Search

Hierarchical Federated Learning (HFL) has emerged to overcome the shortcomings of conventional Federated Learning (FL) due to communication obstacles between the end users and the cloud server and the congestion at the backhaul of wireless network implementations. In this paper, we consider a wireless user-edge-cloud HFL network where the transmissions of the users’ local model parameters to the edge are multiplexed via the Non-Orthogonal Multiple Access (NOMA) technique. The joint problem of association and uplink transmission power allocation of the users to the edge is formulated and solved as a non-cooperative game in satisfaction form. Diverging from the prevailing research that proposes centralized solution concepts, each user makes autonomous decisions regarding its association and power level so as to attain a minimum acceptable tradeoff of three vital network factors. The latter includes the global model’s training accuracy and the users’ consumed energy and time during transmission. Different types of equilibria are explored, i.e., the Satisfaction Equilibrium (SE) and Minimum Efficient Satisfaction Equilibrium (MESE) which not only fulfills users’ minimum tradeoff but also minimizes the overall network’s cost. Algorithms based on Reinforcement Learning (RL) and Best Response Dynamics (BRD) are, then, devised to conclude the SE and MESE points. The proposed framework is evaluated via modeling and simulation, verifying its efficiency in achieving an equitable balance in the network.

Rate-Splitting Multiple Access (RSMA) has been recognized as an effective technique to reconcile the tradeoff between decoding interference and treating interference as noise in 6G and beyond networks. In this paper, in line with the need for network sustainability, we study the energy-efficient power and rate allocation of the common and private messages transmitted in the downlink of a single-cell single-antenna RSMA network. Contrary to the literature that resorts to heuristic approaches to deal with the joint problem, we transform the formulated energy efficiency maximization problem into a multi-agent Deep Reinforcement Learning (DRL) problem, based on which each transmitted private message represents a different DRL agent. Each agent explores its own state-action space, the size of which is fixed and independent of the number of agents, and shares its gained experience by exploration with a common neural network. Two DRL algorithms, namely the value-based Deep Q-Learning (DQL) and the policy-based REINFORCE, are properly configured and utilized to solve it. The adaptation of the proposed DRL framework is also demonstrated for the treatment of the considered network’s sum-rate maximization objective. Numerical results obtained via modeling and simulation verify the effectiveness of the proposed DRL framework to conclude a solution to the joint problem under both optimization objectives, outperforming existing heuristic approaches and algorithms from the literature.

The edge computing paradigm has become extremely popular over the past years, as a means of offloading computationally intensive tasks by users of resource and battery-constrained devices. Nevertheless, the edge networks’ overexploitation by the ever-increasing number of task-offloading users, gradually leads to their performance degradation. In this paper, we leverage on the different levels of available computing capabilities across the network, and we design an incentive mechanism that aims to shift the selfish users’ preference from the edge to the upper fog computing layer, accounting for their level of delay tolerance. To deal with the users’ heterogeneity in terms of their applications’ multi-dimensional distinctive features (including their delay tolerance/sensitivity), a multi-dimensional contract theory modeling is adopted, according to which the edge server determines the bundles of the users’ provided efforts and corresponding offered rewards. In this respect, each user’s effort represents the amount of its initially offloaded task at the edge that is allowed to be further forwarded and processed at the fog. Considering that the users-to-edge server offloading is performed under Non-Orthogonal Multiple Access (NOMA), the problem of joint computation task offloading and uplink transmission power allocation is subsequently addressed via a Stackelberg game, where the edge server and the users are treated as leader and followers, respectively. The aim of the game is to minimize the end-to-end network’s energy consumption and increase its resource utilization efficiency. The incentive mechanism and resource allocation framework is evaluated via modeling and simulation regarding its operation and efficiency under different scenarios.

Lately, the deployment of heterogeneous wireless networks has emerged, as part of the 5G vision, to cope with the users' exaggerated service demands. In this context, the application of Non-Orthogonal Multiple Access (NOMA) technique constitutes a promising solution to facilitate a balance between spectral efficiency and system complexity. In this article, we consider the problem of joint user association and uplink power allocation, in heterogeneous 5G wireless networks, employing NOMA technology. The coupled problem is treated under an incomplete information scenario, where the Base Stations (BSs) have statistical only knowledge of the users' channel conditions. To deal with the incompleteness of Channel State Information (CSI), a Contract Theory (CT) based approach is introduced. A Reinforcement Learning (RL) based methodology, capitalizing on the provided feedback from the communication environment is initially adopted, in order to achieve the users to BS association in an iterative and distributed manner. The problem of uplink power allocation is subsequently formulated as a contract between each BS and its corresponding users. The optimal power is thus obtained as the solution of the optimization of each BS's utility function, while ensuring the optimality of the utility function of each associated user, given the unique communications characteristics and type of each user. Detailed numerical evaluation of the performance of the proposed unified user association and power allocation framework is provided, via modeling and simulation, illustrating its operation, features and benefits, under densely deployed heterogeneous environments.

The ongoing transition towards 5G technology expedites the emergence of a variety of mobile applications that pertain to different vertical industries. Delivering on the key commitment of 5G, these diverse service streams, along with their distinct requirements, should be facilitated under the same unified network infrastructure. Consequently, in order to unleash the benefits brought by 5G technology, a holistic approach towards the requirement analysis and the design, development, and evaluation of multiple concurrent vertical services should be followed. In this paper, we focus on the Transport vertical industry, and we study four novel vehicular service categories, each one consisting of one or more related specific scenarios, within the framework of the “5G Health, Aquaculture and Transport (5G-HEART)” 5G PPP ICT-19 (Phase 3) project. In contrast to the majority of the literature, we provide a holistic overview of the overall life-cycle management required for the realization of the examined vehicular use cases. This comprises the definition and analysis of the network Key Performance Indicators (KPIs) resulting from high-level user requirements and their interpretation in terms of the underlying network infrastructure tasked with meeting their conflicting or converging needs. Our approach is complemented by the experimental investigation of the real unified 5G pilot’s characteristics that enable the delivery of the considered vehicular services and the initial trialling results that verify the effectiveness and feasibility of the presented theoretical analysis.