Abstract:Adaptive bitrate (ABR) using conventional codecs cannot further modify the bitrate once a decision has been made, exhibiting limited adaptation capability. This may result in either overly conservative or overly aggressive bitrate selection, which could cause either inefficient utilization of the network bandwidth or frequent re-buffering, respectively. Neural representation for video (NeRV), which embeds the video content into neural network weights, allows video reconstruction with incomplete models. Specifically, the recovery of one frame can be achieved without relying on the decoding of adjacent frames. NeRV has the potential to provide high video reconstruction quality and, more importantly, pave the way for developing more flexible ABR strategies for video transmission. In this work, a new framework, named Evolutional Video streaming Adaptation via Neural representation (EVAN), which can adaptively transmit NeRV models based on soft actor-critic (SAC) reinforcement learning, is proposed. EVAN is trained with a more exploitative strategy and utilizes progressive playback to avoid re-buffering. Experiments showed that EVAN can outperform existing ABRs with 50% reduction in re-buffering and achieve nearly 20% .

What problem does this paper attempt to address?

The problem that this paper attempts to solve is that in video stream transmission, the traditional Adaptive Bitrate (ABR) technology cannot further adjust the bitrate after making a decision. As a result, when facing network condition fluctuations, it either chooses an overly conservative bitrate, leading to low bandwidth utilization efficiency; or it chooses an overly aggressive bitrate, resulting in frequent rebuffering. These problems affect the user's Quality of Experience (QoE). To solve the above - mentioned problems, the author proposes a new framework - Evolutional Video streaming Adaptation via Neural representation (EVAN). This framework utilizes the Neural Representations for Videos (NeRV) technology and the Soft Actor - Critic (SAC) reinforcement learning algorithm to achieve more flexible bitrate decisions and reduce the occurrence of rebuffering. Specifically, EVAN can gradually play the video during the download process. Even if the model has not been fully downloaded, it can start playing a lower - quality version of the video, thereby avoiding rebuffering when the network conditions are poor. In addition, by adjusting the sparsity of the model, EVAN can dynamically select the optimal video quality according to the current network state, improving the user's viewing experience. The main contributions of the paper include: 1. Proposing a new framework, EVAN, for adaptive video streaming through NeRV, which can perform flexible bitrate selection under changing network conditions. 2. Introducing a progressive playback strategy that allows the reconstruction of the video while downloading the NeRV model, reducing the initial buffering time. 3. Through experimental evaluation, it is proven that EVAN has significant advantages over existing ABR methods in reducing rebuffering time and improving the user's QoE.

EVAN: Evolutional Video Streaming Adaptation via Neural Representation

Enhancement or Super-Resolution: Learning-based Adaptive Video Streaming with Client-Side Video Processing

Deep-Reinforcement-Learning-based User-Preference-Aware Rate Adaptation for Video Streaming

Improving Generalization for Neural Adaptive Video Streaming Via Meta Reinforcement Learning

Adaptive QoE‐based video delivery by optimizing the bitrate of the video course using network bandwidth and user bandwidth

Learning Accurate Network Dynamics for Enhanced Adaptive Video Streaming

Anableps: Adapting Bitrate for Real-Time Communication Using VBR-encoded Video

FedABR: A Personalized Federated Reinforcement Learning Approach for Adaptive Video Streaming.

Learning Tailored Adaptive Bitrate Algorithms to Heterogeneous Network Conditions: A Domain-Specific Priors and Meta-Reinforcement Learning Approach

NANCY: Neural Adaptive Network Coding methodologY for video distribution over wireless networks

NeuSaver: Neural Adaptive Power Consumption Optimization for Mobile Video Streaming

A robust adaptive video encoder based on human visual model

Adaptive Video Encoding For Different Video Codecs

HNeRV: A Hybrid Neural Representation for Videos

Fastconv: Fast Learning Based Adaptive BitRate Algorithm for Video Streaming.

Improving Quality of Experience by Adaptive Video Streaming with Super-Resolution

Buffer Awareness Neural Adaptive Video Streaming for Avoiding Extra Buffer Consumption.

ABRF: Adaptive BitRate-FEC Joint Control for Real-Time Video Streaming

Imitation Learning for Adaptive Video Streaming with Future Adversarial Information Bottleneck Principle

Video Adaptation in Abr Channels with Explicit Rate Feedback Control

Trace-Driven Optimization on Bitrate Adaptation for Mobile Video Streaming