更快更稳！宁诺学者将5G环境下收发信息平均延迟时间缩短27％

Please scroll down for the English version-UNNC researcher tackles 5G connectivity challenge

近日，宁波诺丁汉大学的移动通信专家Chiew-Foong Kwong（关秋峰）博士开发了一种智能切换决策方案，用于缩短5G网络中的延迟时间。这项研究得到了宁波市自然科学基金的支持，现已正式结题。

切换是指移动设备在移动过程中切换与基站的连接以实现不间断数据传输的过程。

根据验证测试结果，关博士的算法被证明能够将切换失败率降低75%，并将不必要的切换数量减少90%以上。更为显著的是，收发信息的平均延迟时间最终减少了27%。目前，他的这项研究成果已经刊发在多家核心期刊上。

据关秋峰博士介绍，5G网络的目标包括为各种不同的应用场景提供超高的连接数密度，极高的数据传输率，卓越的用户体验和毫秒级的端到端时延。其中，最难突破和解决的技术难题就是毫秒级时延，而减少时延的关键在于提高切换的速度及准确性。目前全球范围内的专家学者们都在研发不同的切换算法，使得发送和接收信息之间的平均延迟时间最小化。

“想象一下，汽车以60km/h的平均速度在城市中运行。在这样快速变化的5G车联网的应用场景中，如果切换发生得过早，车辆可能更接近原来的基站，且尚未进入下一基站的最佳覆盖范围，导致在两个基站间频繁切换使得信号不稳定；而如果执行过晚，那么车辆很可能因距离上一个基站过远，完全失去信号。解决这个问题的方法之一是根据需求，推测出即将到来的切换时间和目标，从而在正确的时间执行切换。”

为此，关博士和他的团队成员们提出了一种基于机器学习技术，拥有自适应和预测功能的切换管理算法。这种算法可以产生一系列自动决策，包括是否触发切换机制，切换到哪个基站以及选择正确的切换时机。它还可以准确预测用户的移动轨迹，从而进一步缩短延迟。

关博士告诉我们，像物联网、车联网、外科手术等应用场景，对切换的必要性和切换的时间都提出了极高的要求。高品质的切换非常关键。而他们的算法被证明较现有技术减少不必要的切换达90%以上，同时，收发信息的平均延迟时间减少了27%，但数据处理通过量却增加了9.7%。

目前团队正着眼于将这项研究提升到更高的水平，包括为这种算法申请专利和开发更为节能的切换机制。关秋峰博士还希望与其他研究人员和行业密切合作，将研究成果从实验室应用到各个领域，特别是5G环境下需要极高可靠性应用场景，如自动驾驶技术。

UNNC researcher tackles 5G connectivity challenge

Dr Chiew-Foong Kwong, a mobile communications expert from the University of Nottingham Ningbo China (UNNC), recently developed an intelligent handover decision scheme for 5G network, offering a promising solution to fix its connectivity challenge.

The breakthrough technology of 5G depicts a future where every device is connected and can talk to one another in a seamless fashion. One of the goals of 5G is to minimise the average delay time between the sending and receiving of information, also known as latency, down to less than one millisecond. However according to Dr Kwong, one of the bottleneck problems in reaching the latency goal is to have an efficient handover algorithm.

A handover is a process that occurs when a mobile device switches its connection during movement to achieve uninterrupted data transmission. An ill-designed handover algorithm would adversely impact connectivity and cause increased latency.

“The conventional handover decision mechanism designed during the 4G era doesn’t work very well for 5G.” Dr Kwong said. “It does not sufficiently address the problem of frequent and unnecessary handovers, which affects not only the quality of service, but also increases latency.”

Together with his PhD students Qianyu Liu and Lincan Li, the solution Dr Kwong proposes is an adaptive and predictive handover management algorithm based on machine learning techniques. The algorithm can produce a sequence of enhanced decisions covering whether to trigger a handover, which base station to hand off to, and subsequently the correct timing for handoff. The algorithm also can accurately predict user movement to further shortens latency.

During validation testing, Dr Kwong’s algorithm proved to be able to reduce the handover failure rate by 75% and minimise the number of unnecessary handovers by over 90%. As a result, the latency is reduced by 27%. His work has led to multiple publications on the world’s leading journals indexed by the Science Citation Index.

Dr Kwong is now looking at taking his research to the next level, which includes patenting the algorithm and developing an energy-efficient handover mechanism. He is also closely working with other researchers and industries to bring the research from lab to land, especially to improve the connectivity performance of safety-critical applications such as autonomous driving.

Dr Chiew-Foong Kwong is an Assistant Professor in Electrical & Electronic Engineering at UNNC. His work is supported by Ningbo Natural Science Foundation.