Q. As we know, you worked in the United States for many years. What brought you back to teach in NCTU?
I worked at Telcordia Technologies, formerly Bell Communications Research, Inc. (or Bellcore), which was transformed from AT&T Bell Labs. This Company was founded in 1983, and we take pride in Bell Labs’ heritage of valuing research and invention. However, when Telcordia was acquired and merged by Ericsson in 2011, I started to rethink the direction of my future career. When pondering over the career change, I came to a decision that teaching was what I had been longing to do. While I worked in the US, I kept a close connection with NCTU and was invited back to my old department several times for forums and visits. When the teaching opportunity presented itself, I seized it and returned to teach at my alma mater.
I taught at National Taiwan University of Science and Technology for two years before going to the US. In fact, that teaching experience was what motivated me to study in the US. My original purpose of studying aboard was so that I could acquire a teaching position in Taiwan. I never expected that I would eventually spend nearly 30 years in the US. However, I never forgot my dream, which was to come back to my homeland, to teach in a university, to carry out research, and to work and interact with students.
Q. You have been studying networks for many years and dedicated your efforts to promoting the research and technology on M2M. Can you introduce the content of M2M and the status of current research?
The Chinese term is a translation from English. Actually there are two terms in English: one is Internet of Things (IoT) and the other is Machine-to-Machine (M2M). However, it is impossible to understand what M2M is without going through the changes throughout the entire communication networks.
In the early days, the communication networks were mainly vocal, such as telephone networks. Each family in principle had a telephone line and a telephone number; having two lines for one person was impossible in the fixed network. However, during a short period when the fax was popular, one family might have two telephone numbers, one for the vocal communication and the other for the fax machine. Now that period has passed; fax machines became less important when their function was replaced by emails. The demand for the second telephone line has gradually diminished.
Therefore, once again the two numbers have been reduced to one. I myself have experienced the same process. The next wave of telecom networks involves wireless networks, revolutionizing the concept of one line for one household. In this network structure, each person has a number; that is, in a family of five members, there will be five mobile phones and five numbers. Since everyone has a number, the home telephone may become obsolete. Currently, many young people do not apply for any fixed line when they have a new family; instead, they use a mobile phone or Internet phone such as Skype.
As mentioned above, the communication capacity increased about 10-fold when transitioning from fixed lines and networks to wireless networks and mobile phones. M2M will be the next trend; the capacity will increase an additional 10-fold.
Traditional telecom networks mainly connect people, while M2M networks will expand the subjects of the connection. In the past, people were the only users of network terminals; now machines can also be the users. As a result, the traditional network design may be unable to meet the substantial increase in the demand for M2M networks. As an analogy, normal vocal communication takes about 3 minutes on average; thus, the basic networks were designed accordingly. However, normal connection time to the Internet far exceeds a mere 3 minutes even for simple web browsing. When the demand and flow of the whole communication networks differ, the structure of the network has to change accordingly.
Networks for human users are designed based on human's communication models. To design networks for machines, we need to know the differences between how human and machines do communications. Compared to human, machines exhibit much greater diversity and complexity. Some "machines" have very simple communication models; an example is room temperature sensors whose basic function is to read current temperature every 10 minutes and report the data via the Internet. However, while some machines are very simple, with communication models simpler than human's, many machines are much more complex and there is very large diversity among them. A network design is created according to communication models. To cope with the complexity of these communication applications, the network design needs to consider many factors and should be based on long-term observations and statistics.
But till now, as diverse machines, equipments and sensors are being connected to the Internet, we are yet to know what new model will arise in M2M networks. It’s just the start, and the relevant data are incomplete. We only know that it will impact the networks. The network design in response to the new requirements is still a major issue in current research.
Q. Professor Lin, can you tell us more about the current development of M2M applications?
Principally, M2M networks connect existing machines and sensors. Some technologies have been developed since 10 or 20 years ago. However, until now, the network communication industry has been doing this separately without integration. Many industries that used to have low network needs, such as automotive industry, household appliance manufacturers, hospital managements, and power system operators, have imposed increased demands on networks. Their equipments or back-end computer systems need to be connected to the sales or service systems, while all these machines require access to the Internet. However, as each different industrial sector has its particular requirements, each has developed its own particular way of handling its needs. This results in the bottleneck of the present M2M technologies: The systems developed by different industrial sectors and different applications cannot communicate with each other, even though in fact they all have similar needs, that is to connect the equipments, machines and sensors to the Internet. Therefore, the latest trend is to establish an international standard so that the needs of different industries can be addressed on a common system.
Now the development of M2M standards is well underway. The international M2M standard organization oneM2M was founded in 2012 and started to draw up M2M standards. This is a very important step. With common standards, we can put different industrial systems on the same platform. If this platform succeeds, we can M2M in several generations in a way similar to that of wireless networks, and popularize M2M applications.
As a matter of fact, there are already expanding M2M applications in our lives. For example, conventional TV had no networking functions, while now it can receive cable TV programs, view YouTube clips through the connection to the Internet, or even completely replace computers. While household digital cameras usually require memory cards or cable connection to access photos, some new cameras are equipped with networking functions; after being set up, the photo can be automatically uploaded to Facebook once the user presses the shutter. In fact, these are examples of M2M applications, through which users can skip the complicated operations; the automation is enabled by the convenience of networking capabilities.
The aim of M2M applications is to enable Machine-to-Machine communications, but a shorter-term goal is to allow Machine-to-Human communication, so that machines can provide services to people more actively. For example, if I am interested in cancer issues, I can set up my smart TV to automatically search related programs on the Internet or send email notifications regarding such programs. In sum, Machine-to-Man communication means that machines act more intelligently so that they can judge a user’s needs and provide relevant services.
Machine-to-Machine is also Machine-to-Mobility. An important feature of M2M is the mobility, such as the mobile phones we carry around. During the movements, many data can be collected, transmitted, and then analyzed, leading to very useful results. For example, to monitor a speeding recidivist, the police can install a speed monitoring software on his mobile phone with GPS functions. Whenever the sensor detects the target is driving, it will automatically measure the driving speed and simultaneously report the measured value to the police station. Once the target is speeding, the police station will immediately receive the data, based on which the police can impose a fine or revoke his license. Although such an operation may involve the issue of privacy violation, which is another important topic worth discussing, this is a typical Machine-to-Mobility application, clearly showing that M2M can be practically applied in various aspects of our lives.
Q. Internet of Things (IoT) Credit Program began at the time when you returned to the university. Can you tell us more about this program?
I cannot take credit for establishing the IoT Credit Program, which was already planned by the Department in response to the plan of the Ministry of Education (MOE) before I came back. According to MOE’s plan, NCTU would be the pioneer of this credit program, which was to share relevant resources with other universities after developing some teaching materials and experiments. During the half year before my return, I contacted the professor responsible for this credit program, Chih-Wei Yi, and discussed the need of courses. At that time, I decided to give a course on key technologies for IoT/M2M.
So, as the IoT Credit Program was originally planned by the MOE, I just came back in time and Professor Yi thought me suitable to teach this course as well. For me, this was an appropriate arrangement: on the one hand, it met the MOE’s requirements and, on the other hand, it satisfied my desire to teach this course. Besides, this credit program can coordinate with the project of M2M fundamental technology research center granted by the National Science Council’s Industrial Fundamental Technology Project. With the help of NSC’s funding and research talents, NCTU, as a whole, offers an excellent R&D environment suitable for implementing research projects and promoting this credit program. This credit program can provide students with basic concepts on IoT/M2M, and develop their ability to integrate theories and practical matters.
Q. Finally, can you tell us your expectations for our students in the Department of Computer Science?
Overall, the students at NCTU are excellent. There are always two or three students in each class who can get perfect scores, and the percentage of students pursuing an MS degree is rather high. However, most students pursue an MS degree for very practical considerations, which may generally involve expectations of better-paid jobs. Students rarely aspire to acquire expertise or reach outstanding performance in certain fields, not to mention becoming the pride of Taiwan on the international scale. Perhaps such an expectation is too high. Our department does provide a variety of international exchange opportunities, and the teachers can analyze students’ questions from the global perspective. If our students can be inspired by such information and get motivated, they will be more enthusiastic about learning and exploring this vast realm of knowledge.
In general, Taiwan’s industry is mainly OEM; ODM is gradually starting to grow, while OBM is still quite rare. As the whole industrial structure has had little breakthrough and the profit is limited, the wages of young people are squeezed to a very low level. How can we break through the current situation? The fundamental requirement is to exploit the technologies and to do better than others do. Taiwan’s OEM industry has already shown brilliant performance so that it can attract orders from all over the world. However, our next step should focus on developing higher-level, more creative and more innovative industries. We should enhance our industrial technologies instead of simply marking time because the developing countries are catching up and we are about to lose our competitive edge.
Actually this issue can be linked to the question of many students: Should I pursue my PhD? If a student in a master’s program has learned a great deal and perceives a bright outlook in the field, the notion of advanced study will naturally arise. However, currently this does not seem to be the case, as most students have no motivation to further their studies after a master’s degree. A possible reason is that being PhD students brings little sense of honor. In earlier years, this sense of honor was very strong. For example, in US universities, while master’s and doctoral students attend the same classes; only the best can successfully pass qualification exam and enter a PhD program. Now it seems difficult to have such feelings. The second possible reason may involve worries about future employment after completing a PhD program. I feel that today’s industries in Taiwan are different from those of the past. After my return to Taiwan, I have had contacts with some large-scale enterprises that do value their R&D capacities and thus are interested in hiring PhD researchers. Of course, the wage difference between MS and PhD degrees may not be sufficient to stimulate students’ interests in pursuing an advanced degree. The industries can think about this issue: raising the salary may make a PhD degree more attractive.
I think there are very distinctive differences in the training offered by BS, MS and PhD programs. The BS training involves practice in the problem-solving process since teachers provide both questions and answers; in a MS program, teachers give questions only and students have to find the answers; in a PhD program, teachers provide neither questions nor answers so that students have to look for both. This is the Three-Phase Training mentioned by my advisor, which makes a lot of sense when I look back. PhD researchers should actively seek out possible problems, envision possible solutions and even develop the products before anyone else has even thought of it. Seize the opportunities first! In such a perspective, the training of the PhD phase is extremely important for Taiwan’s industries.
In earlier years in Taiwan, we relied on the contribution of those who had acquired their PhD degrees abroad. Now the universities in Taiwan, with great teachers and abundant resources, have sufficient capacity to train PhD students. Taking NCTU as an example, the excellent teachers and facilities are comparable to outstanding foreign universities. I am not saying that a career in research fits everyone. However, for students who have the qualities and interests for research work, I do recommend that they pursue their PhD degrees, which will equip them with abilities to solve all kinds of problems in the future.