Quality Of Service Requirements Research Articles

Multi-agent DRL-based Multipath Scheduling for Video Streaming with QUIC

The popularization of video streaming brings challenges in satisfying diverse Quality of Service (QoS) requirements. The multipath extension of the Quick UDP Internet Connection (QUIC) protocol, also called MPQUIC, has the potential to improve video streaming performance with multiple simultaneously transmitting paths. The multipath scheduler of MPQUIC determines how to distribute the packets onto different paths. However, while applying current multipath schedulers into MPQUIC, our experimental results show that they fail to adapt to various receive buffer sizes of different devices and comprehensive QoS requirements of video streaming. These problems are especially severe under heterogeneous and dynamic network environments. To tackle these problems, we propose MARS, a Multi-agent deep Reinforcement learning (MADRL)-based Multipath QUIC Scheduler, which is able to promptly adapt to dynamic network environments. It exploits the MADRL method to learn a neural network for each path and generate scheduling policy. Besides, it introduces a novel multi-objective reward function that takes out-of-order queue size and different QoS metrics into consideration to realize adaptive scheduling optimization. We implement MARS in an MPQUIC prototype and deploy in Dynamic Adaptive Streaming over HTTP system. Then, we compare it with the state-of-the-art multipath schedulers in both emulated and real-world networks. Experimental results show that MARS outperforms the other schedulers with better adaptive capability regarding the receive buffer sizes and QoS.

Secure transmission design in cooperative NOMA networks: Orienting toward user requirements

AbstractThis paper presents a full‐duplex and multi‐antenna relay‐aided transmission scheme for the user‐oriented perspective in cooperative non‐orthogonal multiple access (NOMA) systems. Relying on the service requirements, two legitimate users with security demands (SU) and quality of service (QoS) demands (QU) are paired for NOMA execution. The security of SU is enhanced through inverse power allocation of the conventional NOMA. The QoS requirements for QU are guaranteed, simultaneously. Then, the precoding design at the multi‐antenna relay has been exploited to improve the reliability demands of SU. Closed‐form expressions for SU's outage probability (OP) and intercepted probability, as well as QU's OP, are derived. The security–reliability trade‐off is also presented. Additionally, performance analysis under the half‐duplex and multi‐antenna scheme is provided as a benchmark for comparison. Finally, numerical simulations are performed to confirm the accuracy of the theoretical derivations under different schemes. Results demonstrate the superiority of the proposed scheme over benchmark schemes in terms of improving the security and reliability of SU, as well as saving power consumption of the system.

Dynamic hierarchical intrusion detection task offloading in IoT edge networks

AbstractThe Internet of Things (IoT) has gained widespread importance in recent time. However, the related issues of security and privacy persist in such IoT networks. Owing to device limitations in terms of computational power and storage, standard protection approaches cannot be deployed. In this article, we propose a lightweight distributed intrusion detection system (IDS) framework, called FCAFE‐BNET (Fog based Context Aware Feature Extraction using BranchyNET). The proposed FCAFE‐BNET approach considers versatile network conditions, such as varying bandwidths and data loads, while allocating inference tasks to cloud/edge resources. FCAFE‐BNET is able to adjust to dynamic network conditions. This can be advantageous for applications with particular quality of service requirements, such as video streaming or real‐time communication, ensuring a steady and reliable performance. Early exit deep neural networks (DNNs) have been employed for faster inference generation at the edge. Often, the weights that the model learns in the initial layer may be sufficiently qualified to perform the required classification tasks. Instead of using subsequent layers of DNNs for generating the inference, we have employed the early‐exit mechanism in the DNNs. Such DNNs help to predict a wide range of testing samples through these early‐exit branches, upon crossing a threshold. This method maintains the confidence values corresponding to the inference. Employing this approach, we achieved a faster inference, with significantly high accuracy. Comparative studies exploit manual feature extraction techniques, that can potentially overlook certain valuable patterns, thus degrading classification performance. The proposed framework converts textual/tabular data into 2‐D images, allowing the DNN model to autonomously learns its own features. This conversion scheme facilitated the identification of various intrusion types, ranging from 5 to 14 different categories. FCAFE‐BNET works for both network‐based and host‐based IDS: NIDS and HIDS. Our experiments demonstrate that, in comparison with recent approaches, FCAFE‐BNET achieves a 39.12%–50.23% reduction in the total inference time on benchmark real‐world datasets, such as: NSL‐KDD, UNSW‐NB 15, ToN_IoT, and ADFA_LD.

AI-driven resource management strategies for cloud computing systems, services, and applications

Cloud computing is a way for businesses and individuals to It has changed and revolutionized the way we access and use resources. However, efficient resource management in cloud computing systems remains a major challenge due to the scalability, heterogeneity, and dynamic nature of these environments. To address these challenges, artificial intelligence (AI) technology has emerged as an effective solution to improve resource management efficiency. This paper provides an overview of AI-based strategies for efficient resource management in cloud computing systems, services, and applications. This paper first reviews resource management challenges in cloud computing, including scalability, heterogeneity, quality of service requirements, and cost optimization. Below is an overview of the various AI techniques used for resource management. B. Algorithms for machine learning, reinforcement learning, predictive analytics, natural language processing, and genetic algorithms. Next, this paper considers specific AI-based strategies for efficient resource management. These strategies include automated resource provisioning and scaling, intelligent workload planning and task allocation, predictive maintenance and fault detection, and energy-efficient resource management. We also present case studies and applications of AI-driven resource management in various cloud computing scenarios, including large-scale cloud providers, edge computing, serverless computing, and container environments. This paper describes evaluation metrics and performance analysis techniques to evaluate the effectiveness of AI-based resource management approaches. It highlights the importance of ethical considerations, transparency, and explainability in AI-powered resource management systems. Additionally, the integration of AI technologies into existing resource management frameworks is discussed, and future directions are identified, including B. real-time resource optimization and coordination.

Analysis of Wireless Multimedia Big Data Transmission Based on Adaptive Scheduling Algorithm

The increasing demand for wireless multimedia applications has prompted the need for efficient data transmission strategies that can adapt to dynamic network conditions and diverse Quality of Service (QoS) requirements. This paper analyzes wireless multimedia big data transmission using an innovative Adaptive QoS-Aware Scheduling Algorithm (AQOS-ASA). The proposed algorithm aims to optimize the transmission of multimedia data by dynamically adjusting the scheduling based on QoS considerations and real-time network conditions. Through adaptive scheduling, the algorithm ensures that high-priority multimedia data receives optimal transmission treatment, balancing the quality preferences set by users. The algorithm dynamically allocates time slots based on the changing network conditions, allowing for efficient utilization of the available bandwidth. The paper analyses the algorithm's performance in different wireless network scenarios through simulations and evaluations, highlighting its effectiveness in adapting to varying QoS requirements and network conditions. The results demonstrate that the proposed algorithm significantly enhances the quality and efficiency of wireless multimedia big data transmission, making it well-suited for applications with diverse multimedia content and dynamic network environments. The findings contribute valuable insights to wireless multimedia communication and provide a foundation for further advancements in adaptive scheduling algorithms for improved Quality of Service.

MUSCAT: Distributed multi-agent Q-learning-based minimum span channel allocation technique for UAV-enabled wireless networks

We consider a minimum span channel allocation problem (MS-CAP) to overcome spectrum scarcity and facilitate the efficiency of unmanned aerial vehicle (UAV)-enabled wireless networks. Basically, the MS-CAP minimizes the difference between the maximum and minimum used frequency, i.e., the required total bandwidth, while guaranteeing the quality-of-service (QoS) requirements for each wireless link in the network. The conventional optimal minimum span channel allocation (MS-CA) scheme is based on a centralized approach, assuming that global network information is available at the central controller. In practice, however, this may not be feasible for dynamic environments like UAV-enabled wireless networks since the real-time exchange of network information and channel allocation results with dynamically moving UAVs is formidable. Hence, we propose a novel practical MS-CA algorithm based on distributed multi-agent reinforcement learning (MARL), where each agent independently learns its best strategy with its local observations. To the best of our knowledge, the proposed technique is the first work of designing a distributed MARL for the MS-CAP for multi-UAV-enabled wireless networks in the literature. Numerical results reveal that the proposed distributed MS-CA technique can efficiently save the required total bandwidth while ensuring the QoS requirements of each link, represented by the signal-to-interference plus noise ratio (SINR) threshold, even in dynamic wireless networks. It validates the applicability of the proposed distributed MS-CA framework to dynamic networks.

Autonomous integration of TSN-unaware applications with QoS requirements in TSN networks

Modern industrial networks transport both best-effort and real-time traffic. Time-Sensitive Networking (TSN) was introduced by the IEEE TSN Task Group as an enhancement to Ethernet to provide high quality of service (QoS) for real-time traffic. In a TSN network, applications signal their QoS requirements to the network before transmitting data. The network then allocates resources to meet these requirements. However, TSN-unaware applications can neither perform this registration process nor profit from TSN’s QoS benefits.The contributions of this paper are twofold. First, we introduce a novel network architecture in which an additional device acts as a central user configuration (CUC) for TSN-unaware applications and autonomously signals their QoS requirements to the network. Second, we propose a processing method to detect real-time streams in a network and extract the necessary information for the TSN stream signaling. It leverages a Deep Recurrent Neural Network (DRNN) to detect periodic traffic, extracts an accurate traffic description, and uses traffic classification to determine the source application. As a result, our proposal allows TSN-unaware applications to benefit from TSNs QoS guarantees. Our evaluations underline the effectiveness of the proposed architecture and processing method.

UAV-Assisted IoT Applications, QoS Requirements and Challenges with Future Research Directions

ABSTRACT: Unmanned Aerial Vehicle (UAV)-assisted Internet of Things application communication is an emerging concept that effectuate the foreknowledge of innovative technologies. With the accelerated advancements in IoT applications, the importance of this technology became more impactful and persistent. Moreover, this technology have demonstrated useful contributions across various domains, ranging from general to specific applications. Examples include wildfire monitoring, coastal area monitoring, deforestation monitoring, and sensitive military operations, where human access is limited or not feasible. These examples underscore the technology’s importance in scenarios where direct human involvement is challenging or impossible. Although this technology offers numerous benefits, it is essential to note that it also faces several challenges. Among these, Quality of Service (QoS) is a key concern, which limits its useability in various applications. Unfortunately, most researchers in the present literature have overlooked this important factor without giving it considerable attention. To fill this gap, we are presenting a systematic review of the present literature associated with the QoS metrics of this emerging technology from 2015 to 2023 to highlight their contributions and limitations. Based on the systematic review, we highlight the open challenges of this technology to set a roadmap for futuristic research. Finally, we compared each portion of this work with the previously published review articles to confirm the essence of this work, along with an explanation of why this survey is needed and in-time.

Efficient symbol level precoding design for full-duplex integrated sensing and communication systems

Full-duplex (FD) integrated sensing and communication (ISAC) system attracts much attention because adopting the FD mode can enhance the transmission efficiency and reduce the transmission time of the ISAC system. However, the inevitable self-interference (SI) in FD mode would result in severe performance degradation. This paper considers the FD ISAC system and investigates a symbol-level precoding (SLP) design, which minimizes the Cramér-Rao Bound of target angle estimation while considering the constant-envelope power restraint, SI suppression and the quality of service requirement. Due to the non-convex fractional problem and the unit-modulus constraint, the optimization problem is reformulated by implementing the Schur complement, where the optimal waveform is obtained via the semidefinite programming algorithm. The SLP design that optimizes each symbol vector for all users may suffer high computational complexity of matrix multiplication with high numbers of users. A grouped SLP strategy in which the communication users are divided into several groups on the basis of channel correlation is proposed to reduce the complexity. The proposed grouped SLP design only suppresses group-to-group interference and converts the intra-group interference into a constructive interference, which can reduce the dimension of each precoded signal significantly, resulting in a low computational complexity. Simulation results illustrate that the proposed SLP design can effectively mitigate SI, which can improve the accuracy of target detection. Moreover, the proposed grouped-SLP design can further reduce computational complexity without dramatic performance degradation.

Simulation and analysis of C-V2X data transmission in different traffic scenarios based on MAC layer scheduling algorithms

With the rapid development of communication technology, vehicles are becoming more and more intelligent, and Vehicle to Vehicle (V2V) communication has become an important cornerstone for the future intelligence of vehicles. Because of the high-reliability and low-latency characteristics of 5G networks, many applications can be realized by using Vehicle to Network (V2N) communications. Different applications have different requirements for Quality of Service (QoS), which should be met by the network, and the base station can use different scheduling methods to satisfy these different requirements. This paper focuses on the scheduler at the MAC layer in 5G base stations. The scheduler must handle both video signals and other communication signals transmitted by the vehicle, and must also consider Hybrid Automatic Repeat reQuest (HARQ) retransmissions. We performed a C-V2X signaling network transmission simulation by using OMNeT++. We considered urban and motorway, two public transportation scenarios. Different types of data are set to be transmitted in two scenarios. We used different scheduling schemes for data scheduling, then investigated the performance of these scheduling schemes for transmitting data. The methods and simulation experiments mentioned in this paper are important for practical generalization.

QoS-aware edge AI placement and scheduling with multiple implementations in FaaS-based edge computing

Resource constraints on the computing continuum require that we make smart decisions for serving AI-based services at the network edge. AI-based services typically have multiple implementations (e.g., image classification implementations include SqueezeNet, DenseNet, and others) with varying trade-offs (e.g., latency and accuracy). The question then is how should AI-based services be placed across Function-as-a-Service (FaaS) based edge computing systems in order to maximize total Quality-of-Service (QoS). To address this question, we propose a problem that jointly aims to solve (i) edge AI service placement and (ii) request scheduling. These are done across two time-scales (one for placement and one for scheduling). We first cast the problem as an integer linear program. We then decompose the problem into separate placement and scheduling subproblems and prove that both are NP-hard. We then propose a novel placement algorithm that places services while considering device-to-device communication across edge clouds to offload requests to one another. Our results show that the proposed placement algorithm is able to outperform a state-of-the-art placement algorithm for AI-based services, and other baseline heuristics, with regard to maximizing total QoS. Additionally, we present a federated learning-based framework, FLIES, to predict the future incoming service requests and their QoS requirements. Our results also show that our FLIES algorithm is able to outperform a standard decentralized learning baseline for predicting incoming requests and show comparable predictive performance when compared to centralized training.

Energy-aware cloud manufacturing service selection and scheduling optimization

ABSTRACT Cloud Manufacturing Service Selection and Scheduling (CMSSS) is vital for optimizing resource allocation and meeting task requirements. However, inattention to the preheating process of manufacturing equipment has resulted in wasted energy. To reduce manufacturing energy consumption and ensure the Quality of Service (QoS), this paper establishes a bi-level programming model for CMSSS, quantifies the preheating energy consumption of manufacturing equipment by task cohesion, and proposes an energy-aware cloud manufacturing service selection and scheduling approach. The approach selects the service composition for a task from candidate service sets, schedules subtasks to avoid service occupancy conflicts and maximises task cohesion to reduce the preheating energy consumption of manufacturing equipment by Energy-aware Scheduling Generation Scheme (ESGS). Finally, it integrates ESGS into Non-dominated Sorting Genetic Algorithm II (NSGA-II) to determine the optimal task execution solution. Experimental results show that ESGS has a better Pareto front than the previous Feasible Scheduling Generation Scheme (FSGS) under seven types of QoS weights. With almost the same QoS satisfaction level, ESGS consumes, on average, 2% to 8% less energy than FSGS for preheating manufacturing equipment. In cloud manufacturing scenarios with preheating processes, ESGS can meet the QoS requirements of demanders as FSGS but with a better energy economy.

A highly efficient resource slicing and scheduling optimization algorithm for power heterogeneous communication networks based on hypergraph and congruence entropy

New services, such as distributed photovoltaic regulation and control, pose new service requirements for communication networks in the new power system. These requirements include low latency, high reliability, and large bandwidth. Consequently, power heterogeneous communication networks face the challenge of maintaining quality of service (QoS) while enhancing network resource utilization. Therefore, this paper puts forward a highly efficient optimization algorithm for resource slicing and scheduling in power heterogeneous communication networks. Our first step involves establishing an architectural description model of heterogeneous wireless networks for electric power based on hypergraph. This model characterizes complex dynamic relationships among service requirements, virtual networks, and physical networks. The system congruence entropy characterizes the degree of matching between the service demand and resource supply. Then an optimization problem is formed to maximize the system congruence entropy through dynamic resource allocation. To solve this problem, a joint resource allocation and routing method based on Lagrangian dual decomposition is proposed. These methods provide the optimal solutions of the nodes and link mappings of service function chains. The simulation results demonstrate that the proposed algorithm in this paper can greatly enhance resource utilization and also meet the QoS requirements of various services.

Energy efficiency in 5G systems: A systematic literature review

To ensure Energy Efficiency (EE) and better Quality of service (QoS), it is necessary to analyze the energy saving possibilities for low resource utilization in the current networks caused by rigorous QoS requirements and implementing EE approach in the planned model for performance improvement. Distributed Denial of Service (DDoS) attacks aim to exhaust the network’s processing and communication capacity by saturating it with packets and generating malicious traffic. There are numerous advantages that make Digital Twin (DT) and Intrusion Detection technique (ID) an effective remedy for a range of (fifth generation) 5G problems. A DDoS attack must be immediately detected and stopped before a legitimate user can access the target of the attacker for the 5G network to provide an efficient energy service. Although they clearly show promise in assisting with the creation and implementation of the challenging 5G environment, Digital Twins is still a relatively new technology for 5G networks but will increase EE. In this research, a thorough examination of the materials was carried out to identify the most cutting-edge DT and ID methods. The purpose of this study was to comprehend the problems with Energy Efficiency, the need for DT, and the methods for dealing with large-scale attack by DDoS on Energy Efficient networks. Only 94 of the 1555 articles produced by the procedure were determined to be relevant using inclusion and exclusion criteria. The outcome demonstrates that in 5G networks, DT, and its fundamental approaches, like QoS and DDoS attack mitigation, can be used to regulate the network’s Energy Efficiency. Numerous practical applications focusing on 5G Systems use their own principles. The effectiveness of these strategies was evaluated using several assessment criteria, including DT, Intrusion Detection, QoS, Energy Efficiency, and 5G Systems. Each study issue is thoroughly explained, along with typical methods, advantages, disadvantages, and performance metrics. Energy economy, network reliability, privacy, and cost reduction are all considerably increased by the implementation of intrusion detection technology in 5G systems. The decision is supported by the technology’s demonstrated efficacy, scalability, real-time detection capacities, low error level, and personalized learning attributes, all of which contribute to the long-term viability of 5G networks as an entire system.

Improved DRL-based energy-efficient UAV control for maximum lifecycle

Unmanned aerial vehicles (UAVs) operating as airborne base stations (UAV-BSs) provide efficient on-demand services to ground users. UAV-BSs are inherently flexible and mobile, allowing them to be strategically deployed based on ground user distribution and quality of service requirements, including coverage rate, system lifecycle, and user fairness. Owing to the limited battery capacity and coverage range of the UAVs, managing them to extend their operational lifecycle, ensure service fairness, and maintain a specific real-time coverage rate is challenging. Therefore, a multi-objective optimization problem with constrained Pareto dominance is formulated. Subsequently, a novel assisted deep reinforcement learning model is developed to maximize the minimum remaining energy while simultaneously considering user fairness and coverage-rate requirements. The particle swarm optimization algorithm is adopted to assist multi-agent cooperative deep reinforcement learning. Finally, the simulation results show that the proposed model outperforms the other popular methods in terms of user fairness, system lifecycle, coverage rate, and energy efficiency in the context of multi-objective, multi-agent cooperative coverage control deployment.

Efficient Task Scheduling Approach in Edge-Cloud Continuum based on Flower Pollination and Improved Shuffled Frog Leaping Algorithm

The rise of edge-cloud continuum computing is a result of the growing significance of edge computing, which has become a complementary or substitute option for traditional cloud services. The convergence of networking and computers presents a notable challenge due to their distinct historical development. Task scheduling is a major challenge in the context of edge-cloud continuum computing. The selection of the execution location of tasks, is crucial in meeting the quality-of-service (QoS) requirements of applications. An efficient scheduling strategy for distributing workloads among virtual machines in the edge-cloud continuum data center is mandatory to ensure the fulfilment of QoS requirements for both customer and service provider. Existing research used metaheuristic algorithm to solve tak scheduling problem, however, must of the existing metaheuristics used suffers from falling into local mina due to their inefficiency to avoid unfeasible region in the solution search space. Therefore, there is a dire need for an efficient metaheuristic algorithm for task scheduling. This study proposed an FPA-ISFLA task scheduling model using hybrid flower pollination and improved shuffled frog leaping algorithms. The simulation results indicate that the FPA-ISFLA algorithm is superior to the PSO algorithm in terms of makespan time, resource utilization, and execution cost reduction, especially with an increasing number of tasks.

A fine-grained robust performance diagnosis framework for run-time cloud applications

To maintain the required service quality of time-critical cloud applications, operators must continuously monitor their runtime status, detect potential performance anomalies, and diagnose the root causes of these anomalies effectively. However, existing performance diagnosis methods face challenges such as the need for high-quality labeled data, the low reusability and robustness of performance anomaly detection models, and the absence of real-time fine-grained root cause localization. These challenges make fixing performance issues quickly and developing effective adaptation decisions difficult. We provide a Fine-grained Robust Performance Diagnosis (FIRED) framework to tackle those challenges. The framework offers a metrics selection component to filter noise and improve detection efficiency, an anomaly detection component that assembles several well-selected base models with a deep neural network, and adopts weakly supervised learning considering fewer labels exist in reality. The framework also employs a real-time, fine-grained root cause localization component to locate dependent resource metrics of performance anomalies. Our experiments show that the framework can effectively reduce data noise and achieve the best accuracy and algorithm robustness for performance anomaly detection. In addition, the framework can accurately localize the first root causes, with an average accuracy higher than 0.7 for locating the first four root cause metrics.

A digital twin-based energy-efficient wireless multimedia sensor network for waterbirds monitoring

Wetlands play a critical role in maintaining the global climate, regulating the hydrological cycle, and protecting human health. However, they are rapidly disappearing due to human activities. Waterbirds are valuable bio-indicators of wetland health, but it is challenging to monitor them effectively. Wireless Multimedia Sensor Networks (WMSNs) offer a promising technology for monitoring wetlands. Nonetheless, these networks are constrained in terms of energy, and also encounter challenges associated with large-scale deployments under natural environmental conditions. These conditions introduce harsh circumstances that may not have been anticipated during the pre-deployment testing phase. This paper proposes a Digital Twin (DT) based energy-efficient WMSN monitoring system specifically tailored for waterbirds in wetlands. The system utilizes a unique approach that combines local audio identification and image compression with DT technology to optimize network performance and minimize energy consumption. To reduce unnecessary image transmissions, the system employs a real-time, low-complexity local audio identification phase before triggering image capture. A denoising step is employed to achieve highly accurate bird recognition despite surrounding noises. Each image undergoes a low-complexity compression scheme prior to transmission, further enhancing energy efficiency. To enhance the system’s overall efficiency and effectiveness, DT technology is integrated to create real-time replicas of the WMSN and the monitoring application. A synergistic interaction between the two DTs enables cooperative data-making decision that ensures both QoS (Quality of Service) and QoE (Quality of Experience) requirements are met. Transmission rate control is done using a fuzzy logic decision-making technique. Real-time feedback provides rapid and accurate analysis of the current state of the WMSN, allowing for dynamic adjustments. The ”what-if scenarios” feature of the implemented DTs has been effectively leveraged to find the most suitable settings for the controller. The effectiveness and performance enhancements achieved by integrating DT into our WMSN-based surveillance system are validated through comprehensive experiments in scenarios that correspond to a real-world wetland. Comparative analyses demonstrate the undeniable benefits of the DT-integrated system compared to a conventional WMSN-based surveillance setup. In particular, the results demonstrate the system’s superior performance in terms of energy efficiency, real-time monitoring capabilities, and ability to handle multiple video sources.

Five Disruptive Technologies in 6G to Support Digital Twin Networks

With the expected explosive growth of the number of mobile user devices and data traffic in the next decade and beyond, and the expected development of various new applications and services such as vehicle-to-everything, extended reality and smart city, the requirements for Quality of Service (such as ultra-high data transmission rate, ultra-low latency, support for the highest moving speed and seamless connection, etc.) will reach levels never seen before. The sixth generation (6G) of networks is conceived to address the requirements mentioned above. Digital Twin refers to a digital representation to represent a physical entity in virtual space while its corresponding physical object remains in the physical space. In this article, we first briefly introduce some basic knowledge of Digital Twins. We then outline possible innovative technologies expected in 6G networks that will help Digital Twin systems to reach their full potential and deliver the high-level performance needed for the next generation of communications.

Dynamic Multi-Sleeping Control with Diverse Quality-of-Service Requirements in Sixth-Generation Networks Using Federated Learning

The intensive deployment of sixth-generation (6G) base stations is expected to greatly enhance network service capabilities, offering significantly higher throughput and lower latency compared to previous generations. However, this advancement is accompanied by a notable increase in the number of network elements, leading to increased power consumption. This not only worsens carbon emissions but also significantly raises operational costs for network operators. To address the challenges arising from this surge in network energy consumption, there is a growing focus on innovative energy-saving technologies designed for 6G networks. These technologies involve strategies for dynamically adjusting the operational status of base stations, such as activating sleep modes during periods of low demand, to optimize energy use while maintaining network performance and efficiency. Furthermore, integrating artificial intelligence into the network’s operational framework is being explored to establish a more energy-efficient, sustainable, and cost-effective 6G network. In this paper, we propose a small base station sleeping control scheme in heterogeneous dense small cell networks based on federated reinforcement learning, which enables the small base stations to dynamically enter appropriate sleep modes, to reduce power consumption while ensuring users’ quality-of-service (QoS) requirements. In our scheme, double deep Q-learning is used to solve the complex non-convex base station sleeping control problem. To tackle the dynamic changes in QoS requirements caused by user mobility, small base stations share local models with the macro base station, which acts as the central control unit, via the X2 interface. The macro base station aggregates local models into a global model and then distributes the global model to each base station for the next round of training. By alternately performing model training, aggregation, and updating, each base station in the network can dynamically adapt to changes in QoS requirements brought about by user mobility. Simulations show that compared with methods based on distributed deep Q-learning, our proposed scheme effectively reduces the performance fluctuations caused by user handover and achieves lower network energy consumption while guaranteeing users’ QoS requirements.