東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Many-Server Queueing Models With App...

Zhong, Yueyang.

FindBook

Google Book

Amazon

博客來

Many-Server Queueing Models With Applications to Modern Service Operations Management.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Many-Server Queueing Models With Applications to Modern Service Operations Management./
作者:	Zhong, Yueyang.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2023,
面頁冊數:	402 p.
附註:	Source: Dissertations Abstracts International, Volume: 85-06, Section: B.
Contained By:	Dissertations Abstracts International85-06B.
標題:	Statistics. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30691001
ISBN:	9798381159431

Many-Server Queueing Models With Applications to Modern Service Operations Management.
Zhong, Yueyang.

Many-Server Queueing Models With Applications to Modern Service Operations Management. - Ann Arbor : ProQuest Dissertations & Theses, 2023 - 402 p.

Source: Dissertations Abstracts International, Volume: 85-06, Section: B.

Thesis (Ph.D.)--The University of Chicago, 2023.

Service system design is often informed by queueing theory, which helps system managers understand the impact of managerial design decisions on system performance. Traditional queueing theory assumes that servers are inanimate entities and system characteristics are perfectly known. However, these assumptions do not hold generally in modern service systems, where human servers exhibit strategic behavior and system characteristics may not be fully accessible. This thesis accounts for the complexities of human decision-making and the uncertainties of operational environments by integrating human server behavior and statistical learning into classical many-server queueing models, providing a framework for the analysis and optimization of behavior-aware and prediction-driven modern service systems.In Chapter 1, we develop a game-theoretic model to investigate how human server work speed is affected by managerial decisions concerning (i) how many servers to staff and how much to pay them, and (ii) whether and when to turn away customers. We do this in the context of a finite-buffer many-server Markovian queue, where each server selfishly chooses her work speed in order to maximize an expected utility that captures an inherent trade-off between payment, idleness, and effort cost. Then, the work speeds emerge as the Nash equilibrium to a noncooperative game. We establish results on equilibrium existence and uniqueness, and demonstrate non-monotonic behavior. These results indicate that the commonly accepted rule of thumb that reducing workload decreases customer waiting time can be flawed due to servers adapting their work speeds in response to managerial incentives.Chapter 2 studies a learning variant of a canonical scheduling problem in a multiclass many-server general queue with abandonment, when system characteristics are unknown and may be learned, and abandonments are costly. The difficulty is that even when system characteristics are known, characterizing an optimal scheduling policy appears intractable because the state space is very complex. Fortunately, the simple aμ-rule (that prioritizes classes for service in accordance with their cost of abandonment times service rate) is asymptotically optimal (under certain conditions) for large systems that do not have sufficient capacity to serve all customers. We propose a Learn-then-Schedule policy that first learns the unknown service rates and then schedules according to the empirical aμ-rule, which we show achieves the smallest achievable regret relative to the aμ-rule, that is of order log T (where time T is the system run-time).In Chapter 3, we delve into a control problem in the context of a single-class many-server general queue with abandonment. The objective is to strike a balance between operational costs (specifically, abandonment and holding costs) and human server utilization costs (stemming from fatigue). The control question revolves around determining when an available server should commence serving the next customer and when they should take a break. Our analysis of this control problem for large systems motivates that non-idling service disciplines are not in general optimal. To address this, we propose an admission control policy designed to ensure that servers have sufficient idle time. We show that this policy is asymptotically optimal as time as well as the arrival rate and number of servers grow large.

ISBN: 9798381159431Subjects--Topical Terms:

517247
Statistics.
Subjects--Index Terms:

Asymptotic analysis

Many-Server Queueing Models With Applications to Modern Service Operations Management.
LDR:04661nmm a2200409 4500 001 2399426
005 20240916065437.5
006 m o d
007 cr#unu||||||||
008 251215s2023 ||||||||||||||||| ||eng d
020 $a 9798381159431
035 $a (MiAaPQ)AAI30691001
035 $a AAI30691001
040 $a MiAaPQ $c MiAaPQ
100 1 $a Zhong, Yueyang. $3 3769393
245 1 0 $a Many-Server Queueing Models With Applications to Modern Service Operations Management.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2023
300 $a 402 p.
500 $a Source: Dissertations Abstracts International, Volume: 85-06, Section: B.
500 $a Advisor: Ward, Amy.
502 $a Thesis (Ph.D.)--The University of Chicago, 2023.
520 $a Service system design is often informed by queueing theory, which helps system managers understand the impact of managerial design decisions on system performance. Traditional queueing theory assumes that servers are inanimate entities and system characteristics are perfectly known. However, these assumptions do not hold generally in modern service systems, where human servers exhibit strategic behavior and system characteristics may not be fully accessible. This thesis accounts for the complexities of human decision-making and the uncertainties of operational environments by integrating human server behavior and statistical learning into classical many-server queueing models, providing a framework for the analysis and optimization of behavior-aware and prediction-driven modern service systems.In Chapter 1, we develop a game-theoretic model to investigate how human server work speed is affected by managerial decisions concerning (i) how many servers to staff and how much to pay them, and (ii) whether and when to turn away customers. We do this in the context of a finite-buffer many-server Markovian queue, where each server selfishly chooses her work speed in order to maximize an expected utility that captures an inherent trade-off between payment, idleness, and effort cost. Then, the work speeds emerge as the Nash equilibrium to a noncooperative game. We establish results on equilibrium existence and uniqueness, and demonstrate non-monotonic behavior. These results indicate that the commonly accepted rule of thumb that reducing workload decreases customer waiting time can be flawed due to servers adapting their work speeds in response to managerial incentives.Chapter 2 studies a learning variant of a canonical scheduling problem in a multiclass many-server general queue with abandonment, when system characteristics are unknown and may be learned, and abandonments are costly. The difficulty is that even when system characteristics are known, characterizing an optimal scheduling policy appears intractable because the state space is very complex. Fortunately, the simple aμ-rule (that prioritizes classes for service in accordance with their cost of abandonment times service rate) is asymptotically optimal (under certain conditions) for large systems that do not have sufficient capacity to serve all customers. We propose a Learn-then-Schedule policy that first learns the unknown service rates and then schedules according to the empirical aμ-rule, which we show achieves the smallest achievable regret relative to the aμ-rule, that is of order log T (where time T is the system run-time).In Chapter 3, we delve into a control problem in the context of a single-class many-server general queue with abandonment. The objective is to strike a balance between operational costs (specifically, abandonment and holding costs) and human server utilization costs (stemming from fatigue). The control question revolves around determining when an available server should commence serving the next customer and when they should take a break. Our analysis of this control problem for large systems motivates that non-idling service disciplines are not in general optimal. To address this, we propose an admission control policy designed to ensure that servers have sufficient idle time. We show that this policy is asymptotically optimal as time as well as the arrival rate and number of servers grow large.
590 $a School code: 0330.
650 4 $a Statistics. $3 517247
650 4 $a Computer science. $3 523869
650 4 $a Systems science. $3 3168411
653 $a Asymptotic analysis
653 $a Statistical learning
653 $a Many-server queues
653 $a Service systems
653 $a Strategic behavior
653 $a Unknown parameters
690 $a 0310
690 $a 0984
690 $a 0463
690 $a 0790
710 2 $a The University of Chicago. $b Business. $3 1671506
773 0 $t Dissertations Abstracts International $g 85-06B.
790 $a 0330
791 $a Ph.D.
792 $a 2023
793 $a English
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30691001