In recent years, various internet architectures, such as Integrated Services (IntServ), Differentiated Services (DiffServ), Time Sensitive Networking (TSN) and Deterministic Networking (DetNet), have been proposed to meet the quality-of-service (QoS) requirements of different network services. Concurrently, network calculus has found widespread application in network modeling and QoS analysis. Network calculus abstracts the details of how nodes or networks process data packets using the concept of service curves. This paper summarizes the service curves for typical scheduling algorithms, including Strict Priority (SP), Round Robin (RR), Cycling Queuing and Forwarding (CQF), Time Aware Shaper (TAS), Credit Based Shaper (CBS), and Asynchronous Traffic Shaper (ATS). It introduces the theory of network calculus and then provides an overview of various scheduling algorithms and their associated service curves. The delay bound analysis for different scheduling algorithms in specific scenarios is also conducted for more insights.

With the increasingly fierce competition among communication operators, it is more and more important to make an accurate prediction of potential off-grid users. To solve the above problem, it is inevitable to consider the effectiveness of learning algorithms, the efficiency of data processing, and other factors. Therefore, in this paper, we, from the practical application point of view, propose a potential customer off-grid prediction system based on Spark, including data pre-processing, feature selection, model building, and effective display. Furthermore, in the research of off-grid system, we use the Spark parallel framework to improve the gcForest algorithm which is a novel decision tree ensemble approach. The new parallel gcForest algorithm can be used to solve practical problems, such as the off-grid prediction problem. Experiments on two real-world datasets demonstrate that the proposed prediction system can handle large-scale data for the off-grid user prediction problem and the proposed parallel gcForest can achieve satisfying performance.

Content distribution in large-scale vehicular ad hoc networks is difficult due to the scalability issue. A message may need to be carried by several vehicles till it reaches the destination. To select an appropriate next-hop carrier, the current carrier should exchange control messages with a large number of vehicles encountered, and thus the pure ad hoc solution is not scalable. In this paper, we introduce a hybrid-network solution. We first divide the area into regions, and select a hot spot in each region to install a road-side unit (RSU). RSUs can coordinate message exchanges between vehicles, and storage devices are used to temporarily hold a message waiting for the next-hop carrier. The RSUs and the vehicles traveling between them construct an overlay store-carry-and-forward content distribution network. Two types of vehicles exist, one with fixed mobility patterns such as buses, and the other with random patterns such as taxis. Considering one or both types of vehicles, utility-based optimization problems can be formulated to find the optimal routing solutions. Using the bus and taxi traces of Shanghai city, we demonstrate the effectiveness of the hybrid framework in terms of delivery delay, delivery ratio and overhead ratio.