USING FEDERATED MACHINE LEARNING FOR ACCESS POINT SELECTION IN THE HETEROGENEOUS NETWORK

The problem of selecting a wireless access network in a highly heterogeneous environment has been analyzed and solved. A network selection model based on the analysis of a wireless network environment using a federated reinforcement machine learning system is proposed. A model has been developed to estimate the theoretical average capacity available to the user in a highly heterogenic access network. The effectiveness of the proposed method was evaluated using a series of experiments. The article is concluded with a discussion regarding the applicability of the proposed method for IMT-2020 and IMT-2030 networks.

Toward IoT fog computing-enabled system energy consumption modeling and optimization by adaptive TCP/IP protocol

Nowadays, due to the fast-growing wireless technologies and delay-sensitive applications, Internet of things (IoT) and fog computing will assemble the paradigm Fog of IoT. Since the spread of fog computing, the optimum design of networking and computing resources over the wireless access network would play a vital role in the empower of computing-intensive and delay-sensitive applications under the extent of the energy-limited wireless Fog of IoT. Such applications consume considarable amount of energy when sending and receiving data. Although there many approaches to attain energy efficiency already exist, few of them address the TCP protocol or the MTU size. In this work, we present an effective model to reduce energy consumption. Initially, we measured the consumed energy based on the actual parameters and real traffic for different values of MTU. After that, the work is generalized to estimate the energy consumption for the whole network for different values of its parameters. The experiments were made on different devices and by using different techniques. The results show clearly an inverse proportional relationship between the MTU size and the amount of the consumed energy. The results are promising and can be merged with the existing work to get the optimal solution to reduce the energy consumption in IoT and wireless networks.

Dynamic Bandwidth Management for TCP Flows in a Diffserv-Enabled Mobile Wireless Access Network

With the soaring demand to provide global mobility for wide range of services, a growing trend for wireless access networks is to support multiple radio technologies that can be achieved efficiently by using TCP/IP protocols in the access network. In the thesis, we consider a mobile wireless access network where DiffServ is deployed as the QoS solution and Mobile IP is employed as the handover protocol. We first conducted a study on the impact of handover on DiffServ flows. Then we introduced a transient service level for handover flows and propose our QoS scheme and admission control algorithms for handover flows, which protect local flows from losing bandwidth to handover flows by separating the two flows into different service classes. We also proposed a service upgrade algorithm to upgrade the service level of handover flows based on the dynamic information of bandwidth utilization and different service upgrade priorities. To guarantee the proper provisioning for each service class, we proposed a dynamic bandwidth-provisioning algorithm that allows dynamic adjustment of bandwidth allocations to different service classes by adjusting their respective weights configured at the scheduler. We evaluated the feasibility of our QoS scheme and algorithm by simulating different handover situations and results show that the proposed scheme is viable under variety of provisioning scenarios.

Dynamic Bandwidth Management for TCP Flows in a Diffserv-Enabled Mobile Wireless Access Network

With the soaring demand to provide global mobility for wide range of services, a growing trend for wireless access networks is to support multiple radio technologies that can be achieved efficiently by using TCP/IP protocols in the access network. In the thesis, we consider a mobile wireless access network where DiffServ is deployed as the QoS solution and Mobile IP is employed as the handover protocol. We first conducted a study on the impact of handover on DiffServ flows. Then we introduced a transient service level for handover flows and propose our QoS scheme and admission control algorithms for handover flows, which protect local flows from losing bandwidth to handover flows by separating the two flows into different service classes. We also proposed a service upgrade algorithm to upgrade the service level of handover flows based on the dynamic information of bandwidth utilization and different service upgrade priorities. To guarantee the proper provisioning for each service class, we proposed a dynamic bandwidth-provisioning algorithm that allows dynamic adjustment of bandwidth allocations to different service classes by adjusting their respective weights configured at the scheduler. We evaluated the feasibility of our QoS scheme and algorithm by simulating different handover situations and results show that the proposed scheme is viable under variety of provisioning scenarios.

Optical Transport Network to support Digital Communications

The objective of this work is to propose a solution to the problem of the lack of telecommunication services, with adequate quality of service and coverage, in the town of Choclococha, Huancavelica-Peru. For this purpose, the design and evaluation of an optical link as a backbone network and a wireless access network is presented. It is an applied and experimental research in which performance parameters were evaluated. The independent variable is the optical transport network and the dependent variable is data communication. The latter is defined by two indicators: throughput and signal reception power at the users. Both indicators were measured in a sample of the population, obtaining results that were validated with respect to the expected values according to specifications and regulations. Thus, the measured throughput exceeds the 40% data rate guaranteed by the telecommunications operator. As for the reception power, it exceeds the sensitivity level of the receiving equipment. Therefore, it is concluded that the proposed solution supports communications with optimum speed and coverage.