You are here: Home / Project description

Project description

The Future Internet orientation to information and in particular distribution of content is recognized by many analyses and studies. Therefore Content Oriented Networking (CON), or Information Centric Networking (ICN) paradigms are intensively studied and discussed in research groups, academia, industry, etc. Revolutionary or evolutionary approaches are proposed and evaluated in many publications. The DISEDAN project proposes an evolutionary solution to enhance the content distribution via Internet.

DISEDAN will research on an effective solution for the multi-criteria hard problem of best content source (server) selection, considering user context, servers availability and requested content. To this end, DISEDAN will introduce a new concept based on (1) a two-step server selection mechanism (at Service Provider and at End User) making use of innovative algorithms that consider context- and content-awareness and (2) a dual adaptation mechanism consisting of Media adaptation (also called media flow adaptation) and content source adaptation (by switching the streaming server) when the transmission suffers degradation. In this project by Service Provider it is understood a business entity/actor which delivers the content services to the users and possibly manages the transportation network.

A critical issue in collaborative systems (also named ecosystems) between different actors is the distribution of restricted information, especially from the point of view of the Service Provider (SP). Therefore, we conceive an ecosystem where the SP should not be obliged to provide sensitive information to third-parties. In fact, SP can only provide a list of servers (url addresses) where the user may find the desired content, which considers the policies of the SP and the current state of the network and servers. Obviously, if the SP decides to provide information about the state or characteristics of servers and network (e.g., network path), then the user is able to perform better selection and the quality of experience increases. Making this information optional, additionally allows the DISEDAN client to be standalone, providing benefits to the user without the need to modify existing solutions on the SP side.

When the user requests a Multimedia content, the Service Provider sends an xml file which contains Media Description. This file is updated (from the static xml file) for each user request by considering the user profile, the policies of the SP for this user and other information at the SP side (e.g., state of the servers and possibly network-related information). The list of content servers and, potentially, the information about state or characteristics of the servers and the state of the network (if applicable) is written inside the xml file. Also caching server url addresses can be added. The list may be optionally ordered following some desired metrics.

The strategy for content source selection needs fundamental understanding of the parameters (policies, user profile, network measurements, etc.) and algorithms used, so we will go more deeply into multi-objective optimization, and especially we will investigate enhanced multi-criteria decision algorithms delving into the research line just started by us.

When the user’s application receives the Media Description file, it performs the final selection of content server and possibly the network path. This decision can be based on user context or can simply be the first content server in the ordered list.

During the transmission of consecutive chunks, the user’s application can automatically change the rate of the content stream (as it is actually performed in adaptive streaming) and also can automatically switch the content server from where the content is streamed. Often, there exists an overload problem at the server side; in this case, by changing the rate of the chunk, the problem is only partially solved. In this situation, the second adaptation (content source) could maintain the quality of the transmission without reducing bitrate. This is the dual adaptation feature.

Our studies will be directed to develop clear rules for deciding the adaptation to be executed in each case. This decision should consider the monitoring of the stream performed by the application and also some other information provided by the SP, when available.

Currently, the monitoring of the stream at the user application is limited to the control of the bitrate of the streamed content. We will consider also queue length and possibly other parameters (e.g. related to TCP sockets). With all of them, we will specify the rules for deciding about which adaptation to be performed as well as how this adaptation must be performed (parameters). Simulation and analytical studies will be provided in this phase. Moreover, these rules should offer the possibility of increasing bitrate of the following chunk in the case when the last chunk/s were delivered with lower bitrate than the one that the network can offer in this moment (due to, e.g., previous instantaneous overload of the network).

The proposed functionalities will be developed for Dynamic Adaptive Streaming over HTTP (DASH) and an open-source library for easy implementation of these functionalities in video players will be implemented. The library will be able to function as a plugin for the VLC application. Moreover, an implemented solution of MPD server with Multi-BaseURL and advanced server selection will be provided.

Briefly, the objectives of the project are:

  • to conceive a system for two-phase selection of content server: at the SP and at the user sides;
  • to propose new enhanced algorithms for server selection on the basis of information belonging to the SP and information belonging to the user;
  • to analyse dual adaptation capabilities in order to optimize the possibility of adapting Media or content source;
  • to specify DASH Client extension with enhanced server selection algorithm and dual adaptation capabilities;

to provide open-source DASH library compatible with, among others, VLC’s plugin interface.