5G New Radio (NR) network deployment in Non-Stand Alone (NSA) mode means that 5G networks rely on the control plane of existing Long Term Evolution (LTE) modules for control functions, while 5G modules are only dedicated to the user plane tasks, which could also be carried out by LTE modules simultaneously. The first deployments of 5G networks are essentially using this technology. These deployments enable what is known as E-UTRAN NR Dual Connectivity (ENDC), where a user establish a 5G connection simultaneously with a pre-existing LTE connection to boost their data rate. In this paper, a single Federated Deep Reinforcement Learning (FDRL) agent for the optimization of the event that triggers the dual connectivity between LTE and 5G is proposed. First, single Deep Reinforcement Learning (DRL) agents are trained in isolated cells. Later, these agents are merged into a unique global agent capable of optimizing the whole network with Federated Learning (FL). This scheme of training single agents and merging them also makes feasible the use of dynamic simulators for this type of learning algorithm and parameters related to mobility, by drastically reducing the number of possible combinations resulting in fewer simulations. The simulation results show that the final agent is capable of achieving a tradeoff between dropped calls and the user throughput to achieve global optimum without the need for interacting with all the cells for training.

Traditionally, mobile network management has been based on reactive systems, where corrective actions are taken when a failure or a suboptimal network performance is detected. However, the requirements of low latency and high throughput associated with 5G new services, make these reactive systems insufficient to manage the new needs of users. Thus, the focus of mobile network management has changed to a proactive approach, where preventive actions are taken to avoid failures and suboptimal network performance. Moreover, in 5G, management algorithms are expected to make use of large amounts of information sources. The use of this large amount of information and the need to use historical data to generate forecasting models, may cause a loss of accuracy. To address these challenges, a forecasting framework is proposed in this work. The framework is based on the use of automatic feature selection techniques in both spatial and temporal dimensions as a pre-prediction stage. In this way, the proposed framework can improve the accuracy of other prediction schemes when using large amounts of data as input. The proposed scheme has been tested using different forecasting techniques. In addition, the impact of the proposed framework on predictions of different future lags has been analyzed.

This paper proposes a distributed deep reinforcement learning (DRL) methodology for autonomous mobile robots (AMRs) to manage radio resources in an indoor factory with no network infrastructure. Hence, deep neural networks (DNN) are used to optimize the decision policy of the robots, which will make decisions in a distributed manner without signalling exchange. To speed up the learning phase, a centralized training is adopted in which a single DNN is trained using the experience from all robots. Once completed, the pre-trained DNN is deployed at all robots for distributed selection of resources. The performance of this approach is evaluated and compared to 5G NR sidelink mode 2 via simulations. The results show that the proposed method achieves up to 5% higher probability of successful reception when the density of robots in the scenario is high.

In Self-Organizing Networks (SON), load balancing has been recognized as an effective means to increase network performance. In cellular networks, cell load balancing can be achieved by tuning handover parameters, for which a Fuzzy Logic Controller (FLC) usually provides good performance and usability. Operator experience can be used to define the behavior of the FLCs. However, such a knowledge is not always available and hence optimization techniques must be applied in the controller design. In this work, a fuzzy $Q$-Learning algorithm is proposed to find the optimal set of fuzzy rules in an FLC for traffic balancing in GSM-EDGE Radio Access Network (GERAN). Load balancing is performed by modifying handover margins. Simulation results show that the optimized FLC provides a significant reduction in call blocking.

In this paper the behavior and the self-optimization of an LTE network under realistic conditions are investigated. To enhance network performance in a urban environment a controller to auto-tune parameters has been proposed. An urban mobility model has also been defined in order to test the proposed method under realistic conditions. This model allows to investigate some performance features, wich are not visible with a simple mobility model. In this paper, we propose to use a Fuzzy Logic Controller (FLC) for the optimization of handover parameters for adaptive load balancing. Results show that under an agglomeration of vehicles in a main road of the scenario, the proposed method achieves an improvement in the global Call Blocking Ratio (CBR).