As the Next-Generation (NextG, 5G and beyond) wireless network supports a wider range of services, optimization of resource allocation plays a crucial role in ensuring efficient use of the (limited) available network resources. Note that resource allocation may require knowledge of network parameters (e.g., channel state information and available power level) for package schedule. However, wireless networks operate in an uncertain environment where, in many practical scenarios, these parameters are unknown before decisions are made. In the absence of network parameters, a network controller, who performs resource allocation, may have to make decisions (aimed at optimizing network performance and satisfying users' QoS requirements) while emph{learning}. To that end, this dissertation studies two novel online learning problems that are motivated by autonomous resource management in NextG. Key contributions of the dissertation are two-fold. First, we study reward maximization under uncertainty with fairness constraints, which is motivated by wireless scheduling with Quality of Service constraints (e.g., minimum delivery ratio requirement) under uncertainty. We formulate a framework of combinatorial bandits with fairness constraints and develop a fair learning algorithm that successfully addresses the tradeoff between reward maximization and fairness constraints. This framework can also be applied to several other real-world applications, such as online advertising and crowdsourcing. Second, we consider global reward maximization under uncertainty with distributed biased feedback, which is motivated by the problem of cellular network configuration for optimizing network-level performance (e.g., average user-perceived Quality of Experience). We study both the linear-parameterized and non-parametric global reward functions, which are modeled as distributed linear bandits and kernelized bandits, respectively. For each model, we propose a learning algorithmic framework that can be integrated with different differential privacy models. We show that the proposed algorithms can achieve a near-optimal regret in a communication-efficient manner while protecting users' data privacy ``for free''. Our findings reveal that our developed algorithms outperform the state-of-the-art solutions in terms of the tradeoff among the regret, communication efficiency, and computation complexity. In addition, our proposed models and online learning algorithms can also be applied to several other real-world applications, e.g., dynamic pricing and public policy making, which may be of independent interest to a broader research community. / Doctor of Philosophy / As the Next-Generation (NextG) wireless network supports a wider range of services, optimization of resource allocation plays a crucial role in ensuring efficient use of the (limited) available network resources. Note that resource allocation may require knowledge of network parameters (e.g., channel state information and available power level) for package schedule. However, wireless networks operate in an uncertain environment where, in many practical scenarios, these parameters are unknown before decisions are made. In the absence of network parameters, a network controller, who performs resource allocation, may have to make decisions (aimed at optimizing network performance and satisfying users' QoS requirements) while emph{learning}. To that end, this dissertation studies two novel online learning problems that are motivated by resource allocation in the presence uncertainty in NextG. Key contributions of the dissertation are two-fold. First, we study reward maximization under uncertainty with fairness constraints, which is motivated by wireless scheduling with Quality of Service constraints (e.g., minimum delivery ratio requirement) under uncertainty. We formulate a framework of combinatorial bandits with fairness constraints and develop a fair learning algorithm that successfully addresses the tradeoff between reward maximization and fairness constraints. This framework can also be applied to several other real-world applications, such as online advertising and crowdsourcing. Second, we consider global reward maximization under uncertainty with distributed biased feedback, which is motivated by the problem of cellular network configuration for optimizing network-level performance (e.g., average user-perceived Quality of Experience). We consider both the linear-parameterized and non-parametric (unknown) global reward functions, which are modeled as distributed linear bandits and kernelized bandits, respectively. For each model, we propose a learning algorithmic framework that integrate different privacy models according to different privacy requirements or different scenarios. We show that the proposed algorithms can learn the unknown functions in a communication-efficient manner while protecting users' data privacy ``for free''. Our findings reveal that our developed algorithms outperform the state-of-the-art solutions in terms of the tradeoff among the regret, communication efficiency, and computation complexity. In addition, our proposed models and online learning algorithms can also be applied to several other real-world applications, e.g., dynamic pricing and public policy making, which may be of independent interest to a broader research community.

No / We use the method of policy capturing to address three open-ended questions regarding how people judge the fairness of events. First, do people differ in how they judge whether a situation is fair or unfair; second, are fairness judgments stable within-person; and, third, how much insight do people have into how they make fairness judgments? To investigate these questions, we used the method of policy capturing and a representative design that samples situations as well as participants. Forty-nine employees rated the global fairness of 56 performance appraisals sampled from their own organization (N = 2,744 situations), and regression methods were used to infer their judgment policy from their choices. We found that people differed greatly in how they judged fairness but used quite consistent policies across similar situations. Participants also provided self-reports of their judgment policies, and comparisons of these self-reports with actual policies revealed limited levels of self-insight.

Yes / A major challenge for supply chain managers is how to manage sourcing relationships to ensure reliable and predictable actions of distant suppliers. The extant research into sustainable supply chain management (SSCM) has traditionally focused on the transactional and collaboration approaches through which buyers encourage suppliers to act responsibly. However, little effort has been devoted to investigating the factors that underpin and enable effective implementation of these two approaches, or to exploring alternative approaches to help sustain an acceptable level of social performance from suppliers. Building on organisational justice theory, we developed a framework in which we propose that buyers’ justice (i.e. distributive, procedural and interactional) as perceived by suppliers can serve as an alternative and complementary vehicle to the conventional sustainability governance approaches for driving the social justice exhibited by suppliers. The paper sheds new light on an alternative relational approach to help to restrain potentially harmful acts of suppliers. It provides a foundation for new research avenues in the SSCM context and supports more informed decision making by practitioners.

Yes / Research has suggested a fundamental connection between fairness and well-being at the individual, relational, and societal levels. Mattering is a multidimensional construct consisting of feeling valued by, and adding value to, self and others. Prior studies have attempted to connect mattering to both fairness and a variety of well-being outcomes. Based on these findings, we hypothesize that mattering acts as a mediator between fairness and well-being. This hypothesis was tested through Covariance-Based Structural Equation Modeling (CB-SEM) using multidimensional measures of fairness, mattering, and well-being. Results from a Latent Path Analysis conducted on a representative sample of 1,051U.S. adults provide support to our hypothesis by revealing a strong direct predictive effect of mattering onto well-being and a strong indirect effect of fairness onto well-being through mattering. Results also show that mattering is likely to fully mediate the relationship between fairness and multiple domains of well-being, except in one case, namely, economic well-being. These findings illustrate the value of a focus on mattering to understand the relationship between fairness and well-being and to provide future directions for theory, research, and practice. Theoretical implications for the experience of citizenship and participation, along with cross-cultural considerations, are also discussed. / Erwin and Barbara Mautner Endowed Chair in Community Well-Being at the University of Miami

Utilizando a survey junto ao consumidor da universidade de Michigan, obtivemos dados a respeito das expectativas dos agentes não só sobre inflação, mas também sobre nível de atividade (os estudos de Roberts (1997) utilizaram apenas os dados de survey sobre expectativas inflacionárias). Verificamos, então, que grande parcela do custo das estabilizações dos EUA foi antecipado pela maior parte dos agentes, o que nos levou a rejeitar os modelos de Taylor (1979, 1980) e de Calvo (1983), mesmo em suas versões com as hipóteses de falta de credibilidade e informação homogeneamente defasada. Em seguida discutimos como um modelo com fairness, pode explicar este quebra cabeças. Finalmente, examinamos, três possíveis fatores (mutuamente compatíveis): a hipótese de que parte da população tenha expectativas inconsistentes, incerteza knightiana e o modelo com fairness. Nossos resultados empíricos penderam a favor de uma combinação de pelo menos uma das duas últimas alternativas com a primeira. / Using the Michigan Universitys consumer survey, we obtained data about agents expectations of both inflation and output (the latter had not been used in Roberts (1997) studies). With this, we were able to verify that a great part of the sacrifice ratios of the US stabilizations were anticipated by common agents, rejecting the Taylor (1979, 1980) and Calvo (1983) models and, with it, the hypothesis that the only reasons underlying them are staggered contracts, homogeneous sticky information and lack in credibility. WE, then, discuss how a model with fairness can explain this puzzle. Finally, we examine three (mutually consistent) factors: the hipothesis that part of the population have inconsistent expectatitons, Knightian uncertainty and te model with fairness. The results favored the combination of at least one of the two latter alternatives with te former.

Avec l’émergence rapide des applications Internet, il est prévu que le trafic mobile mondial augmente de huit fois entre fin 2018 et 2022. En même temps, les futurs systèmes de communication se devront aussi d’améliorer l'efficacité spectrale des transmissions, le temps de latence et l’équité entre utilisateurs. À cette fin, une technique d’accès multiple non orthogonal (NOMA) a été récemment proposée comme un candidat prometteur pour les futurs accès radio. La technique NOMA est basée sur un nouveau domaine de multiplexage, le domaine des puissances. Elle permet la cohabitation de deux ou plusieurs utilisateurs par sous-porteuse ou sous-bande de fréquence. Cette thèse aborde plusieurs problèmes liés à l’allocation de ressources basée sur NOMA afin d'améliorer les performances du réseau en termes d'efficacité spectrale, de débit et/ou d’équité entre utilisateurs. Dans ce sens, des solutions théoriques et algorithmiques sont proposées et des résultats numériques sont obtenus afin de valider les solutions et de vérifier la capacité des algorithmes proposés à atteindre des performances optimales ou sous-optimales. Après une étude bibliographique des différentes techniques d’allocation de ressources présentée dans le premier chapitre, on propose dans le deuxième chapitre plusieurs stratégies d’allocation de ressource où une réduction de la bande utilisée par les utilisateurs est ciblée. Les résultats de simulation montrent que les stratégies proposées améliorent à la fois l’efficacité spectrale et le débit total des utilisateurs par rapport aux systèmes basés uniquement sur des techniques d’accès orthogonales. Quant au troisième chapitre, il étudie la performance du Proportional Fairness (PF) Scheduler tout en considérant que la bande passante est disponible en totalité. Dans ce sens, plusieurs améliorations basées sur le PF sont proposées, qui offrent au système NOMA des avantages en termes de débit, d’équité entre utilisateurs et de qualité de service. Dans le quatrième chapitre, nous proposons plusieurs techniques d’allocation de ressources qui donnent aux utilisateurs la possibilité de favoriser le débit par rapport à l’équité entre utilisateurs et vice versa. Dans le dernier chapitre, différentes techniques permettant une transmission hybride broadcast/broadband sur la même bande de fréquence sont proposées et comparées à l’état de l’art. / With the proliferation of Internet applications, between the end of 2016 and 2022, total mobile traffic is expected to increase by 8 times. At the same time, communications networks are required to further enhance system efficiency, latency, and user fairness. To this end, non-orthogonal multiple access (NOMA) has recently emerged as a promising candidate for future radio access. By exploiting an additional multiplexing domain, the power domain, NOMA allows the cohabitation of two or more users per subcarrier, based on the principle of signal superposition. This dissertation addresses several radio resource allocation problems in mobile communication systems, in order to improve network performance in terms of spectral efficiency, through put, or fairness. Theoretical analysis and algorithmic solutions are derived. Numerical results are obtained to validate our theoretical findings and demonstrate the algorithms ability of attaining optimal or sub-optimal solutions. To this direction, the second chapter of this thesis investigates several new strategies for the allocation of radio resources (bandwidth and transmission power) using NOMA principle, where the minimization of the total amount of used bandwidth is targeted. Extensive simulation results show that the proposed strategies for resource allocation can improve both the spectral efficiency and the cell-edge user throughput, especially when compared to schemes employing only orthogonal signaling. A context where the total bandwidth is available has also been studied, in the 3rd chapter where we investigate the performance of the proportional fairness (PF) scheduler, and we propose modifications to it, at the level of user scheduling and power allocation that show to improve the system capacity, user fairness and QoS. In the 4th chapter, we proposed new pairing metrics that allow to favor the fairness at the expense of the throughput and vice versa. The proposed metrics show enhancements at the level of system capacity, user fairness, and computational complexity. Different techniques that allow a hybrid broadcast/multicast transmission on the same frequency platform are proposed in the last chapter and compared to the state of the art.

Prior to the Constitutional Court's decision in the Sidumo and another v Rustenburg Platinum Mines Ltd and others (2007) lACC 22 the Commission for Conciliation Mediation and Arbitration applied the "reasonable employer"–test to determine whether a specific sanction, issued by an employer, was fair. The "reasonable employer"–test provided a lot of flexibility to employers to dismiss employees for misconduct, as employers' decisions to dismiss were "protected" from scrutiny by the CCMA. The Constitutional Court replaced the "reasonable employer"–test, which required a measure of deference to the decision of the employer, with that of the "reasonable decision maker"–test, which required an answer to the question whether the decision reached by the commissioner was one that a reasonable decision maker could not reach? This meant that in the event that the decision reached by the commissioner was one that a reasonable decision maker could not reach, that the decision of the commissioner will be overturned on review. The change in test from a "reasonable employer" to that of a "reasonable decision maker" had significant implications for employers who are instituting disciplinary action against their employees and subsequently imposing the sanction of dismissal, as commissioners are no longer allowed to "defer" to the decision imposed by employers. The Sidumo test also have implications for employers who are seeking to take decisions of the CCMA on review, as londo JP held in Fidelity Cash Management Service v CCMA 2008 29 ILJ 964 (LAC) that it will not be often that an arbitration award is found to be one that a reasonable decision maker could not have made. / Thesis (LL.M. (Labour Law))--North-West University, Potchefstroom Campus, 2011.

Prior to the Constitutional Court's decision in the Sidumo and another v Rustenburg Platinum Mines Ltd and others (2007) lACC 22 the Commission for Conciliation Mediation and Arbitration applied the "reasonable employer"–test to determine whether a specific sanction, issued by an employer, was fair. The "reasonable employer"–test provided a lot of flexibility to employers to dismiss employees for misconduct, as employers' decisions to dismiss were "protected" from scrutiny by the CCMA. The Constitutional Court replaced the "reasonable employer"–test, which required a measure of deference to the decision of the employer, with that of the "reasonable decision maker"–test, which required an answer to the question whether the decision reached by the commissioner was one that a reasonable decision maker could not reach? This meant that in the event that the decision reached by the commissioner was one that a reasonable decision maker could not reach, that the decision of the commissioner will be overturned on review. The change in test from a "reasonable employer" to that of a "reasonable decision maker" had significant implications for employers who are instituting disciplinary action against their employees and subsequently imposing the sanction of dismissal, as commissioners are no longer allowed to "defer" to the decision imposed by employers. The Sidumo test also have implications for employers who are seeking to take decisions of the CCMA on review, as londo JP held in Fidelity Cash Management Service v CCMA 2008 29 ILJ 964 (LAC) that it will not be often that an arbitration award is found to be one that a reasonable decision maker could not have made. / Thesis (LL.M. (Labour Law))--North-West University, Potchefstroom Campus, 2011.

Utilizando a survey junto ao consumidor da universidade de Michigan, obtivemos dados a respeito das expectativas dos agentes não só sobre inflação, mas também sobre nível de atividade (os estudos de Roberts (1997) utilizaram apenas os dados de survey sobre expectativas inflacionárias). Verificamos, então, que grande parcela do custo das estabilizações dos EUA foi antecipado pela maior parte dos agentes, o que nos levou a rejeitar os modelos de Taylor (1979, 1980) e de Calvo (1983), mesmo em suas versões com as hipóteses de falta de credibilidade e informação homogeneamente defasada. Em seguida discutimos como um modelo com fairness, pode explicar este quebra cabeças. Finalmente, examinamos, três possíveis fatores (mutuamente compatíveis): a hipótese de que parte da população tenha expectativas inconsistentes, incerteza knightiana e o modelo com fairness. Nossos resultados empíricos penderam a favor de uma combinação de pelo menos uma das duas últimas alternativas com a primeira. / Using the Michigan Universitys consumer survey, we obtained data about agents expectations of both inflation and output (the latter had not been used in Roberts (1997) studies). With this, we were able to verify that a great part of the sacrifice ratios of the US stabilizations were anticipated by common agents, rejecting the Taylor (1979, 1980) and Calvo (1983) models and, with it, the hypothesis that the only reasons underlying them are staggered contracts, homogeneous sticky information and lack in credibility. WE, then, discuss how a model with fairness can explain this puzzle. Finally, we examine three (mutually consistent) factors: the hipothesis that part of the population have inconsistent expectatitons, Knightian uncertainty and te model with fairness. The results favored the combination of at least one of the two latter alternatives with te former.

We explore the political acceptance of taxation in commodity markets. Participants in our experiment earn incomes by trading and must collectively choose one of two tax regimes to raise a given tax revenue. A "uniform tax" (UT) imposes the same tax rate on all markets and is fair in that it yields the same - but low - income to participants in all markets. The "discriminatory tax" (DT) imposes a higher burden on markets with inelastic demand and is therefore efficient but it is also unfair in that incomes are unequal across markets. We find that DT are unpopular, as predicted. Surprisingly, however, DT remain unpopular when they are both efficient and produce a fair (equal) distribution. We conclude that non-discrimination (equal treatment) is a salient fairness principle in taxation that shapes voting on commodity taxes above and beyond concerns for efficiency and equal distribution. (authors' abstract) / Series: WU International Taxation Research Paper Series