5G Radio Access Network (RAN) slicing provides support for resource isolation and dynamic resource optimization to accommodate the diverse Quality of Service (QoS) needs across Internet of Things (IoT) ecosystems, i.e., enhanced Mobile Broadband (eMBB), Ultra-Reliable Low-Latency Communications (URLLC), and massive Machine Type Communication (mMTC) or Best Effort (BE) services. However, existing optimization- and data-driven radio resource allocation (RRA) models often suffer from rigid/suboptimal resource distribution and increased Service Level Agreement (SLA) violations in dynamic, multi-service environments. This paper introduces an Adaptive QoS-aware Priority Scheduling (AQPS) solution, a novel RRA approach designed to optimize resource distribution for multi-service 5G network slicing deployments in IoT-enabled infrastructures. To optimally allocate resource block groups (RBGs) according to time-varying traffic patterns and IoT network conditions, we formulate an NP-hard optimization problem with the objective of minimizing SLA violations while satisfying QoS requirements for diverse IoT service classes. In this context, AQPS optimizes resource distribution for IoT services by incorporating: (i) minimum guarantee allocation (QoS-driven RBG estimation), (ii) weighted urgency-based resource distribution (computing user urgency by incorporating buffer state, QoS factors, spectral efficiency, and service priorities), and (iii) priority-based round-robin allocation. Extensive trace-driven simulation findings from two comprehensive multiuser IoT-dense urban and enhanced IoT coverage zone scenarios with realistic channel conditions demonstrate that the AQPS solution experiences a very low number of SLA violations while enabling SLA assurance rates of 99.07% and 97.58%, respectively, when compared against state-of-the-art benchmarks, including Deep Reinforcement Learning (DRL), Round Robin (RR), and Stepwise Optimal Algorithm (SOA).

Optimal tile-based bitrate allocation improves the Quality of Experience (QoE) for adaptive 360° video streaming across multiple clients in heterogeneous network environments; however, it is challenging as it implies accurate viewport prediction, finest tile-based bitrate reservation, and maintaining QoE fairness, particularly under constrained network conditions. This paper proposes a strategy named EDGE360, that employs an edge-driven Multi-Agent Deep Reinforcement Learning (MADRL) solution for rate adaptation to improve the joint QoE in DASH-based rich media content delivery based on adaptive viewport prediction and Video Multi-method Assessment Fusion (VMAF) corresponding tiling granularity selection. Cooperative strategies among agents in the central critic network are crucial for addressing the complexity of network instances at the edge and optimizing media streaming bitrate assignment in multiple-client scenarios. Therefore, EDGE360 aims to implement the Counterfactual Multi-Agent Policy Gradients (COMA) based on 5G network traces to train agents in policies that optimize individual client QoE and fairness among clients, resulting in an improved rich streaming experience. At the edge, a tile-based quality monitor evaluates viewport trajectories, buffer status, and network throughput, employing deep learning to forecast optimal tile bitrate allocation, which is formulated as an MDP and solved with MADRL. Based on extensive experimentation, EDGE360 surpasses state-of-the-art adaptive bitrate algorithms by achieving the highest average reward, outperforming RAPT360, 360SRL, and BOLA360 by 8.12%, 11.86%, and 18.00%, respectively, demonstrating superior convergence and refinement.

Multiple approaches aim to enhance user experience in the delivery of immersive video content. The popularisation of VR, combined with recent advances in mulsemedia technology has improved access to immersive visual and olfactory stimuli. Synchronising multiple scent dispensers positioned around the user when watching 360° videos can more accurately indicate the location of scent sources, guiding users to move their heads accordingly to the indicated directions. However, the manual annotation process required to add mulsemedia effects is labour-intensive, limiting the availability of content with sensory enhancements, particularly when using multiple scent dispensers from various directions. Addressing this issue, this paper introduces OmniScent-CNN, an innovative solution to automate the diffusion of scents from different directions in a VR environment using Convolutional Neural Networks (CNNs) for scene recognition. Multiple instances of the solution were tested, employing a number of CNN architectures. The results demonstrated that olfaction accuracy can reach up to 71.28% with the ResNet-18 model. Furthermore, user perceptual tests revealed excellent results, with 87.5% of participants agreeing or strongly agreeing that the scents enhanced their enjoyment of the experience. This indicates the feasibility of automating the process of synchronising omnidirectional scents based on 360° scene recognition.

Wildfires rank among the most prevalent natural disasters globally and have emerged as a significant factor in climate change over the past decade. Early detection of wildfires and smoke plumes through satellite imagery is crucial since they are not easily extinguishable which may lead to catastrophic consequences for both wildlife and forest ecosystems. Classic deep-learning models for wildfire and aerosol identification have shown significant progress, but high false-positive rates remain a key limitation. This paper proposes a custom-designed Convolutional Neural Network (CNN) model that aims to improve the identification of wildfires and aerosols, leveraging satellite imagery categorized into cloud, dust, haze, land, seaside, and smoke. Moreover, we considered popular deep learning and transfer learning models, specifically EfficientNet, MobileNetV3, and Inception V3, to identify and distinguish smoke plumes. Hyper-parameter tuning has been performed to achieve more accurate results. The metrics evaluated in this research are accuracy, precision, recall, and f1-score. A comprehensive analysis was performed that aimed to identify the best transfer learning model and the model that closely aligns with the performance of the CNN built. This paper provides valuable insights into the potential use of transfer learning models and of the proposed custom-designed CNN model in early wildfire detection by identifying the smoke plumes and helping in reducing false fire alarms.

This paper introduces the Utility-Based Multipath Transmission Control Protocol (uMPTCP), an innovative extension of MPTCP designed for prioritized extended reality (XR) content delivery, based on monitored Quality of Service (QoS) network metrics. The proposed approach includes an algorithm that assesses subflows’ delivery performance and dynamically selects the most efficient one to deliver prioritized content with reduced latency. A comparison is made against a stateof-the-art solution, the virtual private channel (VPC), and the default MPTCP algorithm. The evaluation considers both singlehomed and multi-homed configurations in scenarios with varied bandwidth requirements, including XR. The study is conducted within the framework of the FRADIS project, aimed at providing a comprehensive solution for 5G and beyond heterogeneous network environments. FRADIS integrates machine learning to optimize service-specific approaches, allowing a choice between traffic engineering with network slicing and protocol-based solutions, including the proposed uMPTCP solution. The framework supports a diverse range of services for smart city monitoring, XR applications, e-health solutions and entertainment.

The emergence of 5G networks signifies a shift towards a range of unique and high-demand network services, which prioritize conflicting Quality of Service (QoS) requirements for diverse applications such as video streaming, Virtual Reality (VR), and gaming services. Network slicing in 5G architecture is crucial in addressing these diverse needs using a common physical infrastructure. The integration of 5G New Radio (NR) introduces promising features such as Bandwidth Parts (BWPs), facilitating flexible resource configuration for each User Equipment (UE) within dedicated bandwidth subsets. However, as network slicing continues to mature, the effective implementation and management of Radio Access Network (RAN) slices for these multifaceted services remains a considerable hindrance. This paper introduces an innovative 5G Slice Streaming solution (5GSS) specifically tailored to elevate content streaming performance and address the unique requirements of various slices of different content types e.g. adaptive video (DASH), VR, and gaming. With a primary focus on video streaming and considering it as one of the pivotal slices, the proposed 5GSS solution simultaneously considers multiple parameters such as buffer impact factor, download impact factor, throughput proximity, and number of quality switches, and applies a Pareto optimization to identify and perform the optimal bitrate selection in a 5G RAN slicing environment. The comprehensive NS3 experimental setup accommodates 5G NR and DASH modules. The experimental results validate the superior performance of the proposed solution, demonstrating noticeable improvements in visual quality, quality smoothness, throughput efficiency and fairness, compared to existing solutions.

Next-generation cellular networks strive to offer ubiquitous connectivity, enhanced transmission rates with increased capacity, and superior network coverage. However, they face significant challenges due to the growing demand for multimedia services across diverse devices. Adaptive multimedia streaming services are essential for achieving good viewer Quality of Experience (QoE) levels amidst these challenges. Yet, the existing adaptive video streaming solutions do not consider diverse QoE preferences or are limited to meeting specific QoE objectives. This paper presents FReD-ViQ, a Fuzzy Reinforcement Learning-Driven Adaptive Streaming Solution for Improved Video QoE that combines the strengths of fuzzy logic and advanced Deep Reinforcement Learning (DRL) mechanisms to deliver exceptional, individually tailored user experiences. FReD-ViQ is a sophisticated streaming solution that leverages efficient membership function modelling to achieve a more finely-grained representation of both input and output spaces. This advanced representation is augmented by a set of fuzzy rules that govern the decision-making process. In addition to its fuzzy logic capabilities, FReD-ViQ incorporates a novel DRL algorithm based on Dueling Double Deep Q-Network (Dueling DDQN), noisy networks, and prioritized experience replay (PER) techniques. This innovative fusion enables effective modelling of uncertain network dynamics and high-dimensional state spaces while optimizing exploration-exploitation trade-offs in adaptive streaming environments. Extensive performance evaluations in real-world simulation settings demonstrate that FReD-ViQ effectively surpasses existing solutions across multiple QoE models, yielding average improvements of 23.10% (Linear QoE), 23.97% (Log QoE), and 33.42% (HD QoE).

The widespread availability of smart computing and display devices such as mobile phones, gaming consoles, laptops, and tethered/untethered head-mounted displays has fueled an increase in demand for omnidirectional (360°) videos. 360° video applications enable users to change their viewing angles while interacting with the video during playback. This allows users to have a more personalized and interactive viewing experience. Unfortunately, these applications require substantial network and computational resources that the conventional infrastructure and end devices cannot support. Recently proposed viewport adaptive fixed tiling solutions stream only relevant video tiles based on user interaction with the virtual reality (VR) space to use existing transmission resources more efficiently. However, achieving real-time accurate viewport extraction and transmission in response to both head movements and bandwidth dynamics can be challenging, which can impact the user’s Quality of Experience (QoE). This article proposes innovative dynamic tiling-based adaptive 360° video streaming solutions in order to achieve high viewer QoE. First, novel and easy-to-scale tiling layout selection methods are introduced, and the best tiling layouts are employed in each adaptation interval based on the prediction-assisted visual quality metric and the observed viewport divergence. Second, a novel proactive tile selection approach is presented, which adaptively extracts tiles for each selected tiling layout based on two low-complex viewport prediction mechanisms. Finally, a practical dynamic tile priority-oriented bitrate adaptation scheme is introduced, which uniformly distributes the bitrate budget among different tiles during 360° video streaming. Extensive trace-driven experiments are conducted to evaluate the proposed solutions using head motion traces from 48 VR users for five 360° videos with tiling layouts of 4 × 3, 6 × 4, and 8 × 6 and segment durations of 1s, 1.5s, and 2s. The experimental evaluations show that the dynamic video tiling solutions achieve up to 11.2% more viewport matches and an average improvement in QoE of 9.7% to 18% compared to state-of-the-art 360° streaming approaches.

Multimedia streaming for the Internet of Vehicles has the potential to enhance road safety and transport efficiency for autonomous vehicles, and the in-car experience for passengers. MPEG-Dynamic Adaptive Streaming over HTTP (MPEG-DASH) framework has been widely deployed to optimize video streaming with respect to end-user Quality of Experience (QoE). However, existing heuristic-based, reinforcement learning-based, or fuzzy-based adaptive algorithms, which use complex control laws and decision-making processes, are not well-suited to handle the non-stationary nature of road traffic environments. Consequently, these solutions often struggle to deliver optimal performance across multiple QoE objectives and under diverse network conditions. In this paper, we introduce FLAME, a novel adaptive multimedia streaming solution based on advanced fuzzy logic. FLAME incorporates interactive membership functions and fuzzy rules in its two variants, FLAME7 and FLAME5, resulting in reduced model complexities and training overheads. FLAME is adaptable to diverse video client settings and QoE goals. Our trace-driven experimental results demonstrate that FLAME solutions offer an improved uninterrupted streaming experience for connected vehicles. On average, FLAME outperforms other state-of-the-art solutions such as PENSIEVE, BOLA, FESTIVE, BBA, and ELASTIC by achieving 11.7% higher QoE.

The latest increase in multi-view video solutions, including those for telepresence, commercial conferencing, remote collaboration, etc. requires support for the high-quality delivery of large amounts of content data. Meanwhile, the extensive proliferation of wireless network access technology and multiple network interfaces on modern devices prompt the network transmission performance over various access networks. Diverse multipath-based multi-view streaming and adaptive delivery solutions were proposed, but they do not enable differentiation between streams. This paper proposes MVP-DASH, a priority-aware adaptive multi-view video streaming scheme based on the MPEG-DASH framework. MVP-DASH enables improved visual quality for high-priority streams, while maintaining acceptable quality levels for low-priority streams, primarily when delivered over a dynamic network environment. The experimental evaluation of the MVP-DASH demonstrates the effectiveness of the proposal in terms of achieving higher video quality and fewer video quality switches in comparison with alternative approaches.

Today, the dynamic network environment poses severe challenging issues to multimedia streaming services that account for an enormous part of network traffic all over the world. Dynamic adaptive streaming over HTTP (DASH) facilitates seamless video playback by allowing for dynamic adjustment of the video bitrate to the ongoing network situation. Despite several attempts, there is still a challenge to design solutions which use DASH to adjust video delivery to the dynamic network environment and achieve high user quality of experience levels. This paper presents a novel DASH-based throughput and buffer occupancy-based adaptation (TBOA) algorithm to provide an improved streaming experience for remote users. TBOA was compared against alternative solutions such as FDASH and SFTM in single- and multiple-client scenarios. Testing results show how TBOA selects higher video bitrates while performing fewer video bitrate switches and reduces the risk of buffer underrun in comparison with the competitors.