ETD system

Electronic theses and dissertations repository


Tesi etd-06022015-171542

Thesis type
Tesi di laurea magistrale
Internet Infrastructure: a Multi-layer Network Analysis
Corso di studi
relatore Prof. Fagiolo, Giorgio
Parole chiave
  • Infrastructure
  • Internet
  • Network
Data inizio appello
Data di rilascio
Riassunto analitico
Nowadays, with the development of the modern Telecom systems, the structure of World interaction has become more interconnected and complex. The extraordinary revolution brought by the Internet era deeply modified the socio-economic interactions. The understanding of those changes requires new methods and approaches. In particular, transfers of information, that does not require face-to-face interaction, could be considered as flows carried by the global telecommunication network.
Therefore, our aim is to disentangle the structural characteristics of these flows among countries, with particular regard to features of the Internet backbone.
In order to disentangle the structural characteristics of the Internet infrastructure, we examine the Submarine Cable System. Our analysis represents a step forward in the understanding of this particular infrastructure. Therefore, we study the network structure among countries that proxies the information flows carried by those cables.

Indeed, the Submarine Cable System is the biggest component of the Internet backbone. This infrastructure is made of a massive network of fiber optic cables that lies deep under the seas.
Nowadays modern cables use optical fiber technology to carry digital data, which include telephone, Internet and private data traffic. The Submarine Cable System is part of an integrated network of different infrastructures that connect the world. For example, the other relevant components of this integrated network are satellites and radio frequency connections.
Submarine cables are faster and cheaper than satellites: modern fiber optic can transmit information at 99.7% the speed of light, and do not have problems of latency and bit loss that characterise satellite transmissions.
For these reasons, Submarine Cable System represents the biggest component of long distance information flows.
Moreover, the Submarine Cable System carries over 97% of intercontinental data traffic. Indeed the capacity used in the Submarine Cable System accounts for the 72% to Internet data, for the 27% to private networks owned by companies, and for only 1% to telephone calls. In our view, given these characteristics we consider cables capacity as a measure of information flows among countries, with particular regard to the Internet traffic.

In our analysis we use a dataset in gravity form, re-constructed from Google Tables. The data represent the network of Submarine Cable System, in which nodes are countries and links are cables with their specific capacity (Tb/s).
The work can be decomposed in three phases: an econometric analysis on four different representation of the Submarine Cable System, multi-layer comparison between Submarine Cable System and Cargo Ship movements (year 2007), and a layer comparison between the Cables and Air Flows (year 2012).

In the phase devoted to the econometric analysis, we study the existence of higher-order properties, unrelated to network characteristics, as country size, income, bilateral relationships, etc.
The first econometric analysis uses data on cable routing from 1989 to 2012. Each observation represents at least one cable construction (with its capacity) in a year between two countries.
The second investigation considers all the possible combinations of countries in cable routing from 1989 to 2012. Moreover, to offer a complete picture of the infrastructure evolution, the employed configuration of the data accounts also for the possibility of not having a cable routing in a year. In this perspective we can estimate the probability of a cable routing to appear.
The last two econometric investigations, take into account the two different layers (2007 and 2012). Therefore each layer represents two different states of the system. Indeed in every layer, the observations represent the existence, or not, of cable connections (with cumulative capacity) between two countries. This allows to estimate the probability to have connections in the particular year.

In the other two phases, we perform the multi-layer network analysis. Therefore we compare the network of Submarine Cables database with two other networks. The first comparison, for layer 2007, is done with respect to the network of cargo ship movements that, here, can be considered as a proxy of international trade. Indeed cargo-ship network accounts for up to 90% of the international exchange of goods (Woolley-Meza et al., 2011).
This comparison aims to understand if trade relations between two countries are linked to the need of more communication channels. Are trade flows able to influence the development of Telecom infrastructure? Indeed, since cargo movements may have geographical constraint similar to submarine cables, this may evidence which country has a strategic position.
The second comparison on layer 2012, instead, employs the air transportation database. The worldwide air transportation network supports the traffic of over three billion passengers travelling between more than 4000 airports on more than 50 million flights in a year (Woolley-Meza et al., 2011).
Because, in principle, the air network does not have the geographical constraints of submarine cables and cargo-ships, this network is used to measure those constraints. Indeed, Notwithstanding the wide use of Submarine cables, this characteristic may jeopardize the possibility for some countries to have good and cheap access to the Internet.
On the other side, people movements could be correlated with the relative information flows among countries. Therefore this comparison aims to disentangle this two features of the infrastructure.
The choice of this two layer years arise from data availability for cargo ship (year 2007) and air transportation (year 2012). Despite the problem of data availability, this two layers are able to capture the maturity of the Submarine Cable System.

In conclusion the network of Submarine Cables among countries is characterized by a large heterogeneity and by the presence of persistent inequalities in countries connectivity. These observed characteristics are, therefore, difficult to disentangle using standard econometrics techniques.

Our study of the network structure of long distance information flows, carried by the submarine fiber optic cables, offers a first picture about the complexity of global communication flows. The market structure and the infrastructural design, in our view, play a fundamental role in communication patterns.
On the other hand, the empirical analysis of its structure among countries may figure out some stylised facts on the global economy.
In conclusion, this empirical work points out the need of more attention for what concerns countries accessibility to modern telecommunication systems. Since a reliable and fast access to the internet telecommunication system is necessary for the development of low income countries, their peripheral position in the network may constitute a serious constraint to their chances of a better future.