Research Associate
School of Data Science
City University of Hong Kong
Email: Qiulin.Lin at cityu.edu.hk
Qiulin Lin is currently a Research Associate at School of Data Science, City University of Hong Kong. He received the Ph.D. degree in Information Engineering from The Chinese University of Hong Kong (CUHK) in 2020, advised by Prof. Minghua Chen. He received B.Eng. degree in Electronic Engineering and Information Science from University of Science and Technology of China (USTC) in 2016.
My current research interests include online optimization and algorithms with applications in AI/ML systems and cyber-physical systems, smart power grid optimization (e.g., energy storage system management and EV charging), and intelligent transportation systems (e.g. ride-sharing and heavy-duty truck routing).
Online optimization handles optimization problems with sequentially revealed information. One has to make online decisions based on the revealed information only, whereas the classical optimization assumes complete information. We propose a novel and general framework CR-Pursuit [SIGMETRICS'19, SIGMETRICS'22] for online optimization under inventory (budget) constraints and show it achieves close performance to the one with complete information. We have applied our approach to various scenarios, including EV charging [INFOCOM'19, T-ITS'21] and peak-demand reduction [INFOCOM'21 (Poster), ACM e-Energy'21]. Our approach also raises the interests of other researchers and is shown useful in cloud resource provisioning, virtual network functions development, etc. We are interested to see broader applications of our approach and online optimization.
Modern ride-sharing systems like Didi Pingche and Uberpool offer an economical and eco-friendly trip mode. Optimizing a ride-sharing system is challenging as it needs to handle tremendous dynamic travel demands in real-time. Meanwhile, the massive amount of data it provides facilitates predicting future demand information. We are interested to see how to apply the predictive demand information to improve the performance of ride-sharing systems. We begin by proposing a new routing paradigm incorporating future demand statistics to optimize the probability and benefit of forming a ride-sharing group for every trip [INFOCOM'18]. We then jointly optimize order dispatching and vehicle routing at system level by taking future demand as a probabilistic resource that can be allocated among drivers [Mobihoc'19]. We show that our approach substantially improves the ride-sharing trip formation and the total profit. Our probabilistic approach is general and can go beyond the ride-sharing scenario to broader application domains.