The telecom sector has experienced numerous automation phases. While previous connections were still made manually by switching cables, hardware later automated this operation. It is no longer a question of whether the rapid development of AI will be affecting or even disrupt the majority of the industry. In addition, the telecom industry is not an exception. Undoubtedly, the emergence of AI, Data Science, and Machine Learning will be enabling telecom firms to operate better, make investments, and additional revenue.

With the increase of intelligent devices leading to increasing demand for traffic, traffic offloading has become a challenging problem. The space-air-ground integrated network (SAGIN) is a superior network architecture to solve this problem. The existing research on SAGIN traffic offloading only considers the single-layer satellite network in the space network. To further expand the resource pool of traffic offloading in SAGIN, we extend the single-layer satellite network into a double-layer satellite network composed of low-orbit satellites (LEO) and high-orbit satellites (GEO). And re-model a four-layer SAGIN architecture consisting of the ground network, the air network, LEO and GEO. Furthermore, we propose a novel Federated Soft Actor-Critic (FeSAC) traffic offloading method with positive environmental exploration to accommodate this dynamic and complex four-layer SAGIN architecture. The FeSAC method uses federated learning to train traffic offloading nodes and then aggregate the training results to obtain the best traffic offloading strategy. The simulation results show that under the four-layer SAGIN, our proposed method can better adapt to the network environment changes by nodes mobility and is better than the existing traffic offloading methods in throughput, packet loss, and transmission delay.

The flexibility introduced with the Open Radio Access Network (O-RAN) architecture allows us to think beyond static configurations in all parts of the network. This paper addresses the issue related to predicting the power consumption of different radio schedulers, and the potential offered by O-RAN to collect data, train models, and deploy policies to control the power consumption. We propose a black-box (Neural Network) model to learn the power consumption function. We compare our approach with a known hand-crafted solution based on domain knowledge. Our solution reaches similar performance without any previous knowledge of the application and provides more flexibility in scenarios where the system behavior is not well understood or the domain knowledge is not available.

This article explores the concepts of online protocol synthesis using Reinforcement Learning (RL). The study is performed in the context of sensor and IoT networks with ultra low complexity wireless transceivers. The paper introduces the use of RL and Multi Armed Bandit (MAB), a specific type of RL, for Medium Access Control (MAC) under different network and traffic conditions. It then introduces a novel learning based protocol synthesis framework that addresses specific difficulties and limitations in medium access for both random access and time slotted networks. The mechanism does not rely on carrier sensing, network time-synchronization, collision detection, and other low level complex operations, thus making it ideal for ultra simple transceiver hardware used in resource constrained sensor and IoT networks. Additionally, the ability of independent protocol learning by the nodes makes the system robust and adaptive to the changes in network and traffic conditions. It is shown that the nodes can be trained to learn to avoid collisions, and to achieve network throughputs that are comparable to ALOHA based access protocols in sensor and IoT networks with simplest transceiver hardware. It is also shown that using RL, it is feasible to synthesize access protocols that can sustain network throughput at high traffic loads, which is not feasible in the ALOHA-based systems. The ability of the system to provide throughput fairness under network and traffic heterogeneities are also experimentally demonstrated.

5G New Radio proposes the usage of frequencies above 10 GHz to speed up LTE's existent maximum data rates. However, the effective size of 5G antennas and consequently its repercussions in the signal degradation in urban scenarios makes it a challenge to maintain stable coverage and connectivity. In order to obtain the best from both technologies, recent dual connectivity solutions have proved their capabilities to improve performance when compared with coexistent standalone 5G and 4G technologies. Reinforcement learning (RL) has shown its huge potential in wireless scenarios where parameter learning is required given the dynamic nature of such context. In this paper, we propose two reinforcement learning algorithms: a single agent RL algorithm named Clipped Double Q-Learning (CDQL) and a hierarchical Deep Q-Learning (HiDQL) to improve Multiple Radio Access Technology (multi-RAT) dual-connectivity handover. We compare our proposal with two baselines: a fixed parameter and a dynamic parameter solution. Simulation results reveal significant improvements in terms of latency with a gain of 47.6% and 26.1% for Digital-Analog beamforming (BF), 17.1% and 21.6% for Hybrid-Analog BF, and 24.7% and 39% for Analog-Analog BF when comparing the RL-schemes HiDQL and CDQL with the with the existent solutions, HiDQL presented a slower convergence time, however obtained a more optimal solution than CDQL. Additionally, we foresee the advantages of utilizing context-information as geo-location of the UEs to reduce the beam exploration sector, and thus improving further multi-RAT handover latency results.

In this article, we advocate for the design of ad hoc artificial intelligence (AI)/machine learning (ML) models to facilitate their usage in future smart infrastructures based on communication networks. To motivate this, we first review key operations identified by the 3GPP for transferring AI/ML models through 5G networks and the main existing techniques to reduce their communication overheads. We also present a novel communication-aware ML framework, which we refer to as Accordion, that enables an efficient AI/ML model transfer thanks to an overhauled model training and communication protocol. We demonstrate the communication-related benefits of Accordion, analyse key performance trade-offs, and discuss potential research directions within this realm. Artificial intelligence (AI) and machine learning (ML) are becoming ubiquitous in our lives. At a consumer level, most used applications in our smartphones rely on AI/ML algorithms for image recognition and video editing. Due to their benefits, vendors and operators are also currently investing on AI/ML models to plan, design, and operate their fifth generation (5G) networks [1]. From an operational perspective, the integration of neural processing units and the execution of AI/ML models in smartphones are becoming commonplace to satisfy the latency constraints of complex applications and/or safeguard their data privacy.

The Neural Architecture Search (NAS) problem is typically formulated as a graph search problem where the goal is to learn the optimal operations over edges in order to maximise a graph-level global objective. Due to the large architecture parameter space, efficiency is a key bottleneck preventing NAS from its practical use. In this paper, we address the issue by framing NAS as a multi-agent problem where agents control a subset of the network and coordinate to reach optimal architectures. We provide two distinct lightweight implementations, with reduced memory requirements (1/8th of state-of-the-art), and performances above those of much more computationally expensive methods. Theoretically, we demonstrate vanishing regrets of the form O(sqrt(T)), with T being the total number of rounds. Finally, aware that random search is an, often ignored, effective baseline we perform additional experiments on 3 alternative datasets and 2 network configurations, and achieve favourable results in comparison.

Boomi, the intelligent connectivity and automation leader, announced that MyRepublic Digital, the digital transformation company powering MyRepublic, an award-winning telecom operator in the Asia-Pacific, has implemented The Boomi AtomSphere Platform to achieve higher levels of efficiency, performance, and ultimately a better experience for customers. "For instance, we've observed a marked reduction in customer waiting time for modems between one and two business days, allowing us to provide faster service." MyRepublic Digital, a recently launched technology business spin-out, helps operators achieve their business outcomes with its readily deployable cloud-native multi-tenant software as a service (SaaS) platform, Encore. In the hyper-competitive telecom market, staying fast and agile when it comes to service delivery is key to success. To enhance the efficiencies of the Encore platform, and its real-time decision-making capabilities, without compromising on scalability or security, MyRepublic Digital selected Boomi's award-winning integration platform as a service (iPaaS).

An enhanced framework for peak-to-average power ratio ($\mathsf{PAPR}$) reduction and waveform design for Multiple-Input-Multiple-Output ($\mathsf{MIMO}$) orthogonal frequency-division multiplexing ($\mathsf{OFDM}$) systems, based on a convolutional-autoencoder ($\mathsf{CAE}$) architecture, is presented. The end-to-end learning-based autoencoder ($\mathsf{AE}$) for communication networks represents the network by an encoder and decoder, where in between, the learned latent representation goes through a physical communication channel. We introduce a joint learning scheme based on projected gradient descent iteration to optimize the spectral mask behavior and MIMO detection under the influence of a non-linear high power amplifier ($\mathsf{HPA}$) and a multipath fading channel. The offered efficient implementation novel waveform design technique utilizes only a single $\mathsf{PAPR}$ reduction block for all antennas. It is throughput-lossless, as no side information is required at the decoder. Performance is analyzed by examining the bit error rate ($\mathsf{BER}$), the $\mathsf{PAPR}$, and the spectral response and compared with classical $\mathsf{PAPR}$ reduction $\mathsf{MIMO}$ detector methods on 5G simulated data. The suggested system exhibits competitive performance when considering all optimization criteria simultaneously. We apply gradual loss learning for multi-objective optimization and show empirically that a single trained model covers the tasks of $\mathsf{PAPR}$ reduction, spectrum design, and $\mathsf{MIMO}$ detection together over a wide range of SNR levels.

Network slicing is a crucial enabler to support the composition and deployment of virtual network infrastructures required by the dynamic behavior of networks like 5G/6G mobile networks, IoT-aware networks, e-health systems, and industry verticals like the internet of vehicles (IoV) and industry 4.0. The communication slices and their allocated communication resources are essential in slicing architectures for resource orchestration and allocation, virtual network function (VNF) deployment, and slice operation functionalities. The communication slices provide the communications capabilities required to support slice operation, SLA guarantees, and QoS/ QoE application requirements. Therefore, this contribution proposes a networking slicing conceptual model to formulate the optimization problem related to the sharing of communication resources among communication slices. First, we present a conceptual model of network slicing, we then formulate analytically some aspects of the model and the optimization problem to address. Next, we proposed to use a SARSA agent to solve the problem and implement a proof of concept prototype. Finally, we present the obtained results and discuss them.