Beijing University of Posts and Telecommunications

About: Beijing University of Posts and Telecommunications is a education organization based out in Beijing, Beijing, China. It is known for research contribution in the topics: MIMO & Quality of service. The organization has 39576 authors who have published 41525 publications receiving 403759 citations. The organization is also known as: BUPT.

Tuning the range, magnitude, and sign of the thermal expansion in intermetallic Mn 3 (Zn, M ) x N( M = Ag, Ge)

Abstract: Neutron diffraction is used to reveal the origin and control of the thermal expansion properties of the cubic intermetallic compounds Mn${}_{3}$Zn${}_{x}$N and Mn${}_{3}$[Zn-(Ag,Ge)]${}_{x}$N. We show that the introduction of Zn vacancies induces and stabilizes an antiferromagnetic phase with huge spin-lattice coupling that can be tuned to achieve zero thermal expansion (ZTE) over a wide temperature range. We further show that the antiferromagnetic ordering temperature (${T}_{N}$) that controls this ZTE can be tuned by chemical substitution, again on the Zn site, to adjust the span of ZTE temperatures from well above room temperature to well below. This establishes a quantitative relationship and mechanism to precisely control the ZTE of a single material, enabling it to be tailored for specific device applications.

Online Proactive Caching in Mobile Edge Computing Using Bidirectional Deep Recurrent Neural Network

Abstract: With emergence of Internet of Things (IoT), wireless traffic has grown dramatically, posing severe strain on core network and backhaul bandwidth. Proactive caching in mobile edge computing systems can not only efficiently mitigate the traffic congestion and relieve burden of backhaul but also can reduce the service latency for end devices. However, proactive caching heavily relies on the prediction accuracy of content popularity, which is typically unknown and change over time. In this paper, we propose an online proactive caching scheme based on bidirectional deep recurrent neural network (BRNN) model to predict time-series content requests and update edge caching accordingly. Specifically, on the first layer, a 1-D convolution neural network (CNN) is devised to reduce the computational costs. Then, BRNN is employed to predict time-varying requests from users. Afterward, a fully connected neural network (FCNN) is harnessed to learn and sample predicts from the BRNN. Finally, we conduct experiments based on real datasets, which demonstrate that the proposed approach can achieve considerably high prediction accuracy and significantly improve content hit rate of end devices.

Cross-Layer Design for TCP Performance Improvement in Cognitive Radio Networks

Abstract: In cognitive radio (CR) networks, the end-to-end transmission-control protocol (TCP) performance experienced by secondary users is a very important factor that evaluates the secondary user perceived quality of service (QoS). Most previous works in CR networks ignore the TCP performance. In this paper, we take a cross-layer design approach to jointly consider the spectrum sensing, access decision, physical-layer modulation and coding scheme, and data-link layer frame size in CR networks to maximize the TCP throughput in CR networks. The wireless channel and the primary network usage are modeled as a finite-state Markov process. Due to the miss detection and the estimation error experienced by secondary users, the system state cannot be directly observed. Consequently, we formulate the cross-layer TCP throughput optimization problem as a partially observable Markov decision process (POMDP). Simulation results show that the design parameters in CR networks have a significant impact on the TCP throughput, and the TCP throughput can be substantially improved if the low-layer parameters in CR networks are optimized jointly.

Deep Q-Learning Aided Networking, Caching, and Computing Resources Allocation in Software-Defined Satellite-Terrestrial Networks

Abstract: With the development of satellite networks, there is an emerging trend to integrate satellite networks with terrestrial networks, called satellite-terrestrial networks (STNs). The improvements of STNs need innovative information and communication technologies, such as networking, caching, and computing. In this paper, we propose a software-defined STN to manage and orchestrate networking, caching, and computing resources jointly. We formulate the joint resources allocation problem as a joint optimization problem, and use a deep Q-learning approach to solve it. Simulation results show the effectiveness of our proposed scheme.