a generic prioritization and resource management algorithm that can be used to prioritize the processing of received packets in vehicular networks. We formulate …

Adaptive Security-Aware Scheduling For Packet Switched Networks Using Real-Time Multi-Agent Systems, Ma'en Saleh Saleh

Dissertations

Conventional real-time scheduling algorithms are in care of timing constraints; they don’t pay any attention to enhance or optimize the real-time packet’s security performance. In this work, we propose an adaptive security-aware scheduling with congestion control mechanism for packet switching networks using real-time agentbased systems. The proposed system combines the functionality of real-time scheduling with the security service enhancement, where the real-time scheduling unit uses the differentiated-earliest-deadline-first (Diff-EDF) scheduler, while the security service enhancement scheme adopts a congestion control mechanism based on a resource estimation methodology.

The security service enhancement unit was designed based on two models: singlelayer and weighted …

Adaptive Radial Basis Function Neural Networks-Based Real Time Harmonics Estimation And Pwm Control For Active Power Filters, Eyad Kh Almaita

Dissertations

With the proliferation of nonlinear loads in the power system, harmonic pollution becomes a serious problem that affects the power quality in both transmission and distribution systems. Active power filters (APF) have been proven to be one of the most successful methods for mitigating harmonics problems. So far, different techniques have been used in harmonics extraction and control of APF to satisfy the fast response and the accuracy required by the APF. Neural networks techniques have been used successfully in different real-time and complex situations. This dissertation demonstrates four main tasks; (i) a novel adaptive radial basis function neural networks …

Load Balancing And Congestion Avoidance Routing, Konstantinos N. Kokkinos

Dissertations

Today’s high speed backbone networks are expected to support a wide range of communication-intensive applications. One of the most important issues in Quality of Service (QoS) is efficient routing. Many QoS routing solutions have been published lately for different criteria of QoS requirements and resource constraints. In this dissertation we focus on the design of regular network topologies and suggest efficient routing schemes to reduce the probability of hot spot creation in the network. Furthermore, we provide a detection of congestion mechanism that reroutes traffic to maintain balancing with small communication cost. Several theoretical results relatively to network traffic balancing …