DSP Challenge Cash
Prize
DSP Challenge Cash
PrizeA fascination with the future of video communications led three students from the University of Maryland in College Park to the final round of Texas Instruments US $100,000 DSP Solutions Challenge.
The TI DSP Solutions Challenge team from the University of Maryland used digital signal processors (DSPs) to design and implement a system that compresses the large volume of data needed for the representation of video signals, making it possible to transmit video signals over narrowband communication channels, such as telephone lines.
There are many real-world uses for such a simple video compression system. A simple application is two-way video communication, such as video telephony or video teleconferencing on the desktop. In the home, family members could see each other even if they are separated by vast geographical distances. In offices, people could hold videoconferences with colleagues in remote areas over their PCs. Other applications would be in remote access to video archives or in telemedicine situations where this system could enable doctors who are not on-site to monitor the process of a surgery remotely, or in the field, allowing a doctor at the hospital to guide paramedics treating an accident victim.
Because video frames are data intensive, advanced video compression systems require a tremendous amount of number crunching at extremely high speeds to keep pace with the amount of data being processed. In short, video compression/decompression requires the capability to perform millions of additions and multiplications per second, a capability that today's general-purpose computers do not have. Although video compression systems are currently implemented with custom-designed VLSI technology, each application is unique and the corresponding system is time-consuming and expensive to design. Furthermore, the custom-designed hardware is algorithm specific and cannot be re-programmed for new video compression algorithms and standards.
The system developed by the Maryland team uses DSPs to provide the fast processing of large quantities of data and to implement a real-time video compression system that can be programmed to handle a variety of algorithms and applications. DSPs provide the fast computations such real-time algorithms require. In addition, they are relatively inexpensive and readily available.
Jerome Johnson (20) is a university sophomore majoring in electrical engineering and computer science who comes from College Park, Maryland. His father and his brother, Brian, are both engineers. "I was fortunate," says Jerome, "in a house of engineers, I got exposed to a lot of problem solving and I was expected to figure things out on my own. Today, I see so many smart students who don't do well just because they have not had the training to solve problems. That's one thing they don't really teach enough in school. So much of learning in schools is just straight memorization. This contest is a way for students to use the resources they have to design projects on their own."
Jerome's brother describes him as "a hard-working, engineering, computer science maniac," and those traits served him well on the DSP Challenge team. For the Challenge, Jerome was given the task of understanding the DSP board and implementing the algorithm on the board. According to Professor Farvardim, "Jerome was the main force behind the implementation of this team's algorithm on the DSP board. He is only 20 years old, and a sophomore, but the work he has done is easily at the level of a master's thesis." What does Jerome think of the DSPs? "Well, I started out knowing nothing and six months later I had the system running on the DSP board," he says. "I guess it isn't too complicated if you can sit down, read the manuals and come up with a working video system."
Ruplu Bhattacharya (24) is a first year graduate student-he was a senior while working on the project-in electrical engineering at the University of Maryland who hopes to stay at Maryland to complete a Ph.D. Ruplu's father is with the World Bank and the family has lived in India and the United States, giving Ruplu insight into the differences and similarities in the two educational systems. How do India and the United States compare in education? According to Ruplu, "Despite the criticisms of American education, the biggest difference is that the United States has so many universities. That gives individuals more choice in deciding what they want to do and where they want to do it. In India, there's intense competition to go to the few universities and programs available, so a majority of the students are working really hard to succeed. Here, the atmosphere is more relaxed, but you find people who are equally bright and accomplished in both countries."
Ruplu comes from a family of engineers and has always wanted to be an engineer. Many of his cousins have come to the United States for their advanced degrees, and he was impressed by the opportunities they had for research. How did he get interested in videoconferencing? "To be honest," he said, "it came from the movies. In many of today's films, there's often a scene in which someone has a videoconference with another person." What did Ruplu do for the DSP Challenge team? "I helped speed up the implementation of the software that Hamid developed so that we could run it fast on the DSP board. I really enjoyed working with the DSP board. It's a great platform, which can be used for all kinds of neat applications."
In his spare time, Ruplu enjoys collecting software and taking it apart to make it run faster. He says, "There are two kinds of software: software that runs, and software that runs faster." He's also a big fan of basketball star Michael Jordan.
Hamid Jafarkani (30) is a Ph.D. candidate in electrical engineering who came to the United States to study at the University of Maryland after completing his undergraduate education in his native country of Iran. Hamid says, "I got interested in engineering when I was in high school. I liked mathematics and physics, and I wanted to know how to apply these sciences to solve problems in the real world." Hamid has been working on image coding since he began his master's thesis, and on the specific algorithm that the Maryland team's video compression system employs since the summer of 1995. When asked about the value of a project like the DSP Solutions Challenge, Hamid replied, "It's becoming impossible to work alone on such advanced projects, so we need to learn how to work with each other and how to give and take ideas in a team environment."
Hamid came to the United States because of the "research facilities that make it very easy to do advanced research." He likes the challenge of working with advanced technologies because "there's always something new being done, and that means you have to continuously update and improve your work." Hamid relaxes by reading literature, playing soccer and table tennis, and he is a member of one of the university's intramural soccer teams.
Nariman Farvardin is Faculty Adviser to the Maryland team. Farvardin earned a doctorate in Electrical Engineering from Rensselaer Polytechnic Institute in Troy, NY. Since 1984, he has been on the faculty of electrical engineering at the University of Maryland, where he is currently a professor. Since May, 1995, he has served as the Chair of the Department of Electrical Engineering. He has previous experience with the Ecole Nationale Superieure des Telecommunications (Paris, France), with AT&T Bell Laboratories, and with the National Institute of Standards and Technology in the United States. Farvardin has published many papers on image coding in IEEE journals and proceedings. In addition, he holds a patent (jointly with S. Tretter and R. Laroia) for a precoding scheme that has been used in the standard for 28.8 kb/s modems. Farvardin received the Presidential Young Investigator Award in 1987.
Reader Inquiry: 1-800-477-8924, ext. 5300
Please refer to Profile #: SPR210
Search the Semiconductor News Release Archives
TI Home
Search
