Ma Lin
Ph.D candidate
Department of Computer Science
School of Computing
National University of Singapore

E-mail: malin at comp.nus.edu.sg
Lab: Soc 1, 04-32, Multimedia Lab

I work with Dr. Ooi Wei Tsang as a member of The Networked Multimedia Systems Research Group (NeMeSys).My research focuses on distributed media streaming (multi-source video streaming). Distributed media streaming model streams video data from multiple sources to single receiver. It provides better load balance and requires less uploading bandwidth at the receiver. Distributed media streaming presents many research problems such as, packet assignment among senders and sender selection. My interest is on error recovery, congestion control and distributed content storage optimization, in the context of distributed media streaming.

Curriculum Vitae:

I plan to submit thesis before May, 2007. Currentely I am looking for a job in IT industry. CV is is available for downloading in both MS-Word and PDF format at below:
MS-Word,
PDF.

Education:

2002~
Ph.D candidate, National University of Singapore, Singapore
1998~2002
Bachelor of Computer Science and Engineering, Zhejiang University, Hangzhou, China

Projects:

Jan 2004 ~ April 2005 Error Recovery in Distributed Media Streaming
Thanks to the multiple senders in the distributed media streaming model, error recovery from different senders is possible. In the traditional single channel model, retransmission or forward error correction are carried out only on the channel where packet is lost. Due to the temporal locality (burst nature) of network congestion, it is highly likely that the recovery effort fails. On the other hand, many works have pointed out the merits of exploiting path diversity for better streaming quality [link]. Therefore it is expectable to achieve better result by using the path diversity principle in error recovery. In this work we test different retransmission and forward error correction methods of using the multiple channels. Part of the results are reported in this paper.
A presentation is HERE.
The experimental code is implemented based on Live555.COM library, which is an RTP/RTCP/RTSP/SIP single source streaming media library in C++. We managed to change it into a multi-source streaming program for MP3 audio streaming. The source code is available: HERE, with preliminary instructions and samples of compilation, setup and conducting tests. Please note, the code was implemented and tested on Fedora Core 3.
May 2005 ~ April 2006 Congestion Control for Distributed Media Streaming
The multiple receivers in distributed media streaming pose new challenges in congestion control. As multiple media flows exist in a single session, conventional congestion control gives the session more bandwidth than a normal TCP flow on the bottleneck. Due to this behavior, users are likely to increase the number of flows to enjoy higher quality video. The abuse of multiple flows introduces more congestion, and harms the stability of the Internet. This work proposes the idea of task-level TCP-friendliness, in which we believe TCP-friendliness should be enforced upon a set of flows in a particular task, instead of upon individual flows. We designed a congestion control scheme to achieve task-level TCP-friendliness in distributed media streaming. By detecting congested links between the receiver and the senders, the scheme identifies the corresponding flows experiencing congestion and dynamically adjusts the throughput of those flows such that their combined throughput is TCP-friendly.
A paper is accepted by IEEE INFOCOM 2007.
A presentation (PPT) is available HERE.
This study is simulation based. ns-2 experimental codes is available HERE.
May 2006 ~ Oct 2006 TCP Urel: A TCP Option for Unreliable Data Streaming
In order to maintain stability of the Internet, multimedia flow should be congestion controlled. While existing TCP provides congestion control, it also ensures reliability via persistent Automatic Retransmission reQuest (ARQ). But persistent ARQ is not necessary in multimedia streaming, which tolerates certain level of packet loss. In this paper we present TCP Urel, an option to the existing TCP protocol, for congestion controlled but unreliable data streaming. TCP Urel sends fresh data in every data segment, no matter the segment is originally constructed to deliver fresh data or to retransmit. The design and implementation of TCP Urel is simple, with less than 750 lines of extra code in TCP stack of FreeBSD 5.4. It removes ARQ and meanwhile retains other functionalities including Additive Increase Multiplicative Decrease (AIMD) congestion control. To evaluate TCP Urel, we carry out emulation on a controlled network testbed. Our experiments show that TCP Urel is friendly to TCP Sack, NewReno and Reno. The independence of congestion control from retransmission allows TCP Urel to remain friendly with possible new TCP versions in the future. Our experiments show that TCP Urel is highly efficient in terms of computational cost, and in terms of bandwidth utilization.
A paper is still under submission.
Source code implemented based on FreeBSD5.4 is available HERE. It includes a self-explanatory README file, a TCP Urel implementation, and a Iperf-2.02 based modification for testing.

Publication:

Retransmission in Distributed Media Streaming,
Lin Ma, Wei Tsang Ooi, NOSSDAV 2005, Skamania, WA, USA, 12-14 June 2005. [pdf]
Congestion Control in Distributed Media Streaming,
Lin Ma, Wei Tsang Ooi, IEEE INFOCOM 2007, Anchorage, AK, USA, 6-12 May 2007. (18%) [pdf]

Awards:

Teaching:

I was a teaching assistant on CS3103, Computer Networks II, in semester 2004/2005 1 (Fall) and 2 (Spring). The student’s feedback on teaching is available here and here.

Courses:

CS5215

CONSTRAINT PROGRAMMING

1, 2002/03

B+

CS5228

KNOWLEDGE DISCOVERY IN DATABASES

1, 2002/03

B+

CS5240

THEORY AND PRACTICE OF MULTIMEDIA

1, 2002/03

B-

CS5246

TEXT PROCESSING ON THE WEB

1, 2002/03

B+

CS5233

SIMULATION & MODELLING TECHNIQUES

2, 2002/03

B+

EE5910

MULTIMEDIA NETWORKING

2, 2002/03

A-

Others:

Album (in Chinese)