Cross-layer design and methodology for satellite broadband networking
- Autores: David Pradas Fernandez
- Directores de la Tesis: María de los Ángeles Vázquez Castro (codir. tes.), Jérôme Lacan (codir. tes.)
- Lectura: En la Universitat Autònoma de Barcelona ( España ) en 2011
- Idioma: español
- Tribunal Calificador de la Tesis: Fotini-Niovi Pavlidou (presid.), Toufik Ahmed (secret.), Guy Lesthiévent (voc.), Michel Bousquet (voc.), Jérôme Lacan (voc.), Harald Skinnemoen (voc.)
- Materias:
- Texto completo no disponible (Saber más ...)
- Resumen
The demand for high-speed networking, driven mainly by the rapid expansion of the Internet, has been growing at an exponential rate. A wide range of technologically innovative wireline and wireless solutions already offer competitive broadband connectivity. Communication satellites emerge as an attractive solution in providing broadband connectivity to a variety of users thanks to its inherent global coverage. The broadcast nature of satellites makes them the natural choice for multicasting services, for interconnecting geographically distributed high-speed networks, and for providing multimedia services to both home and business users.
In order to accommodate the heterogeneity of users and multimedia traffic, broadband satellite networks are moving from the crowded Ku band to the more suitable Ka band. However, this higher frequency band imposes challenging channel conditions (attenuations up to dozens of dBs), for which novel technologies have already been adopted for upcoming broadband satellite networks. The main innovation has been the adoption of adaptive coding and modulation at the physical layer. This is the main driver of this thesis due to the crucial fact that such adaptivity makes traditional satellite system design totally inefficient. In addition, and as a natural way to increase the overall satellite capacity, multibeam technology is also assumed in such broadband satellite networks.
In this thesis, we focus on a different paradigm to address such new challenges due to adaptivity and multibeam technology. Such paradigm of system design is based on a joint optimization across layers of the protocol stack and has been the subject of a profusion of research work in the last years both for terrestrial and satellite networks. The fundamental idea behind this concept is the fact that adaptivity at the physical layer should be followed at upper layers in order to achieve efficient management of the system resources. Furthermore, it is the only way to comply with the stringent Quality of Service (QoS) of new applications services.
Instead of focusing on one aspect only of the cross-layer design for broadband satellite systems, we cover several aspects related to the networking optimization, including interoperability with terrestrial networks (thus also covering hybrid networking). Networking optimization deals with allocating resources efficiently, maximizing the throughput and assuring fairness among all the users, according to channel conditions measured at the physical layer. Delay, jitter and packet loss requirements (straightforward related to video and speech quality) for constrained services are specially challenging when they need to cope with the channel dynamics of the wireless satellite link. In order to meet these requirements, we take profit of adaptive technologies at higher layers (transport and application) such as layered coding and unequal error protection for videostreaming and adaptive codes for Voice over IP (VoIP).
Our efforts have been also focused on choosing the best methodology in terms of: 1) Selection of mathematical tools that best fit with each cross-layer approach. In particular, we have used convex optimization, Network Utility Maximization (NUM), game theory and linear programming.
2) Selection of the most suitable architecture. In particular, hybrid architectures including satellites are of special interest for wireless networks deployment in rural and remote areas. Moreover, conventional centralized vs novel decentralized architectures have also been traded-off.
3) Selection of the best approach to the cross-layer design, implementation, and performance evaluation. In particular, we have first set a taxonomy of approaches so as it is possible to choose the most suitable one for each scenario.
4) Selection, if needed, of truly available standardized tools, hence, ready-to-use for the realistic implementation of our developed schemes.
The contributions of this Ph.D. dissertation can be summarized as follows:
a) We study the resource allocation in multibeam broadband satellite systems. First, we propose to optimize the multicast transmission of the air interface for satellite and hybrid (satellite+terrestrial) networks. We focus on hierarchical allocation algorithms based on Superposition Coding (SC) and Adaptive Coding and Modulation (ACM). We prove that these algorithms can be adapted according to cross-layer information, depending on the channel conditions and users' mobility.
b) We address the challenging problem of capacity drop at Ka Band due to weather conditions, for both the forward and the return link of multibeam broadband satellite systems. In order to do so, we propose a cross-layer DiffServ architecture to optimize the interplay of lower layers, i.e. Physical (PHY), Medium Access Control (MAC) and Network (or IP). We formulate the optimization problem according to a NUM paradigm and we take advantage of Game Theory to investigate the impact of different performances of centralized and distributed approaches. We prove that our solutions maximizes capacity, while at the same time meets QoS requirements (in terms of delay/jitter).
c) The analytical methodology performed and the results obtained allow us to formulate the problem according to fair and efficient policies. We contribute with different tailored trade-offs, such as rate vs. delay or efficiency vs. fairness for different satellite weather conditions.
d) The problem of capacity drop and resources allocation in satellite multicast systems can be also faced up from the higher layers point of view, i.e. Transport and Application (APP) layers, which are able to interact with PHY-layer technologies (adaptive physical layer and Hierarchical Modulation) thanks to our cross-layer approach. Herein, we propose adaptive and scalable solutions in order to fulfil QoS requirements of real-time applications (such as VoIP/Videoconferencing) or strong loss-constrained applications such as videostreaming. In particular, we take advantage of Adaptive Multi Rates coding for VoIP, and Dependency Aware Unequal Error Protection (DA-UEP) and Scalable Video Coding (SVC) for videostreaming. We show that our cross-layer designs outperform noncross-layer designs in terms of video quality (PSNR) and speech quality (MOS).
e) We have also addressed and solved scenarios of particular interest, such as the hybrid transmission to rural areas, or the difficulty to give wireless services in tropical areas (due to strong weather conditions and terrain morphology).
