The applications of Unmanned Aerial Vehicles (UAVs) are rapidly growing in domains such as surveillance, logistics, and entertainment and require continuous connectivity with cellular networks to ensure their seamless operations. However, handover policies in current cellular networks are primarily designed for ground users, and thus are not appropriate for UAVs due to frequent fluctuations of signal strength in the air. This paper presents a novel handover decision scheme deploying Deep Reinforcement Learning (DRL) to prevent unnecessary handovers while maintaining stable connectivity. The proposed DRL framework takes the UAV state as an input for a proximal policy optimization algorithm and develops a Received Signal Strength Indicator (RSSI) based on a reward function for the online learning of UAV handover decisions. The proposed scheme is evaluated in a 3D-emulated UAV mobility environment where it reduces up to 76 and 73% of unnecessary handovers compared to greedy and Q-learning-based UAV handover decision schemes, respectively. Furthermore, this scheme ensures reliable communication with the UAV by maintaining the RSSI above −75 dBm more than 80% of the time.

Carrier-grade networks (CGNs) can leverage the network programmability of software-defined networking (SDN) to ensure fast recovery and high availability. However, for the successful adoption of SDN, the failure recovery requirement must be addressed. Local detouring is a popular approach for faster recovery rather than path-based end-to-end recovery. For fast local recovery, alternate paths must be preinstalled for each individual flow on the link, which in some cases results in storing thousands of alternate path flow rules. Furthermore, the dependence on the controller for dynamic per-flow detouring may delay the recovery. In this paper, we propose local immediate (LIm) and immediate controller dependent (ICoD) recovery mechanisms to address the limitations of OpenFlow-based link recovery approaches. Our proposed mechanisms considerably reduce the alternate path flow rules by aggregating the disrupted flows using virtual local area network (VLAN) tagging. The proposed algorithms achieve recovery within 3 ms and 20 ms, respectively and satisfy the strict 50 ms recovery requirement of CGNs. LIm and ICoD also reduce the alternate path flow storage requirement by up to 99%. Simulation results reveal that the flow-aggregation also reduces the effort of the controller and minimizes the alternate path installation traffic.

Network slicing is an emerging paradigm aiming to provide use case tailored virtual networks. However, it poses new challenges for slice resource assignment. Specifically, new admission control policies are required to model delay-tolerant slices. Fair admission of requests from different willingness-to-pay subscribers while maximizing resource monetization is another major problem. This paper proposes a novel slice admission model with delayed admission and quota constraints to maximize the monetization and minimize the service agreement violation. We investigate a tradeoff between the monetization and the admission rate, that rise due to quota-driven resource allocation. We map the slice admission to the multiple knapsack problem, formulate an integer programming model, and present a partially adaptive greedy (PAGE) algorithm to tackle this combinatorial optimization problem. Simulation results validate the efficacy of the PAGE over classical heuristics.

Containers virtually package a piece of software and share the host Operating System (OS) upon deployment. This makes them notably light weight and suitable for dynamic service deployment at the network edge and Internet of Things (IoT) devices for reduced latency and energy consumption. Data collection, computation, and now intelligence is included in variety of IoT devices which have very tight latency and energy consumption conditions. Recent studies satisfy latency condition through containerized services deployment on IoT devices and gateways. They fail to account for the limited energy and computing resources of these devices which limit the scalability and concurrent services deployment. This paper aims to establish guidelines and identify critical factors for containerized services deployment on resource constrained IoT devices. For this purpose, two container orchestration tools (i.e., Docker Swarm and Kubernetes) are tested and compared on a baseline IoT gateways testbed. Experiments use Deep Learning driven data analytics and Intrusion Detection System services, and evaluate the time it takes to prepare and deploy a container (creation time), Central Processing Unit (CPU) utilization for concurrent containers deployment, memory usage under different traffic loads, and energy consumption. The results indicate that container creation time and memory usage are decisive factors for containerized micro service architecture.

Importance of user mobility has rapidly increased in 5G due to reduced cell sizes, management of Multi-access Edge Computing (MEC), and ultra-low latency services. Reactive nature of existing management systems is a bottleneck, and it can be solved by building proactive systems that exploit temporal characteristics of time-series mobility data to predict long-term user movement (i.e.path). However, user mobile path prediction with useable accuracy is a challenging task, particularly for lengthy target trajectories. This paper adopts general approaches to propose two models for predicting mobile path with high accuracy. Step Forward Iteration (SFI) model is based on recursive approach, whereas Encoder-Decoder (ED) model follows multi-output approach, and both the models use Long-Short Term Memory (LSTM) as the learning unit. Training and testing of these models is done on mobility datasets from the wireless network of Pangyo ICT Research Center, Korea and one of the Korean mobile operators. The experiment results show viability of the proposed models for leveraging mobile network management, as they outperform state-of-the-art GRU with attention (GRU-ATTN) and Transformer Network (TN) models. The highest prediction accuracies achieved for 3, 5, and 7 steps of target sequences (i.e.predicted mobile path) in the campus dataset are 96%, 90%, and 87%, respectively.

Fast data aggregation is crucial for facilitating critical Internet of Things services as it enables the collection of sensory data within strict volume and time constraints. Over the past decades, the data aggregation scheduling problem for minimum latency has garnered significant research attention. Existing approaches to this problem typically schedule all data transmissions based on an aggregation tree, which is constructed without secondary interference. However, such interference can introduce delays when scheduling a transmission from a node to its parent in the tree. To this end, this study proposes an approach called active neighbor exploitation (ANEX) that enables sensor nodes to switch their parents by identifying active neighbors for potential connectivity, irrespective of the receivers established in the tree. Additionally, the scheme prioritizes scheduling nodes with the fewest unscheduled active neighbors, thereby allowing for more concurrent transmissions. ANEX is evaluated through theoretical analysis and extensive simulations under various scenarios. The results demonstrate that ANEX achieves up to 86% faster aggregation compared to the state-of-the-art approach while maintaining an equivalent time complexity.