Electrical and Electronic Engineering - Theses
Permanent URI for this collection
Search Results
1 - 2 of 2
-
ItemOptical wireless integration: network design challenges( 2013)Today, telecommunication networks are expected to provide high bandwidth services to anywhere and at anytime that the services are requested. Among the many solutions available, the optical-wireless converged network has emerged as a popular solution to enable these mandatory features through combining high capacity optical backhaul and flexible wireless access network. However, as the bandwidth demand in the wireless network increases drastically day by day, wireless access technological advances and new requirements continue to evolve. Therefore, to keep pace with the ever advancing wireless access networks and its services, rapid advancements in the optical backhaul network and a tighter integration of optical and wireless access networks are required in order to provide a quality-of-service (QoS) guaranteed ubiquitous access to end users. In particular, numerous challenges need to be addressed in order to deploy optical-wireless converged networks efficiently. These challenges include provisioning adequate backhaul capacities to guarantee QoS of essential services, facilitating efficient intercommunication between base stations to enable traffic diversions and coordination functions, achieving cost-effective deployment and improving energy efficiency of the converged network. This thesis explores efficient solutions to overcome these key challenges. In particular, approaches to achieve cost effective deployments, to guarantee QoS, and to preserve energy efficiency of the converged networks, are investigated. A denser small cell deployment is considered as a future proof solution to cope with the unabated growth of mobile traffic. However, the deployments of small cells are challenging due to the cost associated with the backhaul, powering, and real estate requirements. Therefore, these challenges need to be tackled effectively in order to achieve all the potential benefits of small cell deployments. To this end, this thesis explores cost-effective approaches that leverage the resources associated with the existing fibre infrastructure to provide fibre backhaul for small cells. By analysing the geographic information system (GIS) data of a real network, we show that the existing fibre terminals can be strategically exploited for the small-cell deployment. Moreover, to reduce the dark fibre usage associated with backhauling, we present an approach to select an optimal subset of fibre terminals for small-cell deployments, which yields maximum possible coverage. In addition, to further reduce the total deployment cost of small-cell backhauling, cost-optimal deployment of passive optical networks (PONs) on top of the existing infrastructure is also explored in this thesis. We demonstrate the applicability of our proposed approaches by using them to plan small-cell deployments in a portion of a large carrier's network in the USA. The results reveal that our proposed approaches can save half of the deployment costs associated with small-cell backhauling. Quality-of-service is an essential requirement for next-generation broadband services such as e-Health and Internet protocol television (IPTV). Therefore, in this thesis, we investigate how the characteristics of both optical and wireless networks can be exploited to provide tighter integration and hence, guaranteed QoS in optical-wireless converged networks. In particular, resource handling methods that strategically exploit the frame structures and resource allocation information retrieved from the wireless network to allocate bandwidth in the converged network, are investigated. Moreover, since the direct communication between base stations is identified as one of the major considerations in the next-generation wireless access networks, we investigate how the widely deployed tree topology-based PON can be used to enable these complementary features. To this end, we explore different PON-wireless converged network architectures. In addition, we also investigate an architecture discovery enabled resource allocation mechanism that can be implemented irrespective of the architecture used for the converged network deployment. Our simulation results indicate that the proposed approaches can significantly improve the QoS performance of both the uplink and inter-communications links between base stations. Moreover, as the energy consumption of telecommunication networks is identified as a major contributor towards the global energy consumption, this thesis also explores the energy efficient aspects of the converged networks. The energy consumption of proposed converged network architectures is investigated under various deployment scenarios by analytically modelling the energy consumption of each of the architectures. In addition, we also explore how the energy saving mechanisms can be implemented in the converged network without compromising the QoS. Overall, our studies reported in this thesis, provide insight into deployment strategies that can be used to realise cost-optimal, energy-efficient, and QoS rich next generation optical-wireless converged networks.
-
ItemFlow control and performance optimization for multi-service networks( 2010)As networks grow and evolve, there are various emerging applications and services available. Taking the Internet as an example, real-time applications (e.g., VoIP and online video clips) nowadays have become increasingly popular besides traditional data transmission services. Given multi-service networks, flow control is a key design issue to ensure the performance of heterogeneous applications as well as a fair network resource allocation without congestion. Specifically, from the flow control perspective, the applications in communication networks can be broadly categorized as either elastic traffic or inelastic traffic based on their Quality of Service (QoS) requirements. This in turn implies that future communication networks will have to support a multitude of applications or services with different QoS characteristics, for both elastic and inelastic traffic. The majority of flow control strategies are mainly designed to cater for elastic traffic, thus far from sufficient in multi-service networks. It is primarily because the QoS utility function of inelastic traffic does not satisfy the strict concavity condition. It is also found to possibly yield an unfair bandwidth allocation even for elastic traffic. To address these limitations, this thesis is concerned with flow control and performance optimization for multi-service networks. Since heterogeneous applications are engaged, it is no longer desirable to allocate bandwidth simply according to the traditional fairness criteria in terms of bandwidth. Instead, networks are expected to guarantee the performance of different applications. The utility function is hence assumed to be strictly increasing only, which is applicable to elastic traffic and inelastic traffic, but may not necessarily be strictly concave as strongly required by the existing flow control approaches. In this thesis, we develop a utility fair flow control framework to allocate bandwidth such that their associated utilities achieve certain fairness criteria, i.e., the fairness is considered in terms of utility rather than bandwidth. Indeed, the utility-based fairness criteria generalize and strengthen the bandwidth-based ones. The framework is first considered in a general wired network setting, like the Internet, with both single-path routing and multi-path routing scenarios. It involves an efficient and fair flow control scheme, consisting of source algorithm and congestion feedback mechanism, to achieve utility proportional fairness and/or utility max-min fairness and provide QoS guarantees. In addition, a sliding mode control-based algorithm is also devised to obtain utility max-min fairness with low overhead and rapid convergence. Furthermore, the theory of utility fair flow control is adapted from wired networks to wireless networks, wireless sensor networks in particular. As the capacity region of wireless networks is usually unknown and complex, and critically depends on the underlying MAC and physical layers, it is not a direct application. We tackle the difficulties in both a layered and cross-layered manner. Through the layered approach, we formulate the flow control and resource allocation problem in heterogeneous sensor networks by characterizing the channel capacity and energy consumption properly, and then derive the corresponding algorithms and evaluate their performance. Through the cross-layered approach, we not only present an elegant queue backpressure-based algorithm that jointly optimizes transport layer flow control and MAC layer scheduling policy, but also design a first-ever flexible and practical transmission protocol that can efficiently handle elastic and inelastic traffic for wireless sensor networks.