Use this URL to cite or link to this record in EThOS:
Title: Optimal resource management in communication networks : theory and algorithm designs
Author: Zafari, Faheem
ISNI:       0000 0004 9357 0810
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2020
Availability of Full Text:
Access from EThOS:
Access from Institution:
Communication networks have enabled human beings and machines to communicate with each other using either wired or wireless technologies. With the ever-growing reliance on the networks, there is a need for efficiently managing and using the available infrastructures as the demand for the resources is more than their supply. Due to the advent of novel paradigms{,} such as {the} Internet of Things (IoT) and edge computing, and the growing demand for resources for data analytics and machine learning (among many other applications), existing resource management algorithms and frameworks may no longer be viable. Therefore, there is a need for novel resource management frameworks that will enable the 'optimal' utilization and timely provision of available resources to different applications to maximize a single or multiple objectives. In particular, there is a need for multi-objective resource sharing frameworks that will enable different service providers (SPs), despite having different objectives or utilities, to share their resources in order to improve their utility and achieve higher application (user) satisfaction. To achieve the above, we first study the single-objective optimization problem of allocating storage, communication and computation resources to reduce energy consumption in a communication network. This issue is important because energy efficiency is a fundamental requirement of communication systems, as reflected in much recent work on performance analysis of system energy consumption. However, most work has only focused on communication and computation energy consumption without considering data caching costs. Given the increasing interest in cache networks, this is a serious deficiency. We consider the problem of energy consumption in data communication, compression and caching (C3) with a quality-of-information (QoI) guarantee in a communication network. Our goal is to identify the optimal data compression rates and cache placement over the network that minimizes the overall energy consumption in the network. We formulate the problem as a 'Mixed Integer Non-Linear Programming' (MINLP) problem with non-convex functions, which is NP-hard in general. We propose a variant of the spatial Branch-and-Bound algorithm (V-SBB) that can provide an ε-global optimal solution to the problem. By extensive numerical experiments, we show that the C$3$ optimization framework improves the energy efficiency by up to 88% compared to any optimization that only considers either communication and caching or communication and computation. Furthermore, the V-SBB technique provides comparatively better solutions than some other MINLP solvers at the cost of added computation time. We then study the problem of multi-objective resource sharing in a communication network with multiple SPs, such as 'in an' edge computing setting, where resources belong to different SPs that have their own objectives or 'utilities' to optimize. On the one hand, certain SPs may not have sufficient resources to satisfy their applications according to the associated service-level agreements (SLAs). On the other hand, some SPs may have additional unused resources. For the case where SPs treat their native and non-native (belonging to other SPs) applications uniformly, we propose a bargaining-theory based resource-sharing framework that enables different SPs to optimally manage their resources and improve the satisfaction level of applications subject to constraints, such as communication costs for sharing resources across SPs. For a specific class of concave utility functions, we present an N-person 'Nash Bargaining Solution' (NBS) for resource management and sharing among SPs with the {Pareto} optimality guarantee. Furthermore, we propose a 'distributed' algorithm to obtain the NBS by proving that the strong-duality property holds for the resultant resource-sharing optimization problem. Using synthetic and real-world data traces, we show numerically that the proposed NBS framework not only enhances the ability to satisfy applications' resource demands, but also improves the utilities of different SPs. Finally, we propose a cooperative game-theoretic framework for resource sharing among SPs with multiple objectives that only requires the SPs to have monotonic, non-decreasing and non-negative utility functions. In contrast with our proposed NBS framework, the cooperative game-theoretic framework is applicable to more general settings at the cost of added computational complexity. Furthermore, the cooperative game-theoretic framework can support two different strategies employed by SPs. In the 'uniform priority' strategy, SPs do not differentiate between their native and non-native applications for resource management. In the 'non-uniform priority' strategy, SPs first allocate resources to their native applications and then share remaining resources with other SPs, if needed. For the uniform priority strategy, we prove that the proposed resource-sharing game is 'canonical' and 'cardinally' convex. Hence, the 'core' is not empty and the grand coalition of SPs is stable. We propose 'Game-theoretic Pareto optimal allocation for the Uniform Priority strategy' (GPUS), a centralized algorithm to obtain a Pareto optimal allocation from the core for the uniform priority strategy. We then modify our game-theoretic framework to enable the SPs to employ non-uniform priority as a strategy. We prove that despite the modification, our proposed resource-sharing game is canonical and cardinally convex. We propose two algorithms, referred to as the 'Game-theoretic Pareto optimal allocation} (GPOA) and 'Polyandrous-Polygamous Matching based Pareto Optimal Allocation' (PPMPOA), also to provide allocations from the core. Hence the obtained allocations are 'Pareto' optimal and the grand coalition of all SPs is stable. Experimental results confirm that our proposed resource-sharing framework improves utilities of SPs and the degree of application request satisfaction.
Supervisor: Leung, Kin Kwo Sponsor: Engineering and Physical Sciences Research Council ; Distributed Analytics and Information Sciences-ITA
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available