CONCERT: A Context-Adaptive Content Ecosystem Under Uncertainty

The objective of CONCERT is to develop a content ecosystem encompassing all relevant players which will be able to perform intelligent content and network adaptation in highly dynamic conditions under uncertainty. This ecosystem will have as basis emerging information-/content-centric networking technologies which support intrinsic in-network content manipulation. The project will consider uncertainty aspects in the following two application domains: (1) social media networks based on user generated content and (2) CDN-like professional content distribution (Content Distribution Networks). Three dimensions of uncertainties will be addressed: (1) heterogeneous and changing service requirements by end users, (2) threats that may have adverse impacts on the content ecosystem, as well as (3) opportunities that can be exploited by specific players in order to have their costs reduced. In order to manage and exploit the uncertainty aspects, CONCERT defines a two-dimensional content and network adaptation framework that operates both cross-layer and cross-player. First, the decision on any single adaptation action needs to take into account context information from both the content application layer and the underlying network. Second, we consider joint content and network adaptation in order to simultaneously achieve optimised service performance and network resource utilisation. Finally, some complex uncertainty scenarios require coordinated content and network adaptation across different ecosystem players. In this case, inconsistent or even conflicting adaptation objectives and different levels of context knowledge need to be reconciled and are key research issues. In order to achieve adaptation solutions capable of coping with different uncertainties, the project will develop advanced learning, decision-making and negotiation techniques. Learning is required for deriving accurate system behavioural patterns according to the acquired context knowledge. This will then drive decision-making functions for taking the most appropriate adaptation actions to address these uncertainties. Negotiation techniques are required for resolving potential tussles between specific content/network adaptation objectives by different players in the content ecosystem. The project will consider both centralised and distributed approaches in which learning and decision-making processes on adaptation actions can be performed either at the central adaptation domain controller or in a decentralised manner across multiple network elements. In the latter case, emerging information- /content-centric networks will become much more intelligent, with content-aware devices performing self- adaptation according to their own context knowledge but through coordination in order to achieve global near- optimality and stability.

More than 75% of the Internet traffic nowadays carries multimedia content (mostly, video). The CONCERT project an international collaborative basic research project involving University College London, University of Surrey, EPFL Lausanne, and AAU Klagenfurt set out to develop a (multimedia) content ecosystem able to perform intelligent content and network adaptation in highly dynamic conditions under uncertainties, with the goal, e.g., to deliver as high quality video content as possible. The uncertainty aspects considered were: 1) heterogeneous and changing service requirements by end users, e.g., use of diverse end devices over diverse wireless networks; 2) threats that may have adverse impacts on the content system, e.g., content going viral or attacks on networks; and 3) opportunities that can be exploited, e.g., the assistance by end devices in distributing highly popular content. Three application domains were addressed: 1) social media networks (user generated content); 2) professional content distribution; and 3) multi-view video streaming (content streams from multiple cameras on the same scene). As a technical basis, emerging Information-Centric Networking (ICN) technologies were often assumed for content delivery; ICN supports intrinsic in-network content caching and manipulation. We (AAU) decided to focus on one of the considered application scenarios, namely professional content distribution, more specifically on Dynamic Adaptive Streaming (DAS) of videos over Named Data Networks (NDN), a promising ICN variant. (In todays Internet, DAS is widely deployed over HTTP for video streaming, and then is denoted as DASH.) In this setting, we performed a variety of research and development studies, including: - Modelling applications, content usage and popularity patterns, and uncertainties in the considered multimedia content system. - Developing DAS-over-NDN simulation models (in the simulation tool ndnSIM) and performing manifold simulation experiments to obtain performance results. - Creating mathematical DAS-over-NDN performance models to derive upper performance bounds of DAS over NDN without and with caching of content in NDN. - Developing and evaluating a novel packet forwarding strategy in NDN, called Stochastic Adaptive Forwarding (SAF), that is able to adapt traffic in the network to congestion and failures and outperforms other forwarding strategies. Extending SAF to exhibit context awareness can, e.g., take into account the traffic type for the sake of prioritized delivery. - Building an ICN/NDN emulation testbed, i.e., hardware for performance experimentation. - In joint work with AI experts, researching a principled intelligent agent approach for NDN nodes, based on a general knowledge representation and reasoning technique. The FWF project resulted in more than 20 peer-reviewed publications. One of the guiding principles was to make all project results available to the scientific community, including open access publications, datasets, software, and hardware guidelines.