Large-scale Internet of Things (IoT) networks enable intelligent services such as smart cities and autonomous driving, but often face resource constraints. Collecting heterogeneous sensory data, especially in small-scale datasets, is challenging, and independent edge nodes can lead to inefficient resource utilization and reduced learning performance. To address these issues, this paper proposes a collaborative optimization framework for energy-efficient federated edge learning with small-scale datasets. We first derive an expected learning loss to quantify the relationship between the number of training samples and learning objectives. A stochastic online learning algorithm is then designed to adapt to data variations, and a resource optimization problem with a convergence bound is formulated. Finally, an online distributed algorithm efficiently solves large-scale optimization problems with high scalability. Extensive simulations and autonomous navigation case studies with collision avoidance demonstrate that the proposed approach significantly improves learning performance and resource efficiency compared to state-of-the-art benchmarks.

Identifying the top-K flows that require much more bandwidth resources in a large-scale Software-Defined Network (SDN) is essential for many network management tasks, such as load balancing, anomaly detection, and traffic engineering. However, identifying such top-K flows is not trivial, not only because of the fluctuations in flow bandwidth requirements but also because of the combinatorial explosion of problem instance sizes. In this paper, we weaken the tradeoff between exploration and exploitation and innovatively define the online top-K flows identification problem as identifying the top-K arms in a Combinatorial Multi-Armed Bandit (CMAB) model. Then, we propose a general greedy selection mechanism with some identification strategies that focus on temporal variations in the rewards. Extensive simulation experiments based on real traffic data are conducted to evaluate the performance of different strategies. In addition, the results of numerical simulations demonstrate that our proposed greedy selection mechanism significantly outperforms existing counterparts on top-K arms identification.

Service Function Chaining (SFC) is regarded as an important concept for next-generation communication networks because it can flexibly tackle diverse usage scenarios. Due to SFC requests’ life-cycle and resource adjustment, the distribution of the remaining physical resources may become unbalanced, which brings negative effects to subsequent SFC requests as well as network operators. In this paper, we investigate the network SFC migration problem in the core cloud under the premise of considering the migration cost and the balance of physical resource distribution. We first model the SFC migration problem as an integer linear program and propose an aggressive migration strategy that can effectively reduce the imbalance of physical resource distribution. Then, we employ two state-of-the-art heuristics to allocate resources for subsequent SFC requests. The simulation results show that migrating SFC requests in the initial service queue can bring favorable feedback to subsequent requests as well as network operators. Compared to the conservative migration strategy, our proposed migration strategy can mitigate the imbalance of physical resource distribution more effectively, and thus the acceptance ratio of subsequent SFC requests, physical resources utilization, and the long-term profit of network operators can be further improved.

Summary The paper illustrates a parallel and distributed scheme for computing a planar Delaunay triangulation using a divide‐and‐conquer strategy in Cloud environment, which combines the incremental insertion algorithm and the divide‐and‐conquer method. The proposed hybrid algorithm for Delaunay triangulation construction is easy to be parallelized due to the dynamic pruned characteristic of the binary tree model used. Moreover, the Cloud platform decreases the communication overhead and improves data locality by making use of a data partitioning and integrating scheme offered by the map‐reduce architecture. The implementation of the parallel and distributed version of the algorithm relied on a robust data structure called quad‐edge, which implies the geometric relationship among the edges and vertexes adjacent. More importantly, the data are serialized easily and transmitted efficiently between different Cloud nodes; the algorithm is executed conveniently on PC clusters. We tested the parallel version of the algorithm on GeoKSCloud, a geographical knowledge service Cloud developed by our research team. Experimental results show that the proposed hybrid algorithm is efficient and competitive; it can be easily migrated and deployed in distributed and parallel computing environment, such as grid and Cloud. The parallel implementation of the hybrid algorithm has a good speed‐up, while data communication is the crucial factor for the efficiency of the parallel version. Overall, the parallel version outperforms both the sequential divide‐and‐conquer algorithm and the sequential incremental insertion algorithm.

Delaunay triangulation (D-TIN) is an important graphic tool in computational geometry, which is not only widely used in many real applications, but also very significant for many spatial data mining algorithms. However, constructing Delaunay triangulation is time-consuming for most practical applications. Distributed and parallel computing mechanism is becoming a good choice to solve large scale and compute-intensive D-TIN applications. This paper proposes a novel hybrid algorithm (HA) for D-TIN construction in cloud computing environment, which is based on a balanced binary-tree model and an elegant data structure called quad-edge. HA combines the divide & conquer approach and the incremental method. Moreover, a distributed and parallel version of Delaunay triangulation computing service in cloud is designed and implemented. The hybrid algorithm performed in both centralised and in cloud environments are compared. Experimental results showed that the hybrid D-TIN service outperforms both the the divide & conquer one and the incremental one, and it can effectively provide higher data mining services with fundamental D-TIN construction function in cloud.