Image and video Authentication for P2P application

The study of information hiding aims to embed data imperceptibly into various types of multimedia data. Because information hiding can be used for many critical applications like covert communication, copyright protection and data authentication, information hiding has become one of the major topics in the fields of multimedia and security research. In this sub-project, we intend to design new information hiding techniques with active characteristics to protect network information security.

In the past few years, most of the research works on information hiding have been concentrated on passive information hiding, which is a complement of active information hiding. Passive information hiding aims to hide passive data that cannot perform any task actively by itself. However, active information hiding tries to hide active data into certain cover multimedia data to perform specific tasks for various security applications. Such active data may be an applet or an executable program, called an active agent. For example, assume that an active agent is imperceptibly embedded in a video. When the video is downloaded by a user from the Internet, the corresponding media player will invoke the active agent during the first request by the user. The user would then be asked for a proper key. When the user fails to provide a valid key, the agent may simply delete the video from the user storage. Otherwise, the user will be allowed to preview the video.

In addition to the basic requirements for information hiding techniques like imperceptibility of embedded data, active information hiding requires high data hiding capacity and low probability of extraction errors. These two additional requirements pose the challenge of designing active information hiding techniques. An active information hiding technique must have the ability to embed not only secret data, but also the designated active agent. Accordingly, the former requirement of active information hiding is easy to understand because the size of an applet or an executable file is at least several hundred bytes. Generally, if an active agent is designed to perform several tasks, its size will be larger than the agent that performs only one simple task. On the other hand, due to the sensitivity of errors in executable files, it is required that active information hiding techniques are with low probability of extraction errors.

Due to an active agent can be designed to perform several tasks, we will also design active agents to perform different tasks like data authentication, copyright protection, and covert communication (steganography).

The results of our researches are summarized as follows:

1. Propose new active information hiding techniques for different types of multimedia.

2. Design active agents to perform different tasks for security protection.

3. Image and video authentication for P2P application.

Security knowledge warehousing for P2P intrusion detection and prevention

Security knowledge including attack features and detection methods are evolutional in a rapidly changing network environment. Main research tasks about application-level network security include: (i) dynamic security knowledge acquisition method; (ii) security knowledge warehousing; (iii) knowledge aggregation and generalization techniques; (iv) worm immune service; and (v) the improvement of misuse intrusion detector's power and efficiency.

P2P gateway systems for security protection

We detect a P2P flow and packets sent from a P2P connection, such as eDonkey, eMule, Kademlia, KaZaA, BT, and Gnutella SoulSeek, by using content-based algorithm. The test system is now implemented in Intel IXDP425 embedded system.