Artificial Intelligence in HTTP Adaptive Streaming: A review within the evolution of cable television distribution

This article reviews the evolution of cable television distribution, from coaxial networks to virtualized architectures based on the internet protocol and the cloud, and examines how artificial intelligence has transformed the bitrate adaptation (ABR) algorithms that underpin HTTP Adaptive Streaming (HAS) throughout that transition. Peer-reviewed academic literature and industry documentation were reviewed, and the managed and best-effort distribution models on which these algorithms operate were characterized, alongside the ongoing transport-layer shift from TCP to QUIC. The analysis identifies four fronts where artificial intelligence intervenes at different points of the distribution chain, bitrate decision-making through deep reinforcement learning, content-aware encoding, client-side super-resolution, and energy-aware adaptation, each backed by quantitative gains of up to 43% in quality of experience over a strong reinforcement-learning baseline and up to 57% in energy consumption relative to state-of-the-art algorithms. The article concludes that the sophistication of these models is systematically in tension with the real-time decision requirement of live streaming, which positions coordination across fronts, rather than the isolated improvement of each one, as the central challenge facing the industry.