An SDN-based device-aware live video service for inter-domain adaptive bitrate streaming

A Contemporary Survey on Live Video Streaming from a Computation-Driven Perspective

“…In addition to the eiciency achieved by mCast, NFV, SDN, DNN, and ABS approaches, the optimization problems among one or several of the four key parameters were analyzed. In Section 5.8, we reviewed the security and privacy perspectives within the LVS scope, which are based on blockchain technologies as well as non-blockchain [32] and [17], respectively [122,123] Redundant nodes for clouds Achieved annual downtime of 83.798 h and 49.05 h from [123] and [122], respectively CDN, ABS using HLS Achieved 11.58% improvement in throughput and 0.25% degradation in packet loss ratio [5] CDN, ABS using Hampel ilter Achieved better 20% true positive rate than baselines [102] Large-scale CDN Achieved a better than 70% throughput than conventional benchmarks E2E Latency [108] HTTP/2-based LVS Improved video quality, smoothness, and E2E latency performance [105] Packet bufering delay was minimized to 2 seconds [151] A novel ABS proposal Obtained closed-form expression for the average bufering delay [171] MultiLevel ABS algorithm Achievable E2E latency was reduced to only 0.1 second [156] ABS using DRL Estimated optimal video bitrate for ultra-low-latency intention [69] ABS using QARC and DRL Improved video quality by 18ś25 % and decreased the E2E latency by 23ś45 % [98,134] ABS using SVC Provided coding bitrate decrements by 38% leading to a reduction in the E2E latency [103] ABS using HEVC Achieved approximately double encoding eiciency of the SVC with 56.7% [145] ABS using SHVC Gained around 20% decoding speedup in signiicantly diminishing the E2E latency [142] ABS using VVC Provided a low bufering delay of approximately 0.21 seconds [18] Edge computing Minimize the E2E latency and response time as well as provide the resource savings [177] Endśedgeścloud coordination Satisied low-latency, accurate analytic, LVS quality, and computing resource constraints [10] Fog architecture Low-latency support as well as ultra-reliability communication [110] MEC, lexible transcoding ABS Latency of 15ś75 ms under limited computing, caching, and bandwidth constraints Network QoS/QoE [48] Adaptive SDN architecture Achieved end-user QoE improvement and high-accuracy MOS estimation [5] ABS with large-scale CDN Ofered 92% ac...…”

“…A higher video bitrate is expected to achieve a higher video quality for the LVS experience. Owing to the unstable characteristics of many parameters afected by video quality, many authors have attempted to adopt the ABS concept [5,57,102,149] to guarantee video quality as high as possible by adapting to bandwidth/throughput luctuations owing to changes in network condition. In the ABS method, a transmitted video is simultaneously encoded at various levels of bitrates, from which these streams are divided into multiple segments and stored on an HTTP server, and the client will be assigned appropriate bitrate segments considering the network conditions [16,104].…”

“…The numerical results in [5] have revealed that the Hampel ilters can achieve 92% accuracy of the QoE impairment, which introduces an approximate 20% improvement in the true positive rate compared with baseline methodologies. In the same context, [102] considered a large-scale prototype with up to 500 LVS from users dynamically adapted at various bitrates under CDN integration. The system yielded an average user throughput increase of up to 70% compared with conventional benchmarks.…”