User and content dependent resource management in a multi provider environment (Acreo)
|Acreo AB - Acreo AB, Kista
|Funding from Vinnova
|SEK 375 000
|January 2010 - December 2012
Purpose and goal
The present project will make an effort to bring end-to-end control of Internet based content services. This is performed by creating a self adapting network. In this context the customer Quality-of-Experience (QoE) is vital. Current technologies for the deployment of connectivity and best-effort based Internet services require however further technological development to meet the stability and reliability demands for successful deployments of real time services e.g. IPTV. The project will tackle the cross-layer problematics, from the choices of technology development and enhancement, to the service delivery workflows and networking aspects, and the development of a scalable prototype covering the access network end to end scope - monitor, analyze, control - for a successful service delivery following the given Service Level Agreements (SLAs) and end user requirements specifications.
Expected results and effects
New standardized solutions, built on open interfaces for network resource management, will be developed. The solutions will provide for better use of available resources in the access networks. It renders possible the detection and attendance of malfunctions that affects the users perceived service quality before the user is aware of the quality reduction. The architecture relies on the coordination of several modules, applications and services, for network management on different levels. The main beneficiary of the outcome of this project will be the end-users. A self-adaptive network that adjusts itself to the individual user will fully exploit the available network capacity. The customers agenda will change from content and connectivity buyer to just have to order delivery of certain content services. The user sees only the quality of the applications, and will have the opportunity to give direct response to the network, and this in real time. Operators can use the result to fulfil demands from governmental authorities, like PTS, to deliver a product with minimum characters instead of maximum, as is the state of today. It will be practically possible to set up SLAs for the content services at sale, because the SLA can be objectively evaluated according to per service measurements.
Planned approach and implementation
Control of applications will be a key issue for the system. When an application disturbance is detected (either by an end-user or a QoE monitor), the resource manager will need to check whether this disturbance comes from transmission impairments inside the access network or from the core network. If the access network is responsible, actions will be taken to resolve the impaired QoS. There are two ways to solve the problem, both including cross-layer issues: Physical layer adaptation and Scheduling of applications. In the project, we will establish a relation between QoE parameters measured on higher layers and the adaptation of parameters controlling physical layer behaviour. Scheduling of applications is a suitable action when the access link is congested. Measurements and control signals might not always be found in the same layer in the OSI reference model. Therefore, the project will involve research and development of cross layer techniques. The objective is to get control of which parameter in one layer that affects the uncontrollable parameter in another. Cross-layer network models will be investigated in a modified well-established network simulator, like ns-2 or OPNET. Also, another objective of this project is to work through a demonstrator, a case study, to investigate the dependencies, develop models, and attempt cross-layer control of the QoE. The network and resource management system is by nature distributed and modular. In order to meet the real-time demands, both monitors and resource managers should be intelligent entities and independently take decisions on actions depending on the current network situation. This calls for management control architecture, with control protocols, and security mechanisms. The security architecture is mainly out of the scope of this project; however, some issues as personal integrity will be addressed.