One of the pillars of the 5G architecture is network slicing, in which hardware, radio, and power resources are virtualized as a logical network taking into account the requirements of diverse applications. While ensuring performance isolation among different slices, resource allocation in 5G Radio Access Networks (RANs) is associated with different challenges due to network dynamics and the different applications requirements. In this paper, we have considered the allocation of power and radio resources to rate-based as well as resource-based users. We propose an energy-efficient deep reinforcement learning-assisted resource allocation (EE-DRL-RA) method for RAN slicing in 5G networks. The main idea of the proposed method is to exploit a collaborative learning framework that includes deep reinforcement learning (DRL) and deep learning (DL) to decide on resource allocation in the RAN. Specifically, we use DL for decision-making on resource allocation on a large time-scale and DRL for decision-making on resource allocation on a small time-scale. The asynchronous advantage actor-critic (A3C) and the stacked and bidirectional long-short-term-memory (SBiLSTM) network are used as DRL and supervised DL methods, respectively. Furthermore, we determine the optimal power and resource blocks (RBs) for rate-based users by formulating the energy-efficient power allocation (EE-PA) problem as a non-convex optimization problem and solve it by an efficient iterative algorithm. Our proposed approach is unique in that it simultaneously allocates power and RBs while ensuring slice isolation with low computational and time complexity. Simulation results show that EE-DRL-RA yields better performance compared to a state-of-the-art published method in terms of convergence speed, computational complexity, energy efficiency, and the number of accepted users as well as the degree of inter-slice isolation.

One of the key foundations of 5 ^th Generation (5G) and beyond 5G (B5G) networks is network slicing, in which the network is partitioned into several separated logical networks, taking into account the requirements of diverse applications. In this context, resource management is of great importance to instantiate and operate network slices and meet their performance and functional requirements. Resource management in Radio Access Networks (RANs) is associated with a range of challenges due to network dynamics and the specific requirements of each application while ensuring performance isolation. In this paper, we present a survey on state-of-the-art works that employ Machine Learning (ML) techniques in RAN slicing. We begin by reviewing the challenges, then we review the existing papers on resource management in a comprehensive manner, and classify the papers based on the used ML algorithm, the addressed challenges, and the type of allocated resources. We evaluate the maturity of current methods and state a number of open challenges and some solutions to address these challenges in RAN resource management.

In 5G wireless networks, cooperative non-orthogonal multiple access (NOMA) and wireless power transfer (WPT) are efficient ways to improve the spectral efficiency (SE) and energy efficiency (EE). In this paper, a new cooperative NOMA scheme with WPT is proposed, where EE optimization with a constrained maximum transmit power and minimum required SE is considered for the user grouping and transmit power allocation of users. We obtain a sub-optimal solution by decoupling the original problem in two sub-problems: an iterative algorithm is considered for the user grouping, while, in addition, we utilize the Bat Algorithm (BA) for solving the power allocation problem, where BA was proved to be able to achieve a higher accuracy and efficiency with respect to other meta-heuristic algorithms. Furthermore, to validate the performance of the proposed system, analytical expressions for the energy outage probability and outage probability of users are derived, confirming the effectiveness of the simulation results. It is demonstrated that the proposed cooperative NOMA with WPT offers a considerable improvement in terms of SE and EE of the network compared to other methods. Finally, the effectiveness of BA in solving the EE optimization problem is demonstrated through a high convergence speed by comparing it with other methods.

Non-orthogonal multiple access (NOMA) and millimeter-wave (mmWave) communication are two important technologies considered for the fifth generation (5G) of mobile networks. In this paper, we propose a machine-to-machine (M2M) communication underlaying heterogeneous network (HetNet) with mmWave-NOMA transmission, where small cell users (SCUs) and machine type communication (MTC) devices share the same resource block (RB) to increase the network capacity. We formulate an optimization problem to select efficient users and allocate transmission powers to SCUs and MTC devices. The object of optimization is to maximize the energy efficiency (EE) subject to the minimum spectral efficiency (SE) requirements and maximum transmission powers. We present a low computational complexity sub-optimum solution based on the non-cooperative game. An iterative algorithm with fast convergence speed and a reduced complexity search-based algorithm are respectively presented to solve the power control and user selection problems. The comparison between the proposed solution and optimum solution for different number of transmit antennas indicate that there is a small gap between SEs and EEs of the proposed and optimum solutions. Further, the results of user selection scheme demonstrate that the location of strong user affects the performance. We also derive analytical expressions for the outage probability which confirm Monte Carlo simulations.

Sleep has great effect on physical health and quality of life. Electroencephalogram (EEG) signal is used in studying sleep process and recently, time–frequency transforms are increasingly utilised in EEG signal analysis. This study proposes an efficient method for sleep stages classification based on a time–frequency transform, namely Stockwell transform. In the introduced method, at first, the Stockwell transform is used to map each 30 s epoch of EEG signal into the time–frequency domains, which results in a complex‐valued matrix. Then, the frequency domain is divided into different non‐overlapping segments, leading to several matrices. After that, entropy features are extracted from the obtained matrices. In order to determine the sleep stage of each epoch, the computed features are applied to classifier. Support vector machine, weighted K ‐nearest neighbour, and ensemble bagged tree classifiers are considered. The Pz–Oz and Fpz–Cz channels of EEG signal from Sleep‐EDF data set and C3–A2 channel from ISRUC‐Sleep data set are used in this research. The results indicate that the proposed method outperforms the recently introduced methods.