Currently, most consumer videos in Standard Dynamic Range (SDR) format undergo media compression. High Dynamic Range (HDR) reconstruction from such videos remains challenging due to the existence of block-based, entangled luminance and texture distortions caused by sequential luminance quantization and block-by-block media encoding. Specifically, block-based luminance banding caused by quantization and block blur introduces fake block contours; meanwhile, block-based texture distortions from cross-block moving objects are inconsistent and cause luminance shifts. Facing these mutually incurred, entangled distortions, most existing methods neglect one or both aspects, as well as their block-based nature, and the remaining multi-stage ones suffer from error propagation. Therefore, we propose a novel Parallel Block-level Enhancement HDR Framework (PBE-HDR) for HDR video reconstruction, conducting independent parallel compression-specific, block-based luminance recovery and texture restoration without cross-distortion entanglement. Without texture involvement, it tackles luminance banding via inter-block global Multi-frame Luminance Equalization (MLE) and intra-block progressive luminance compensation guided by a block prior termed Block Luminance Knowledge (BLK). Meanwhile, it employs implicit distorted-texture enhancement through a multi-scale mutual shuffle-and-attention update mechanism, with dual-frame mutual enhancement for luminance-error neutralization. It also employs pixel shuffling combined with channel-wise attention across multiple scales for implicit inter-/intra-block texture composition and reconstruction, which is infeasible for explicit alignment methods. The final adaptive fusion produces the entanglement-refrained results. Extensive experiments demonstrate that the proposed method outperforms state-of-the-art reconstruction methods.
Volumetric video streaming has recently emerged as a key immersive media technology, enabling content consumers and users to have immersive six degrees of freedom (6DoF) experiences. However, smooth streaming of volumetric content with an enhanced sense of presence is challenging because of the large amount of volumetric data, complex pre-processing, unreliable network transmission, and resource-intensive decoding operations. Current streaming solutions often fail to optimize real-time rendering of the volumetric content and therefore lead to poor streaming experiences, further exacerbated by inefficient bitrate adaptation policies. In this paper, we introduce an MPEG-DASH compatible adaptive volumetric video delivery solution that includes: 1)an interactive web-based volumetric video streaming player that incorporates parallel Draco decoding capabilities and employs the WebGL-based Three.js framework to perform level of detail rendering and playback; 2)an MPEG-DASH-compatible volumetric video delivery framework that uses Draco-encoded frames to formulate DASH chunks with Media Presentation Description (MPD) manifest generation and HAProxy-based traffic management for scalable content delivery; and 3)an enhanced Throughput- and Buffer Occupancy-based Adaptation (TBOA+) algorithm that integrates real-time network conditions and buffer state analysis to optimize the quality selection for immersive human-centric volumetric content. Comprehensive experimental evaluations using four volumetric videos (i.e., Longdress, Redandblack, Loot, and Ricardo10) from the 8i Voxelized Full Bodies (8i VFB v2) and MVUB datasets, along with the scaled 5G Irish dataset, reveal that TBOA+ achieves outstanding performance with a 32.73% higher volumetric bitrate, a 21.15% higher average volumetric quality level, and a 31.52% higher throughput efficiency, all while ensuring more stable playback compared to conventional approaches.
Multi-scale feature compression seeks to eliminate task-insensitive redundancy in multi-scale features while maintaining downstream task performance. However, existing methods struggle to dynamically extract inter-scale correlations, leading to insufficient redundancy elimination. Moreover, they typically fail to accurately reconstruct fine-grained spatial details and to adequately preserve key object semantics, limiting their practical coding performance. To address these issues, a Semantic-consistent Holistic Perception Network (SHPNet)-based multi-scale feature compression is proposed, which not only dynamically captures robust inter-scale correlations for compact feature fusion but also significantly enhances the quality of reconstructed features. Specifically, an inter-scale decorrelation-based adaptive fusion module is first designed to reduce task-irrelevant inter-scale redundancy by dynamically mining global and local correlations across scales. Then, a context-guided detail reconstruction module is developed to recover fine-grained spatial details using diverse contextual cues. Finally, we propose a sensitivity-aware signal-semantic fidelity loss that incorporates sensitivity maps to explicitly guide our SHPNet in preserving task-crucial object semantics. Extensive experiments on both instance segmentation and object detection tasks demonstrate that our proposed method achieves superior coding efficiency and significantly outperforms state-of-the-art multi-scale feature compression methods.
Dynamic 3D Gaussian Splatting (4DGS) has emerged as a promising representation for real-time rendering of dynamic scenes, yet its adaptive delivery over bandwidth-constrained networks remains largely under-explored. In this paper, we present a progressive 4DGS streaming system that jointly addresses scalable representation and adaptive transport. On the representation side, we introduce a gradient-aware sensitivity metric to estimate the contribution of each Gaussian primitive to rendering quality. Guided by this metric, Gaussian primitives are hierarchically organized into a compact base layer and successive enhancement layers through iterative layered partitioning. A hybrid refinement strategy is further adopted to recover reconstruction quality after partitioning, while a mixed quantization pipeline compresses the layered representation for low-bitrate transmission. On the transport side, we propose Progressive Adaptive Transmission (PAT), which adapts its scheduling policy to the estimated bandwidth regime. PAT jointly schedules forward streaming and quality backfilling under abundant bandwidth, and falls back to buffer-aware per-segment selection under constrained bandwidth. Extensive evaluations on multiple dynamic datasets and real-world network traces demonstrate the effectiveness of the proposed design. On the N3V dataset, our progressive representation requires only 1.6% of the storage of the non-progressive baseline at the base layer and 11.8% at the highest quality level, while maintaining near-full reconstruction quality. Unlike adaptive bitrate (ABR) based quality switching, which transmits independent higher-quality versions, our incremental refinement reuses all transmitted data and reduces quality-upgrade traffic by up to $3\times $ . Equipped with the PAT scheduler, our system achieves superior QoE compared with both classic ABR and progressive streaming baselines.
With growing demands for higher data transmission rates, spectrum resources are becoming increasingly scarce. Geographically segmented localcasting (GSL) within Single Frequency Networks (SFNs) offers a promising solution by leveraging Layered Division Multiplexing (LDM) to enable Local Content Insertion (LCI), thereby addressing the constraint of conventional SFNs that transmit only identical content from all transmitters. In the GSL framework, a wide-area broadcasting service is transmitted via the LDM core layer to establish a large-scale SFN, while local multimedia and data broadcasting services are delivered to specific areas via the enhanced layer. Due to the lower power allocation of the enhanced layer, the overlap between adjacent transmitter coverage areas is reduced, thereby mitigating co-channel interference (CCI). However, in high-order modulation scenarios, successive interference cancellation (SIC) may still fail. To address this challenge, this paper extends the Iterative Joint Detection and Decoding (IJDD) scheme to higher-order modulation and analyzes its theoretical performance. To manage the computational complexity associated with higher-order constellations, a performance-complexity trade-off is investigated, and a parallel IJDD algorithm is proposed. Simulation results demonstrate that the proposed method significantly outperforms conventional approaches in mitigating CCI under higher-order modulation, validating its potential for localcasting, spectrum reuse, and 6G wireless systems.
In high-throughput satellite (HTS) communication systems, dynamic power allocation is a core part of radio resource management, especially for supporting data-intensive applications that require reliable connectivity across wide geographic areas. However, existing solutions face notable limitations: traditional heuristic algorithms (e.g., PSO, GA) are constrained by high computational complexity and inability to leverage historical experience, while deep reinforcement learning (DRL) struggles with poor generalization to unseen request patterns and insufficient adaptability to dynamic traffic. To address these challenges, we propose an enhanced end-to-end self-learning framework for HTS dynamic power allocation. The framework first adopts a Large Language Model (LLM)-based module to parse user natural language requests, automatically extract and standardize key parameters, and intelligently schedule downstream processes. It then employs a hybrid optimization architecture: a Transformer model captures spatiotemporal correlations of traffic demands to generate high-quality initial power solutions, which are further refined by PSO-GA heuristic algorithms under power constraints to minimize the mismatch rate (MR). Additionally, an online self-learning mechanism based on Low-Rank Adaptation (LoRA) uses high-quality heuristic optimization results to incrementally fine-tune the Transformer, enhancing generalization without excessive computational overhead. Extensive simulations on a GEO HTS platform show that the proposed framework outperforms baseline methods (PSO, GA, DRL) in terms of allocation precision, cross-scenario generalization, and operational efficiency, effectively adapting to the dynamic demands of HTS power allocation.
In professional broadcasting and Over-the-Top (OTT) streaming, non-linear coupling among perceptual modules in hybrid architectures fundamentally limits end-user Quality of Experience (QoE). Current heuristic tuning and brute-force searches cannot properly model cross-module dependencies. Encoders thus often fall into sub-optimal local minima. We propose a systematic relevancy-guided decoupling framework for perceptual video coding to address this issue. We introduce a quantitative Inter-module Relevancy Metric based on a full-factorial design. This metric explicitly models the interaction intensity among heterogeneous coding modules. Guided by this metric, our Influence-Based Ascending Optimization (IBAO) strategy transforms the NP-hard joint optimization into a deterministic sequential tuning path that mitigates cumulative errors. We also design a strictly bounded online adaptive parameter model to address video non-stationarity. It tracks local optimal operating points in real time with minimal runtime overhead. Experiments on the industrial HEVC/x265 platform validate an average 1.94 BD-VMAF gain over the standard x265 Slow preset. Critically, pixel-level objective fidelity remains undegraded. These results confirm our approach effectively resolves inter-module coupling conflicts. Ultimately, it delivers a stable, Pareto-optimal rate-distortion solution suited for real-world multimedia distribution.
This paper proposes low-complexity detection and decoding based on the maximum $a$ posteriori (MAP) criterion for satellite broadcasting systems that use layered division multiplexing (LDM). The first component of this scheme exploits pilot signals to compensate for nonlinear distortion caused by satellite transponders. Specifically, received data signals suffer from nonlinear distortion and are classified into several bit-dependent clusters, of which the centroids can be obtained from the corresponding received pilot signals. The distance between each centroid and a received data signal yields the log likelihood ratio (LLR) of each data bit with sufficient accuracy. Since this component can almost nullify the nonlinear distortion, it is expected to outperform conventional successive interference cancellation (SIC), which has been widely used with the LDM system. To further improve LDM performance, the second component of the proposed scheme performs iterative decoding that reduces the otherwise prohibitive computational complexity in the MAP decoding. To cope with the short-loop issue inherent in the iterative decoding for LDM, the decoding process and associated parameters are optimized on the basis of numerical analyses. Specifically, a new decoding scheme that uses the $a$ posteriori LLR as a criterion is derived from an analysis of the relationship between prior and extrinsic information in iterative decoding. This scheme can determine whether specific extrinsic information should be used for the iterative decoding or not. Numerical analyses also show how inter-layer interference dominance depends on the inter-layer power ratio and how the resulting LLR reliability affects the decoding performance in LDM. Computer simulations over nonlinear channels created by emulating actual satellite transponder characteristics demonstrate that the pilot-aided compensation (the first component) can successfully mitigate the nonlinear distortion, while the optimized iterative decoding (the second component) can further improve the bit error rate (BER).
Recently, learned video compression (LVC) has achieved remarkable progress. Accurate temporal prior modeling is crucial for the rate-distortion (RD) performance of LVC. However, existing methods often suffer from imprecise temporal priors due to limited reconstruction ability of frame decoders and insufficient exploitation of temporal information from reconstructed frames and features. To address these issues, we first design an enhanced feature reconstructor (EFR) by integrating content-aware depthwise separable convolution (CADSC) that excels at local modeling with efficient linear attention duality (ELAD) that facilitates complete global modeling. Based on the EFR, an enhanced frame decoder (EFD) is constructed to improve transform capability, thereby generating high-quality reconstructed frames and features. Furthermore, to fully mine temporal information from these high-quality reconstructions, we propose a multi-type temporal prior mixer (MTPM). Specifically, the MTPM fuses short-term temporal information obtained via optical-flow-based feature warping with long-term temporal information captured by a proposed state-update-based method, thereby further improving the accuracy of temporal prior modeling. The temporal prior produced by the MTPM not only provides high-quality conditional guidance for frame encoding and decoding, but is also combined with the hyper prior, spatial prior, and the discrete latent representation of the previous frame to construct a diverse prior fusion entropy model (DPFEM), which further enhances entropy estimation accuracy. Experimental results demonstrate that the proposed method outperforms existing LVC models and achieves an average performance gain of 29.64% compared with the LDB configuration of the H.266/VVC reference software VTM20.2.
This paper presents a coverage analysis for the deployment of TV 3.0 in Brazil, which adopts the Advanced Television System Committee (ATSC) 3.0 physical layer with mandatory cross-polarized Multiple Input Multiple Output (MIMO) indoor reception. We evaluate the minimum median equivalent field strength (Emed) required to ensure reliable indoor fixed reception while maintaining population coverage comparable to the current Integrated Services Digital Broadcasting Terrestrial (ISDB-Tb) system, originally planned for outdoor reception. In the S & atilde;o Paulo metropolitan region, we investigate two deployment strategies: a High-Power High-Tower (HPHT) configuration and Single Frequency Network (SFN) architectures incorporating Low-Power Low-Tower (LPLT) complementary stations. We perform propagation simulations using the CRC-Predict model with terrain, clutter, building entry loss (Lbel), and receiver antenna gain (GRx) variability. We match, in TV 3.0, the baseline TV 2.x (ISDB-Tb) coverage for S & atilde;o Paulo (51 dB mu V/m with one HPHT and 10 LPLT stations) either with 120 LPLT stations at Emed = 73.75 dB mu V/m or with only 10 LPLT stations at Emed = 62.9 dB mu V/m. Our Monte Carlo sensitivity analysis indicates that carrier-to-noise ratio (C/N) has a medium effect on Emed, while GRx exhibits a small effect, with no statistically significant interaction between parameters. We have demonstrated that our TV 3.0 deployment strategies utilizing SFN configurations with complementary LPLT stations offer solutions for achieving coverage parity with legacy systems through substantial field strength reduction compared to standalone HPHT deployments.
Multibeam satellite systems, benefiting from wide coverage, enable ubiquitous connectivity. Meanwhile, the promising hierarchical rate-splitting multiple access (HRSMA) technique, with robust interference management capabilities, can effectively handle both inter-beam and intra-beam interference caused by multibeam satellite systems. To simultaneously support diverse information services in multibeam satellite systems, this paper proposes a non-orthogonal broadcast, multicast, and unicast (NOBMU) joint transmission scheme based on HRSMA. Specifically, we formulate a maximum sum rate (MSR) problem and develop a block coordinate descent (BCD) iterative optimization (IO) algorithm based on weighted minimum mean square error (WMMSE). The closed-form expressions for beamforming vectors are derived using the Lagrangian expression and Karush-Kuhn-Tucker (KKT) conditions. To enhance the sum rate while ensuring fairness, a two-stage optimization framework is proposed: the first stage solves a max-min fairness (MMF) problem to obtain a guaranteed minimum user rate, which is then applied as a constraint in the second-stage MSR problem. Finally, simulation results demonstrate that the proposed HRSMA-assisted NOBMU transmission scheme significantly outperforms other multiple access schemes.
The next generation of the Brazilian Digital TV System, called TV 3.0 or DTV+, transforms the TV receiver into a comprehensive platform, the Application-Oriented Platform (AoP). That platform provides several new functionalities to viewers, ranging from accessing broadcasters’ content as applications, viewer profile creation, connection with smart devices using a local network, and increased content immersion with the possibility of rendering sensory effects and/or additional content in virtual reality (VR). Given the wide range of functionalities supported by AoP, its important to have it available to the academia, allowing learning the new standard, application testing, experimenting with TV services, and even the proposal and implementation of new functionalities for future evolutions of TV 3.0. This article presents a distributed architecture for the creation of a testbed for the TV 3.0 application coding layer. This architecture is designed to be extensible such that scholars/developers can easily create/extend it including new AoP functionalities. To indicate its usefulness and versatility, this work presents two use cases implemented over this proposed architecture. It also evaluates the proposed architecture regarding delay, and consequently media synchronization.
In 360 degrees video transmission, the "tile-based" mode has proven effective in enhancing viewport quality under bandwidth constraints. However, this mode cannot support real-time generation and large-scale distribution of multi-bitrate and multi-tile video versions within the latency limits required for live streaming. This paper introduces a "tile-free" mode to reduce encoding redundancy and lower computation, caching, and communication overhead. With this mode, we design an edge-assisted LIVE system for processing and transmitting 360 degrees videos, where effective signal superposition of the video layers enables users with poor channel quality to achieve a high quality of experience (QoE). The joint strategy of user grouping, bandwidth allocation, and bitrate decision is formulated as a multi-objective mixed-integer nonlinear programming problem to maximize the users' QoE and minimize the system energy consumption. After deriving the latency equivalence properties among non-dominated solutions, we propose the Non-dominated Sorting Genetic Relaxation (NSGR) scheme, which leverages enhanced genetic algorithms with continuous slackness for improved optimization. Compared with state-of-the-art schemes, NSGR makes 360 degrees video LIVE feasible, enhancing total QoE by 18 similar to 23% and reducing the energy consumption by 50 similar to 62%. Additionally, we propose a constant-time complexity alternative that achieves similar gain.
Point clouds have emerged as a crucial representation for 3D visual content, driving innovations in immersive media applications and broadcasting. The development of no-reference point cloud quality assessment (PCQA) methods is crucial for automating quality monitoring and enhancing user experiences. However, the advancement of PCQA is hindered by the limited scale and diversity of existing databases, coupled with the insufficient exploitation of the intrinsic data information and relationships. To overcome these limitations, we first introduce the PCL-PCQA database, a newly constructed large-scale PCQA database comprising 1,599 subjectively evaluated distorted point cloud samples with diverse geometric and textural distortions, derived from 39 high-quality reference point clouds and labeled with reliable mean opinion scores (MOS) following the ITU-R BT.500 standard. Second, we propose a novel ranking-based no-reference PCQA model, Rank-PCQA, which leverages prior knowledge from inherent quality rankings embedded in distortion levels through a two-stage learning pipeline. This approach involves a pre-training stage with pairwise ranking loss and level-wise boundary regularization to learn hierarchical quality relationships, followed by fine-tuning for accurate MOS regression. Extensive experiments on the SJTU-PCQA, WPC, LS-PCQA, and our PCL-PCQA databases demonstrate that our Rank-PCQA achieves state-of-the-art or highly competitive performance against existing advanced methods, validating its effectiveness and strong generalization capability. The database and source code will be available at https://github.com/KunChun/Rank-PCQA
This paper presents the mathematical analysis and simulation results for a time-domain Combined Integral and Fractional Estimator (CIFE) for the Carrier Frequency Offset (CFO) of Orthogonal Frequency Division Multiplexing (OFDM) communication systems. The proposed method is based on manipulating the number of null subcarriers in a single OFDM preamble symbol, instead of using duplicate symbols as required by most methods found in the literature. The results obtained, including the analysis of the Advanced Television Systems Committee 3.0 (ATSC 3.0) bootstrap, recently adopted by Brazilian TV 3.0, showed that the fractional and integer values of CFO can be accurately estimated in the range of multiple SubCarrier Spacing (SCS), even in low Signal to Noise Ratio (SNR) scenarios. The proposed method can be considered a generalization of the Maximum Likelihood Estimation (MLE) and Schmidl and Cox Estimation (SCE) methods. Furthermore, the proposed symbol structure reduces the preamble overhead while remaining adequate for time synchronization and frame detection.
Currently, there is an increasingly high demand for high-quality images across various streaming media applications, but visual signals are impaired by composite noise. The existing dataset-based image denoisers, requiring a huge number of clean-noisy image pairs from diverse scenes to well train a large-size deep network, often perform unwell in the real-world noisy image restoration due to the complexity and variability of scenes, limited computing resources and the difficulty in obtaining clean images. So this paper tries to propose a dataset-free highly efficient zero-shot image denoiser with a lightweight CNN, called HEZID, to achieve the effective real-time real-world noisy image restoration. First, for the problem of the BSN strategy that discards the center pixel to avoid identity mapping for image denoising, we propose a new solution by devising a dual-mode pixel-realignment strategy, which sequentially develops a ‘rough’ pixel-realignment method for smooth regions and a ‘fine’ pixel-realignment method for edge regions, therefore fully exploiting the center pixel and preventing identity mapping. Second, we develop a super-lightweight CNN-based denoiser based on the strip convolution, which is motivated by the high sensitivity of human eyes to horizontal and vertical stimuli for visual inputs, followed by introducing the residual learning and separable convolutions to severally facilitate feature propagation and reduce the computational burden. Experiments on multiple real-world noisy image datasets confirmed that our HEZID outperformed a dozen state-of-the-art peers in terms of restoration quality and running efficiency.
HTTP-based adaptive video streaming has become a key contributor to the Internet traffic. The streaming quality however is heavily affected by the throughput and stability of the underlying TCP. Being compatible and fair, multipath TCP (MPTCP) has recently been suggested as a promising enhancement to the conventional TCP, particularly considering that multi-home is becoming ubiquitous for modern networked devices. The opportunities and challenges of adaptive streaming over MPTCP however remain largely unclear to the research community. This paper presents a systematic study on adaptive streaming over MPTCP. We start from realworld experiments with Dynamic Adaptive Streaming over HTTP (DASH) and analysis on its performance over MPTCP. We show that DASH can greatly benefit from the improved aggregated throughput by MPTCP; yet the inter-path throughput difference and the intra-path throughput fluctuation have noticeable (negative) impact, too. Without a proper design of path selection and adaptation in MPTCP, they can easily confuse the adaptation logic of DASH, resulting in low bitrates or frequent rebuffering even if high-bandwidth paths are available. We present MPTCP+, an extended multipath TCP solution to offer high quality and smooth playback for adaptive HTTP streaming. MPTCP+ incorporates a path use decision algorithm that smartly disables/enables a path to minimize the inter-path difference, and a novel congestion control algorithm that smooths congestion window evolution with multiple paths. We have implemented MPTCP+ in the MPTCP Linux kernel, with minimum change on the server-sideMPTCPmodule only. It is fully compatible with the existing MPTCP clients and requires no change on the upper-layer protocols, too. Our experiments suggest that MPTCP+ increases the quality of experience (QoE) of DASH by up to 50%.
The proliferation of immersive services, such as augmented and virtual reality, poses unprecedented challenges for wireless networks due to their stringent requirements for high data rates, low latency, and high reliability. This paper investigates the resource optimization problem in an integrated network architecture that combines a high-power high-tower (HPHT) broadcast network and low-power low-tower (LPLT) cellular networks for delivering multi-perspective immersive streams. We formulate a novel multi-objective optimization problem that simultaneously caters to the utilities of three key stakeholders: content providers (minimizing server congestion), network operators (balancing spectral and energy efficiency), and end users (maximizing quality of experience). To centrally coordinate resource allocation across both networks, we introduce an agent, Convergence Intelligent Processing Entity (CIPE), that collects information from all parties and makes globally optimized decisions. To tackle this complex, non-convex, and large-scale problem, we propose an efficient block coordinate descent (BCD)-based algorithm that decomposes the problem into tractable subproblems: HPHT parameter configuration, MBSFN zone formation and user association, and LPLT resource allocation. Extensive simulation results demonstrate that the proposed algorithm is significantly superior to several well-known baselines, including a single-cell point-to-point (SC-PTP) only scheme, an HPHT-only broadcast scheme, and various hybrid schemes that partially leverage heterogeneous resources, such as SC-PTP with broadcast, and SC-PTP with multi-point multicast. Furthermore, we established an external field experimental environment to verify the feasibility and performance gains of the integrated network architecture in a real-world setting.
eXtended Reality (XR) Applications are characterized by stringent Quality of Service (QoS) demands, including low latency, high reliability, and sustained high data rates. These requirements are particularly challenging to satisfy in dynamic 5G network environments, where conventional Over-The-Top (OTT) rate control strategies—such as adaptive bitrate streaming and packet pacing—operate without visibility into network conditions. While recent efforts like CAMARA APIs aim to expose QoS capabilities, they primarily offer coarse-grained, reactive control mechanisms that do not address user-specific service degradation. This work introduces NetXRate, an O-RAN-compliant Network as a Service (NaaS) framework designed to enable user-centric rate adaptation for XR traffic. NetXRate integrates 3GPP Network Data Analytics Function (NWDAF) insights with a fairness-aware xApp deployed within the O-RAN Radio Intelligent Controller (RIC), enabling rate recommendations from the network to the XR application. Extensive simulation-based evaluations demonstrate that, compared to OTT adaptation methods, NetXRate achieves up to 96% outage reduction when XR sessions are not prioritized and up to 85% outage reduction when XR traffic is prioritized. Our results also demonstrate that by making more efficient use of spectrum, introducing NetXRate also enhances the performance of eMBB services in cells where both XR users and eMBB users coexist.
