Session IT-01

Invited Talk 1

Conference
1:30 PM — 3:00 PM CST
Local
Aug 10 Mon, 1:30 AM — 3:00 AM EDT

A Comprehensive Optical Mobile Fronthaul Access Network

Weisheng Hu (Shanghai Jiao Tong University, China)

1
Both cloud radio access network (C-RAN) and next generation fronthaul interface (NGFI) are the key solutions for the 5G deployment. In both architectures, the baseband units (BBUs) are centralized, and remote radio units (RRUs) are separately allocated, where the BBUs and RRUs are connected through a fronthaul network with Common Public Radio Interface (CPRI) and evolved eCPRI. In this work, we proposed a comprehensive optical mobile fronthaul access network (COMFAN) to meet the various fronthaul requirements. To support both the CPRI and eCPRI interfaces, several low-cost high bitrate optical transmission techniques are comparatively studied.

Toward a Trustworthy and Evolvable Future Internet

Hongbin Luo (Beihang University, China)

0
Although the Internet has made great success since its inception, it faces many serious issues such as the lack of trustworthiness, the rigidity in deploying novel technologies at layer 3, as evidenced by the proliferation of various cyberattacks and the difficulty in deploying IPv6. These issues makes it extremely difficult to further expand the Internet to satellite networks, industrial networks and vehicular networks because, as widely recognized, IP does not perform well in these network environments. In this talk, we present the core ideas of an architecture for a Trustworthy and Evolvable Future Internet.

Resource Orchestration of Optical Networks with Multi-Access Edge Computing

Shanguo Huang (Beijing University of Posts and Telecommunications, China)

0
With the advent of the 5G, the traffic pressure on the bearer network is increasing. Meanwhile, the rapid development and large-scale application of IoT devices have brought about low-latency, high-reliability information processing and transmission requirements. Multi-access Edge Computing (MEC) introduced by sinking cloud resources from the Remote Cloud to the edge of the network is one of the solutions to support 5G low-latency applications. Optical networks with MEC is considered a promising candidate to meet the demanding bandwidth and latency requirements of future communications. At present, for optical networks with MEC, a key issue is how to provide services with lower latency and higher efficiency for end-users. Based on this, we investigate the resource orchestration and benefits of optical networks with MEC. This presentation first introduces the basic principle and characteristics of the optical network and Multi-access Edge Computing, then several resource orchestration schemes are explained in detail, and the simulation results are discussed at the end. The results show that the proposed schemes can effectively improve the resource utilization of the system while reducing user latency.

Session Chair

Xiaoxue Gong

Session IT-02

Invited Talk 2

Conference
3:10 PM — 4:40 PM CST
Local
Aug 10 Mon, 3:10 AM — 4:40 AM EDT

Optimal Scheduling of Mobile Edge Computing for Space Information Networks

Qinyu Zhang (Harbin Institute of Technology, Shenzhen China.)

1
Mobile Edge Computing (MEC) is a promising solution to tackle the upcoming computing tsunami in space information network (SIN), by effectively utilizing the idle resource at the edge. In this work, we study such a multi-hop D2D-enabled MEC scenario for SIN, where mobile devices at network edge connect and share resources with each other via multi-hop D2D. We focus on the micro-task scheduling problem in the multi-hop D2D-enabled MEC system, where each task is divided into multiple sequential micro-tasks, such as data downloading micro-task, data processing micro-task, and data uploading micro-task, according to their functionalities as well as resource requirements. We propose a joint Task Failure Probability and Energy Consumption Minimization problem (called TFP-ECM), which aims to minimize the task failure probability and the energy consumption jointly. To solve the problem, we propose several linearization methods to relax the constraints, and convert the original problem into an integer linear programming (ILP). Simulation results show that our proposed solution outperforms the existing solutions (with indivisible tasks or without resource sharing) in terms of both the total cost and the task failure probability.

Key Technologies of Full-Band Optical Transmission Systems and Networking

Gangxiang Shen (Soochow University, China)

1
The transmission technology based on the traditional C-band standard single-mode fiber (SSMF) has approached its transmission capacity limit. However, the remaining capacity of an SSMFЎЇs low-loss window is still abundant, up to 400 nm. To explore this potential capacity, this talk will introduce the key technical aspects that enable the full utilization of this full-band. The related technologies on transmission systems and networks are discussed.

Neural Network-based equalizer for intensity modulation and direct detection systems

Lilin Yi (Shanghai Jiao Tong University, China)

1
The neural network (NN) has been widely used as a promising technique in fiber optical communication owing to its powerful learning capabilities. Specifically, the NN-based equalizer is qualified to mitigate mixed linear and nonlinear impairments, providing better performance than traditional algorithms, especially in intensity modulation and direct detection (IMDD) systems. Many demonstrations employ a traditional pseudo-random bit sequence (PRBS) as the test data. However, it has been revealed that the NN can learn the generation rules of the PRBS during training, resulting in abnormally high performance. So it is important to distinguish whether data features are learned by an NN model, what type of dataset can be used to avoid the above problem. After solving the data training issue, optimizing the NN structure to improve the equalization performance without improving the complexity becomes an important objective. In this talk, we analyze the detailed learning process when an NN is trained using a PRBS and determine the effect of the detection of generation rules. We then provide a mutual verification strategy to verify the training effectiveness and propose a combination strategy to construct a strong random sequence that will not be learned by the NN or other advanced algorithms.

Session Chair

Xu Zhang

Made with in Toronto · Privacy Policy · © 2022 Duetone Corp.