Transport Layer solution for bulk data transfers over Heterogeneous Long Fat Networks in Next Generation Networks

This compendium thesis focuses its contributions on the learning and innovation of the New Generation Networks. That is why different contributions are proposed in different areas (Smart Cities, Smart Grids, Smart Campus, Smart Learning, Media, eHealth, Industry 4.0, among others) through the application and combination of different disciplines (Internet of Things, Building Information Modeling, Cloud Storage, Cybersecurity, Big Data, Future Internet, Digital Transformation). 

Specifically, the sustainable comfort monitoring in the Smart Campus is detailed, which can be considered my most representative contribution within the conceptualization of New Generation Networks. Within this innovative monitoring concept, different disciplines are integrated in order to offer information on people's comfort levels. . This research demonstrates the long journey that exists in the digital transformation of traditional sectors and New Generation Networks. 

During this long learning about the NGNs through the different investigations, it was possible to observe a problematic that affected the different application fields of the NGNs in a transversal way and that, depending on the service and its requirements, it could have a critical impact on any of these sectors. This issue consists of a low performance operation during the exchange of large volumes of data on networks with high bandwidth capacity and remotely geographically separated, also known as Elephant networks, or Long Fat Networks (LFNs). Specifically, this critically affects the Cloud Data Sharing use case. That is why this use case and the different alternatives at the transport protocol level were studied. For this reason, the performance and operation problems suffered by layer 4 protocols are studied and it is observed why these traditional protocols are not capable of achieving optimal performance.  

Due to this situation, it is hypothesized that the introduction of mechanisms that analyze network metrics and efficiently exploit network’s capacity meliorates the performance of Transport Layer protocols over Heterogeneous Long Fat Networks during bulk data transfers. 

First, the Adaptive and Aggressive Transport Protocol (AATP) is designed. An adaptive and efficient transport protocol with the aim of maximizing its performance over this type of elephant network.. The AATP protocol is implemented and tested in a network simulator and a testbed under different situations and conditions for its validation. 

Once the AATP protocol was designed, implemented and tested successfully, it was decided to improve the protocol itself, Enhanced-AATP, to improve its performance over heterogeneous elephant networks. In addition, in order to upgrade the behavior of the protocol, its fairness system is improved for the fair distribution of resources among other Enhanced-AATP flows. Finally, this evolution is implemented in the network simulator and a set of tests are carried out. 

At the end of this thesis, it is concluded that the New Generation Networks have a long way to go and many things to improve due to the digital transformation of society and the appearance of brand-new disruptive technology. Furthermore, it is confirmed that the introduction of specific mechanisms in the conception and operation of transport protocols improves their performance on Heterogeneous Long Fat Networks.