Theory and technology of multi-agent systems. Software tool for multi-agent systems design. Development of applied multi-agent systems. Multi-agent systems for business logistics. Multi-agent systems for computer network security. Mathematical model of distributed attacks against computer networks. Data and information fusion for situation assessment. Methods and technology for distributed data mining and distributed decision making. Transparent embedding data into digital images for hidden communications and watermarking.

Recent Results

1. New (forth) version of software tool for applied multi-agent system development, MASDK 4.0, is developed.
2. Virtual P2P Emulation Environment for testing and verification of the implemented and ready-to-deploy distributed agent applications. This environment creates on single host virtual overlay networks of communication providers, instances of agent platforms and agents and emulates changes of communication P2P network topology. These changes can be caused by failure or appearance of new nodes or communication channels, by nodes movement in space and so on. Due to the fact, that original agent platform is used (non modified (non “simulation” version)), it is possible to investigate real distributed agents applications as well as their simplified simulation versions under conditions when topology of underlying communication network is dynamic and changes in time.
3. Agent installer, which manages configurations of agents on local host. Agent installer can create configuration files for agent platform, prepare installation packages of multiagent applications, install and uninstall agents on local host.
4. The P2P agent platform constituting an intelligent software infrastructure (middleware) providing transparent interaction of distributed agents of multi-agent systems. The developed P2P agent platform is a pioneering implementation of FIPA NA WG6 reference model. This platform is open, i.e. it supports dynamic population of network nodes, agents set up on top of it and services to be provided by agents). The distributed search services supported by the platform are adaptable to different variants of P2P network topology. P2P agent platform can be ported to many operating systems, e.g. Windows, Linux, and other Unix– like operating systems. The platform functionality is completely decoupled from interfaces that make it possible to develop user interfaces which most completely and advantageously use the means of various operating systems like Windows Mobile, Windows XP, Vista, Fedora Core, etc. The complete package of documentation (user and developer guides, usage cases, scientific literature), as well as run time code and useful links are posted at the Intelligent System Laboratory web site at space.iias.spb.su/ap.
5. New (third) version of software tool for applied multi-agent system development, MASDK 3.0, is developed. This software tool supports the main stages of the applied MAS life cycle of development and maintenance of applied MAS, that are analysis, design, implementation, deployment, maintenance and modification during its practical use. Peculiarities of the developed multi-agent software tool consist in the following: (1) it is built on the basis of the modern methodology called GAIA-methodology; (2) it supports graphical style of the system development; (3) it uses the reusability conception for software components that are standard for many applied MAS; (4) the result of the system formal specification is presented in terms of XML language. this specification is further automatically translated into source code on C++ language and then - in executable code; (5) it is applicable for development of the wide range of applied distributed intelligent systems. This software tool was presented in an International Exhibition of the multi-agent software tools (Erfurt, Germany).Currently this software tool is used for development of applied MAS in the area of detection of intrusions in computer networks and for business processes monitoring and management, in particular, in the area development of new products in electronics. The customers of these works are foreign industrial companies. The main results on this result were published in Russian and international publishers ([5, 6, 9]).
6. Theoretical basis and algorithms for real time situation assessment are developed. In particular, situation formal model is developed. Its main peculiarities are as follows: (1) asynchronous mode of input information presented in form of data streams; (2) Situation input data are heterogeneous and received from many data and information sources; (3) different input data streams have different life time and can contain missing values; (4) situation assessment process is of multi-level character; in this process, fusion of decisions received from many classifiers producing their decision on the basis of different data sources is the most important one, because such decisions are produced at different time; (5) update of the situation assessment is made at every moment when new portion of information is received by the situation assessment system. Application areas of the developed model and algorithms are such as crisis management, natural and man-made disaster, threat assessment in the area of anti-terrorism activity, military applications, computer network security, etc. The developed model and algorithms are used for development and implementation of the system destined for assessment of the security status of a computer network attacked by malefactors. This result was highly evaluated of specialists. In particular, the paper [2] formulating the basic results received on the task in question was presented on the one of the important international conferences (with more than 450 presentations) received The Best Paper Award. The main results on this result were published in Russian and international publishers ([1,-4, 7, 8]).
7. The basic components constituting together a formal model aiming at simulation of distributed coordinated attacks are developed. These components include: (1) Conceptual models of agent-based specification of sophisticated distributed coordinated attacks, in particular, DDoS attacks; (2) Formal model destined for specification of adversary interactions of teams of hackers; (3) Fragments of ontology of attacks and mechanisms for protection against them; (4) Variety of agents' team structures, mechanisms of their interaction and coordination, the specifications of hierarchy of action plans and ways of roles assignment and agents' plans distribution. The basic models were validated via simulation of such interactions on the basis of the developed multi-agent software prototypes.
8. The analysis of development of deception information systems (DIS) was elaborated. The general functions and architecture of advanced DIS were offered. The generalized models and algorithms of disguised counteraction to remote non-authorized access to information resources of computer systems were developed. The software prototypes of DIS were developed and implemented.