Abstract

Traditional Content Delivery Networks (CDNs) built with traditional Internet technology are
less and less able to cope with today’s tremendous content growth. Enhancing infrastructures
with storage and computation capabilities may help to remedy the situation. Information-Centric
Networks (ICNs), a proposed future Internet technology, unlike the current Internet, decouple
information from its sources and provide in-network storage. However, content delivery over in-network
storage-enabled networks still faces significant issues, such as the stability and accuracy
of estimated bitrate when using Dynamic Adaptive Streaming (DASH). Still Implementing new
infrastructures with in-network storage can lead to other challenges. For instance, the extensive
deployment of such networks will require a significant upgrade of the installed IP infrastructure.
Furthermore, network slicing enables services and applications with very different characteristics
to co-exist on the same network infrastructure.
Another challenge is that traditional architectures cannot meet future expectations for streaming
in terms of latency and network load when it comes to content, such as 360° videos and immersive
services. In-Network Computing (INC), also known as Computing in the Network (COIN), allows
the computation tasks to be distributed across the network instead of being computed on servers to
guarantee performance. INC is expected to provide lower latency, lower network traffic, and higher
throughput. Implementing infrastructures with in-network computing will help fulfill specific
requirements for streaming 360° video streaming in the future. Therefore, the delivery of 360° video and immersive services can benefit from INC.
This thesis elaborates and addresses the key architectural and algorithmic research challenges
related to content delivery in future networks. To tackle the first challenge, we propose algorithms
for solving the inaccuracy of rate estimation for future CDNs implementation with in-network
storage (a key feature of future networks). An algorithm for implementing in-network storage
in IP settings for CDNs is proposed for the second challenge. Finally, for the third challenge,
we propose an architecture for provisioning INC-enabled slices for 360° video streaming in next-generation
networks. We considered a P4-enabled Software-Defined network (SDN) as the physical
infrastructure and significantly reduced latency and traffic load for video streaming.