الفهرس 
IntroductionOnly 14 pages are availabe for public view
Abstract
This thesis proposes a parallel multi-hop routing protocol for 5G network using cloud computing and High Performance Computing (HPC). Efficient 5G network management depends on the proposed protocol. The parallel implementation is tested by different communication network sizes using two platforms, ERI-HPC and virtual cluster on ERI-OpenStack cloud.
This thesis provides five contributions in the following points:
• Developing a parallel protocol for selected serial multi-hop routing algorithm as a suitable communication algorithm that meets diversified 5G service requirements.
• Testing the parallel protocol on real ERI-HPC cluster system and deploying virtual cluster system using ERI-Openstack cloud.
• Analyzing and comparing the efficiency of cloud and HPC clusters in the speed-up and performance enhancement provided by the parallel multi-hop routing protocol.
• Evaluating the scalability of (HPC and cloud) parallel platforms and the impact of ultra small cells densification using different network sizes.
• Setting recommendations for the adoption of cloud computing and high performance computing platforms to be utilized by 5G protocols.
This thesis is divided into six chapters in addition to the lists of contents, tables and figures as well as list of references and appendices.
• Chapter 1: This chapter is the introduction of the thesis and provides a brief for thesis motivation to adopt HPC and cloud computing platforms for 5G network management challenges. This chapter also includes thesis contributions and outlines.
• Chapter 2: This chapter demonstrates an overview for HPC systems as parallel computing platforms and their architectures (Shared-memory, Distributed-memory). In addition to the parallel programming models and some metrics for parallel implementation performance. It also provides an overview of cloud computing and its characteristics, and introduces OpenStack architecture and its services to understand the virtualization overhead among virtual machines. Then highlights some of cloud challenges, such as HPC over cloud and cloud-based network in 5G.
• Chapter 3: This chapter discusses 5G and its key technologies such as ultra dense networks and cloud radio access networks and also presents 5G network management as one of vital challenges, especially efficient multi-hop routing protocol in 5G in the light of the related work and thesis objectives.
• Chapter 4: This chapter shows a detailed illustration about the serial multi-hop routing algorithm and the steps for parallel implementation. The serial algorithm is divided into two main functions. Firstly, it finds the widest paths from a given source communication node to destination node by calling Bellman-ford algorithm once. Secondly, it selects the path whose links have the smallest maximum-power. Bellman-Ford algorithm is computational intensive procedure that critically needs parallelization to speed up, especially in UDNs. The parallelization process begins with performance and code analysis for the serial algorithm to highlight hotspot tasks. Then, it is followed by concurrency analysis for tasks to expose data dependencies, and finally parallel implementation is used by suitable parallel programming model. Parallel implementation depends on the underlying hardware infrastructure of the system, and a ”distributed-memory system” is used because it is scalable in terms of the number of processors. The master-slave model is used as a programming model for a distributed memory system, and it consists of three main tasks: pre-processing, computation, and post-processing. First, the master processor set up the data and configures the number of hops in the network to be sent to the slave processors using collective communication. Secondly, each processor initializes and processes its own distributed data and then updates the local data in every hop. Each slave processor sends its results to the master processor. Finally, the master processor, in the post-processing, chooses the path from a given start node to the destination in the network, taking into account the lowest maximum transmit power along the path.
• Chapter 5: This chapter explains the platforms that used to test our parallel implementation. The results of each platform are presented and compared with each other.
• Chapter 6: This chapter highlights thesis conclusions and recommends some future work.
Key words: Cloud computing, High Performance Computing (HPC), 5G routing protocol, 5G backhauling, Ultra Dense Network (UDN), Cloud Radio Access Network (C-RAN), Heterogeneous Cloud Radio Access Network (H-CRAN), Parallel programming, Message Passing Interface (MPI).