The unique initiatives of FPT students for graduation thesis

29

Ceremony to present graduation thesis is an important “frontier” of student life. The graduation ceremony of Summer 2019 has gathered many interesting topics with high applicability.

In the group of Information Technology, students quickly grasp the outstanding and highly applicable technologies of the 4.0 technology revolution such as IoT, AI, Blockchain to put into their projects. Let’s take a look at some interesting projects presented by the students at the 2019 Graduation thesis project.

Tools to help businesses detect security holes

5 Information Security students conquered the Review Council with the product “Solution to manage security holes for enterprises”.

Stemming from the increasing number of cyber-attacks in Vietnam and around the world, causing a lot of damage to users, the student group consists of Nguyen Anh Viet (leader), Tran Anh Duc, and Le Dinh Manh, Nguyen Khac Hung, and Nguyen Duc Anh, under the guidance of lecturer Ha Bach Nam, proposed the idea of ​​building a system to manage security holes.

Members make a product presentation.

In fact, the market has many solutions to this problem such as hiring security solutions from third parties, developing own privacy policies or hiring security services to periodically monitor security holes, etc. This solution has encountered the problems of cost, operating manpower,… causing many difficulties for customers, especially for corporate customers.

To overcome that limitation, the FPT University student group has built a platform that is capable of controlling 24/7 security holes, durable and is easy to use. At the same time, it also has the ability to update regularly and expand the number of nodes to increase the performance of the entire system easily.

In addition, the team also built a Web portal interface that makes it easy for users to monitor and manage vulnerabilities that are scanned by tools in the system.

New to the solution is that the tools will be divided into stages and the inputs of one tool will depend on the output of another. More realistic running flow helps the tool get the correct input to run out an accurate output.

Due to being a student research group, the members encountered many difficulties in the process of implementing the project because of limited knowledge and skills. To overcome this problem, you actively assigned research areas to each member. After that, the whole group got together and talked to all to understand the problem. The group also actively started to work on a project early to timely improve its products.

Before coming to the graduation project, the team brought a test product on FIS Internal system and discovered 43 security holes, FPT Public detected 150 vulnerabilities and this number in CMC Public was 14.

Evaluating the group’s project, the Review Council said that this is an innovative solution with practical significance but also needs further improvement to really work effectively when put into use.

Doremon Vietnamese version

Spending many days, 4 members, Phan Minh Duong, Hoang Van Thang, Truong Minh Giang, and Le Hoang Nam, went to different regions of Vietnam to collect Vietnamese voice data for the mechanical cat. Doraemon – Products that control smart homes with Vietnamese scripts.

This is a product that helps users control their homes with Vietnamese voice with 15 scripts such as: turning on/off lights, closing/opening doors, turning on/off the TV, turning on/off fans… This product will help replace English voice control devices that are now available. The product consists of 2 parts: the main controller and the device to control the appliances in the house. The main controller recognizes the voice and issues control commands to the control devices. After the receiving device will execute the issued command, for example when saying “Turn on the light“, the main controller will recognize the word “Turn on the light” and issue the “Turn on the light” to the device.

Doraemon products of the group.

In order to help Doraemon recognize group voices, he has used Deep Learning technology to create voice recognition models with up to 98% accuracy. To get this accurate result, the team gathered data from more than 300 people from the North – Central – South regions across the country.

Sharing about the highlight of the product, Phan Minh Duong – a member of the group said that with the use of Deep Learning, the model was created very light, suitable for today’s embedded computers. Thanks to that, the voice will be recognized right on the main unit, helping to give a stable response speed. In particular, the products of the group can recognize voices in an environment without Internet connection other than products on the market today such as Google Home,… Besides, the product does not need to send voice data on the server to identify to help increase the privacy of users, not afraid of eavesdropping and ensuring the safety of the home.

Demo of product features in front of the Review Council at the project defense.

Like other student groups, the process of creating a group product encountered many difficulties. Perhaps the most difficult is the process of going through the Vietnamese S-band to collect Vietnamese voice data. The members of the group are all in the North, so it is not easy to collect voices, but with the goal that any Vietnamese people can use the product easily, the group has sent each other to the Central and Southern regions. This is a big challenge for each member to move on his own, to meet each person to collect voice. After many days, the group has collected the voices of more than 300 people of different ages and genders across the three regions of the country.

To increase the commercialization of the product, the group will add and improve some features such as increasing recognition accuracy, increasing the number of scripts to control more devices. The team hopes that Doraemon will become a user-friendly virtual assistant in the era of technology 4.0.

Surely everyone knows the mechanical cat from the future of Doraemon in Japan with countless talents. Based on this image, the team wanted to create the Vietnamese version of Doraemon cat with Vietnamese users. This will be a friendly machine cat, helping people control the devices in their house by voice. In addition, Doraemon shape will create a sense of friendliness and closer to users”, Minh Duong added.

It is known that Doraemon cat controlling smart home with Vietnamese script helped Duong group win first prize in IoT Showcase Contest organized by FPT Edu in Can Tho recently. In addition, the group’s research was published in the international journal Scopus and received awards from the University.

Shocking warning application

Student group Vu Van Cuong, Le Hong Dung, Lam Hai Vu and Nguyen Duy Tri (FPT Education) have developed a shocking warning solution for drivers to minimize traffic accidents. The app, called Smart Traffic, works smoothly on the Android platform and is highly commercially available.

According to statistics, every 17 minutes, England spent 432 pounds to overcome the consequences caused by bad roads. In 2011, 20 million potholes were recorded throughout European routes. In Vietnam, the road situation is not flat, making it difficult to travel and has been reflected in the mass media.

Due to this fact, 4 FPT Education students proposed solutions to identify potholes and elephant pockets on the road and give timely warning to the driver to avoid accidents or damage to the car. The solution consists of 2 Android applications. An application that collects the position of the bumpers on the road via accelerometer and aggregates them into a network/map. The other app gives voice alerts or messages to drivers when approaching the bump.

Members of the group give presentations on how to use the application.

The group shared that in fact, there were many solutions to warn the bump on the road, but the limitation of this product is that it is expensive and operating equipment is difficult to develop in developing countries where traffic infrastructure is limited.

Overcoming these problems, the Smart Traffic app has an eye-catching, easy-to-use, and extremely user-friendly interface. Users have just received alerts to be more proactive when traveling on the road and can participate in contributing information to the network through simple operations on smartphones. Besides, the application also integrates directions feature or gives advice on less bumpy routes.

When tested on Thang Long Boulevard (Thach That District, Hanoi), the application has detected 40 shock points with almost absolute accuracy. Every 200 – 300 meters from each bump, the application automatically emits a warning to drive at a reduced speed or to steer to reduce collisions.

The operation interface of the application in practice.

Witnessing the product running a stable demo, intelligent algorithms and user-friendly application interface, Mr. Hoang Xuan Son, lecturer of FPT Education, said that the product has the potential for commercialization: “I see your application very good, has practical implications. I have also asked the question why such a useful function does not appear in big applications such as Google Maps. Therefore, if we can improve the features and pack the product completely, it is clear that the prospect for the product is very bright.”

According to FPT Education students, another way of commercializing the product is to sell it to car manufacturers so that they can integrate this function into the vehicle’s sensor system. “The group will take the comments of the Review Council to continue to improve the features, and create more versions of the app on iOS so that more people can use and contribute information to the system”- Duy Tri said.

FPT Education

Related posts: