IP-friendly cross-layer optimization of adaptive satellite systems

Objectives

The project focuses on the novel DVB-S2 standard, which uses channel state information to adapt coding and modulation to the time-varying physical layer conditions experienced by a specific user. The goal of the project is to investigate how providing cross-layer information may help upper layers to take full advantage of this innovative physical layer solution. The study should provide a first feedback on upper layer behaviour and performance in this novel system, both with and without applying cross-layer solutions. The study will ultimately propose a cost/benefit analysis of the studied technology respect to the traditional ones.

 

The evaluation of the different considered solutions will be performed with the aid of an in-lab emulator of DVB-S2/DVB-RCS systems to which real user equipment will be connected to measure performance parameters with real end-to-end applications. For such evaluation, several scenarios will be tested covering the different cross-layer study cases.

In short, expected project results are to:

  • Assess performance of different real-time and non real-time applications over a realistic DVB-S2/DVB-RCS scenario
  • Evaluate and propose cross-layer techniques which may led to a relevant performance increase
  • Produce an emulation tool, based on a previous developed one by Indra and thanks to a scalable and modular design, shall be an effective starting point for future cross-layer techniques evaluation tasks or DVB-S2 related design options performance assessments

Challenges

The key issue of the cross-layer concept is to improve overall performance by exchanging relevant information between layers.

Benefits

The study proposed by ESA responds to the current awareness that due to the inexistence of communication across layers improvements introduced for lower layers may not be fully exploited by upper layers. In this sense, the project shall identify cross-layer techniques which allow taking the full benefit from novel DVB-S2/RCS ACM system features.

Features

The study considers a broadband access Ka-band satellite system scenario, using a transparent satellite. The system is composed of a number of gateways, giving service to user terminals distributed over different beams (multi-star network).

 

The forward link, i.e. from the gateway to the user terminal, is DVB-S2 compliant, while the return link, i.e. from the user terminal to the gateway, is based on DVB-RCS.

As studied cross-layer optimizations will focus on the effects of rain events on the overall system performance, the physical channel is modeled in great detail. Variations of SNIR experienced by user terminals are calculated using models providing probability distributions of the rain attenuation for any latitude and longitude and considering among other aspects fading frequency of occurrence, fading duration and how long it takes since a rain event starts till it reaches its maximum intensity.

Four study cases are analyzed to detect cross-layer potentialities:

  • End-to-end TCP: This study case focuses on cross-layer mechanisms which may improve well-known TCP performance problems, for both short and long-lived connections.
  • TCP Splitting: The aim is to optimize Performance Enhancement Proxy (PEP) functionality (used to improve TCP performance) with cross-layer information (see FIGURE 1 for an example reference scenario).
  • Real-time services: This study case focuses on real-time applications (VoIP, videoconference) and their particularities.
  • Queuing and Quality of Service: Adaptive Coding (and Modulation) in DVB-S2 and DVB-RCS requires a tight interaction between IP and layer 2 queuing and scheduling, in order to take advantage of users with better channel conditions while at the same time assuring QoS.

During the project the two study cases and cross-layer mechanisms which are considered most promising will be selected, analysed and evaluated in depth.


click for larger image

figure 1: TCP Splitting Reference Scenario

Plan

The tasks within this project are performed in two different stages:

 

The first stage is focused on performing a detailed analysis of cross-layer approaches for various study cases (TCP, TCP with PEP, real-time traffic, queuing/scheduling). This phase includes:

  • A review of cross-layer mechanism applied to satellite and terrestrial systems, covering the state of the art of these technologies. Analysis of the suitability for the reference satellite scenario.
  • Definition of all potential study cases, the needs and justification for cross-layer design, the criteria of evaluation, the preliminary assessment of performances, the impact on the system architecture and the selection of cases to be further investigated.
  • In depth detailed analysis of most promising cross-layer mechanisms for each study case, including performance evaluation comparing to the reference system.

The output of this stage is the selection of a promising cross-layer technique for two study cases and a detailed specification and design of the cross-layer mechanism.

The second stage is centred in the emulation of selected cross-layer mechanisms and the analysis of emulation results. Two distinct phases can be distinguished:

  • Emulation platform development. Subtasks include emulator specification, architectural & detailed design, development and validation.
  • Test campaign and critical analysis. This phase encompasses test campaign definition for cross-layer performance evaluation, actual test campaign and results evaluation and analysis.

The result of this phase is, on one hand, an operative emulation platform for DVB-S2/RCS systems and, on the other, a cross-layer evaluation document showing the performance results obtained with selected cross-layer mechanisms. Results will be presented in a Final Presentation to be held in the Agency's premises.

Current status

The performance evaluation of the two selected cross-layer mechanism (QoS queuing and TCP splitting) has shown that cross-layer approaches significantly provide many improvements in comparison to layered approaches. Moreover, they provide an elegant approach to dealing with adaptive physical layer systems, especially adaptive Broadband Satellite Multimedia (BSM) systems.

The emulator was quite important in providing in-depth validation to the cross-layer optimizations but it could also prove quite useful in testing applications over DVB-S2/RCS platforms.

Future work should follow, both in developing new systems based on cross-layer optimizations as well as upgrading current ones, which would also require evolving these optimizations into deployment grade maturity. Methodologies developed during this project also open the way for new improvements in other areas and applications that were outside of the scope of the project such as mobile scenarios.

Status date

Thursday, September 4, 2008 - 11:57