Resource Management for Broadband Satellite Networks

宽带卫星系统在多媒体应用的传输中起着重要的作用。在这个建议中，我们集中在GEO（地球同步）卫星网络。基于GEO的系统通过一颗卫星向特定区域提供连续的服务。它们在公共和私有互联网中扮演着越来越重要的角色，主要是由于它们的大地理覆盖范围、固有的广播能力和快速部署。它们在支持数据、音频和视频流方面具有吸引力；批量数据传输，例如软件更新或Web缓存的分发；以及涉及有限交互性的应用，如远程学习。它们还具有吸引力，可以为那些无法到达地面网络的用户，或对广播/多播应用或快速部署有特殊需求的用户提供宽带接入。在本提案中，我们将重点介绍下一代卫星网络，该网络由一颗（或多颗）具有某种形式的机载处理（OBP）的地球卫星组成。关于宽带卫星网络的两个基本问题是1)端到端资源管理和2)网络可用性。资源管理是在提供足够效率的同时向服务提供可接受的服务质量（QoS）的关键。它是任何卫星系统的核心，无论是弯管还是机载处理（OBP）卫星系统。实际上，为了给突发应用提供灵活性和效率，多址上行链路的容量必须使用带宽按需（BoD）方案进行管理。我们将网络可用性定义为网络在降级期间向某些应用程序提供某种级别服务的能力。大多数下一代系统将使用Ka波段，该波段以其非常困难的传输特性而闻名。已经提出了许多解决方案，包括可变速率编码、功率控制和/或可变调制，以使该频段对宽带应用友好。这个建议不会直接研究那些1/2层机制，而是研究它们的使用如何影响BoD和更普遍的端到端资源管理。因此，我们提案的最终目标是设计和评估下一代卫星网络的端到端资源管理，这些网络与空中接口设计者提出的一系列解决方案有效地互连，以提高传输的可用性。特别是，我们想要了解效率和系统复杂性之间的权衡。为了达到这一最终目标，我们将研究：大型GEO OBP系统的静态端到端资源管理。术语静态指的是编码和调制是固定的。这里的重点将是开发和评估可扩展、健壮、灵活和保护OBP（即空中交换机）的机制。我们将在之前为大型GEO弯管系统开发可扩展、强大、灵活的端到端资源管理的领域开展广泛的工作。网络可用性。卫星网络设计者主要关心的问题之一是，尽管该频段具有困难的特性，但如何为用户提供“适当”水平的服务。然而，“适当的服务级别”的含义并不十分清楚，因为它可能因用户或应用程序而异。大多数研究都集中在联机可用性上，而没有考虑到联网和应用方面。我们相信，协调底层与上层将为我们提供一个灵活且经济高效的解决方案，允许系统根据用户的需求和支付意愿向用户提供不同级别的网络可用性。因此，网络可用性可以看作是aQoS需求，应该这样定义和研究。为包含1/2层机制的系统设计端到端资源管理，动态提高上行链路可用性，为用户提供灵活、经济的服务。我们需要提出解决方案并对其进行评估，以确保复杂性的增加在效率方面是值得的。我们还将提出和研究一些与按需计费相结合的动态定价方案，以便在总可用容量因链路损害而变化的情况下，允许用户表明他们为容量付费的意愿。

Broadband satellite systems have an important role to play in the delivery of multimedia applications.In this proposal we concentrate on GEO (geo-synchronous) based satellite networks. GEO based systemsdeliver continuous services to a specific region with a single satellite. They play an ever-increasing role inthe public and private Internets, due mostly to their large geographic coverage, inherent broadcastcapabilities and fast deployment. They are attractive to support data, audio and video streaming; bulk datatransfer such as software updates or dissemination of Web caches; and applications involving limitedinteractivity such as distance learning. They are also attractive to provide broadband access to users who areeither beyond the reach of the terrestrial network, or have particular needs for broadcast/multicastapplications or fast deployment. We will focus in this proposal on Next Generation Satellite Networks that comprise one (or more) GEOsatellite with some form of onboard processing (OBP). Two of the essential issues regarding broadbandsatellite network are 1) end-to-end resource management, and 2) network availability. Resource management is key to deliver acceptable Quality of Service (QoS) to services while providingadequate efficiency. It is central to any satellite system, be it a bent pipe or an onboard processing (OBP)satellite system. Indeed, in order to offer flexibility and efficiency to bursty applications, the capacity of themultiple access uplink has to be managed using some bandwidth on demand (BoD) scheme. We define network availability as the ability for the network to offer some level of services to someapplications even during degraded periods. Most of the next generation systems will be using the Ka band,which is known for its very difficult transmission characteristics. Many solutions involving variable ratecoding, power control, and/or variable modulation have been proposed to make this band friendly tobroadband applications. This proposal will not study directly those layer 1/2 mechanisms but rather studyhow their use affects BoD and more generally end-to-end resource management. Hence the final objective of our proposal is to design and evaluate end-to-end resource managementfor the Next Generation of Satellite Networks that interworks efficiently with the range of solutionsproposed by the air-interface designer to improve the availability of the transmission. In particular, wewant to understand the trade-off in terms of efficiency and system complexity. To achieve this ultimate objective, we will study:1. Static end-to-end resource management for large GEO OBP systems. The term static refers to thefact that the coding and modulation are fixed. The focus here will be to develop and evaluatemechanisms that are scalable, robust, flexible and protect the OBP (i.e., the switch in the sky). Wewill build upon our extensive previous work in the domain where we developed scalable, robust,flexible end-to-end resource management for large GEO bent pipe systems.2. Network availability. One of the main concerns of satellite network designers is to offer an"appropriate" level of service to the users in spite of the difficult characteristics of the frequencyband. However it is not very clear what is meant by "an appropriate level of service" since it mayvary from a user to another or from an application to another. Most of the studies have focussed onlink availability without taking into account the networking and applications aspects. We believethat coordinating the lower layers with the upper layers will give us a flexible and cost-efficientsolution allowing the system to deliver different levels of network availability to the usersdepending on their need and their willingness to pay. Hence network availability can be seen as aQoS requirement and should be defined and studied that way.3. Designing end-to-end resource management for a system comprising layer 1/2 mechanisms todynamically improve the uplink availability so as to offer flexible and cost efficient services to theusers. We need to propose solutions and to evaluate them to ensure that the increase in complexityis worthwhile in terms of efficiency. We will also propose and study some dynamic pricing schemesintegrated with the BoD so as to allow the users to indicate their willingness to pay for capacity in acontext where the total available capacity is variable due to link impairment.