Page 62 - Undergraduate Research Opportunities Program Proceedings 2017-18

School of Engineering
Department of Computer Science & Engineering

Department of Computer Science & Engineering

Congestion Control for Data Center Networks

Supervisor: BENSAOU BRAHIM / CSE
Student: AGGARWAL Ashish Kumar / COMP Course: UROP1100, Spring

This is a project done as a part of Undergraduate Research Opportunities Program 1100 under the supervision of
Professor Brahim Bensaou on the topic of Datacenter Networking. The project focused on mitigating congestion in
datacenter networks especially incast congestion generated due to small, bursty datacenter traffic. After reading various
research papers and tools to understand the current techniques used and proposed by other researchers to mitigate
datacenter congestion, I gained an understanding of the current state of art solutions as well as the causes of latency in
datacenter network. Then I worked to implement a simple solution to using DPDK library to reduce virtualization latency.
The project will be continued in the summer 2017-18.

AI meets Big Data: User Analytics and Personalized Recommendation Based on Location Data

Supervisor: CHAN Gary Shueng Han / CSE
Student: CHAO Wen-shuo / SENG Course: UROP1100, Summer

There are plenty of parking lots face the issue of inefficient entrance methods, which use the parking ticket. In order to
solve this problem, we propose the idea of auto-payment through image recognition. With license Plate recognition,
people could easily pay the fee without complicated procedure. Moreover, it helps the user to locate their car more
conveniently. To achieve the target, we try to use the machine learning method. Not only does it provide accurate
results, it is able work efficiently. We have tried Haar cascade training and analyzed the performance. Although there
are still lots of things could improve, we have gained the basic structure already.

Supervisor: CHAN Gary Shueng Han / CSE
Student: ZHOU Yeju / COMP Course: UROP1100, Summer

The objective of the project is to detect multiple HK license plates at a time from a video frame and recognize the license
number on the plates. The operation contains mainly two procedures: First, detect the license plates and allocate the
coordinates of the plates in the given image; Second, extract the license plates (and segment the texts if necessary) and
recognize the texts. Two methods both based on OpenCV library did I try to implement the first procedure: One is to
operate pre-processing, color extracting, morphological changes and edge detection on the given image, which is
considered as the “traditional method”; the other is to use Cascade Classifier to train a classifier by means of AdaBoost
(a means of machine learning). This report will only discuss about the first procedure and the results and challenges of
the two methods respectively.

AI meets Big Data: Analytics and Recommendation for User Location Data

Supervisor: CHAN Gary Shueng Han / CSE
Student: BAE Juyoung / COMP Course: UROP1100, Fall

GPS has been used in wide range of application in outdoor localization since it was first introduced. In the same context,
localization in indoor environments has attracted growing interest in the past decades due to its promising application

52