Prof. Vladimir Gorodetsky St.Petersburg Institute for Informatics and Automation (Russian Academy of Sciences), St.Petersburg, Russia

Agent and Mining Technologies for Smart Space: Perspectives and Challenges

Abstract

In the talk, smart space is understood in a broad interpretation, i.e. as a distributed ambient, possibly, hierarchical environment with existing, inside it, dynamic multitude of living and nonliving entities solving own and collective tasks. This environment should be capable to provide for them with context–dependent information, communication, services, reminders and personalized recommendations in user–friendly mode where and when needed. This definition encompasses, in addition to smart home and like environments, many others, e.g., environments to be ad-hoc created for emergent situation mitigation and recovery, security systems of the future, smart home with assisting robots, hierarchy of smart spaces like smart city–smart building–smart room + personal smart environment of a human, car smart environment, etc. Moreover, it also includes what is now called a network–centric battle space environment.

Currently the smart space theory and practice is still at infancy age so far. In the existing research and developments, as a rule, simplified modes of smart space and limited multitudes of not–subtle applications are considered. Usually, in the existing research, a smart space population consists of a single entity and services provided are relatively simple. Human behavior is analyzed to detect particular actions, but not his/her complex intention–driven behavior. It can be said that now the smart space research and developments investigate and accumulate particular facts, approaches, models, technologies and experience to in–depth understand this many–sided and infinitely complex problem.

A specificity of the smart space research and development field is diversity of models, frameworks and technologies to be used and therefore most challenging problems are here associated with the necessity of their integration. Smart space is formed by the use of a huge volume of heterogeneous information obtained from distributed sensors measuring environment characteristics (temperature, humidity, etc.), physiological properties of human, his/her localization and movements, statuses of home appliances and so on. It should include many subsystems performing sophisticated fusion of context sensitive information, existing knowledge, as well as knowledge to be online extracted and accumulated. It should efficiently search for and compose web services. It should be capable of extraction of human behavior patterns and predict behavior. To learn multiple personal human profiles in order to develop context–dependent recommendations. All these tasks remain to be challenging.

This talk is about the author’s vision of some perspectives concerning benefits that could be provided for smart space by integrated use of agent and data mining technologies in distributed context dependent and human intention–aware information processing intended for providing intelligent assistance and personalized recommendations.

According to the author’s opinion, future smart space architecture should be composed of multiple overlay networks of autonomous intelligent entities (agents) set up on top of an infrastructure(s) layer supporting distributed interactions and communication of heterogeneous software, hardware and information sources (sensors, data bases, web resources). Each layer should correspond to a virtual ad-hoc network of nodes solving jointly a specific designated task in coordinated and cooperative mode and communicating in peer–to–peer (P2P) style. As a rule, each such overlay network should be dynamic with regard to both its population of nodes and topology. Therefore, a typical task caused by the aforementioned fact is self configuration intended for better coordination and cooperation in favor of overlay network node assigned task. This task can be efficiently solved based on agent–based distributed data mining technology. In fact, this task is reduced to optimal neighbor selection by each network node. The second task that should be locally solved by each node is improvement of its target task solution using experience accumulated and off–line and on–line learning. In fact, both these tasks are data mining task. These tasks are typical for many parts of smart space system and integrated agent and data mining technology can efficiently cope with them. One more typical task is data, information and decision fusion to take into account multiple–sources heterogeneous information of various levels of abstractions. The talk is about how integrated use of data mining and agent technologies can contribute to smart space in order to solve the aforementioned typical tasks.

In the talk, a brief overview of current results in the area of agent and data mining integration will be given and its contribution to the three aforementioned tasks of smart space will be demonstrated on the basis of some concrete smart space–related tasks examples. In parallel, a context-dependent (ontology and data–driven) efficient data mining approach specifically appropriate for smart space related knowledge discovery will be presented.

Special attention will be paid to the behavior pattern mining that is now a subject of research in the Behavior Informatics scope. The idea here is to present intention–driven multiple scenario behavior of a human in terms of dedicated specification language that makes it possible to represent behavior patterns of a human by the special structure called Scenario Knowledge Base. This structure represents behavior knowledge as a structure of human goals to be achieved by performing this or that actions. This behavior representation makes it possible to predict human behavior as well as to detect his/her intention. In fact, this model makes it possible to introduce, along with situational awareness, a new useful notion, intentional awareness, and use it as a mean to provide user with personalized context– and intention– driven assistance and recommendations. It is important to note that in current research this important issue of smart space research is paid too little attention. An example of agent system demonstrating this issue will be given.

In conclusion of the talk, a benefit of integrated use of agent and data mining technologies forming a core architecture and technology of future smart space will be summarized.

	Dr. Alexey Vinel St.Petersburg Institute for Informatics and Automation (Russian Academy of Sciences), St.Petersburg, Russia

Intelligent Transportation Systems from Cyber Physical Systems Perspective

Abstract

Recent advances in information and communication technologies made it possible to introduce a vast variety of Intelligent Transportation Systems (ITS) aimed to reduce traveling time, increase safety and ecological compatibility of different means of transportation. ITS is a cross-disciplinary area, which makes use of the up-to-date results in traffic engineering, wireless communication, data processing, automation and human behavior studies. This fact leads to a situation, when different components of ITS are often designed independently of each other. However, an increase of the ITS performance can be achieved by a tight coupling of computing and communication aspects of the system with its physical dynamics, what corresponds to the conception of Cyber Physical Systems (CPS). In this tutorial-style talk we first provide definitions and classifications of ITS and CPS. Then by the example of road traffic cooperative safety systems architecture we explain conventional design and performance evaluation approaches as well as enhance them for the case when mutual effects of communication, computation and physical processes are considered. We illustrate the benefits and drawbacks of CPS-based vision as well as its applicability to the design of different types of ITS.

Bio

Alexey Vinel is a Head of telecommunications technologies and computer networks group and a Senior research scientist of Saint-Petersburg Institute for Informatics and Automation (Russian Academy of Sciences). He received his Ph.D. (kandidat nauk) (2007) degree in technical sciences from the Institute for Information Transmission Problems (Russian Academy of Sciences, Moscow). Prior to that he was studying at Saint-Petersburg State University of Aerospace Instrumentation, where received his Master (2005) and Bachelor (2003) degrees in information systems, both with honours. He did internships in Germany at Siemens AG in Munich (2004-2005), University of Wuerzburg (2006-2007) and carried out numerous short-term visits with different European organizations. He has conducted scientific research in the areas of wireless network performance evaluation, development of random multiple access algorithms and has published articles in internationally recognized journals, such as IEEE ComMag, IEEE JSAC as well as many conference proceedings. He is Fellow of Alexander von Humboldt Foundation, Member of Organizing and Technical Committees of numerous international conferences.

	Prof. Chung-Ming Huang Distinguished Professor National Cheng Kung University, Taiwan

Demodulating and Encoding Heritage (DEH): Multi-Disciplinary Cross-Domain Research, Development and Deployment over Computer Science/Engineering and Human Literature Science for U-Heritage

Abstract

Two elements that currently need to be emphasized in Computer Science and Engineering (CSE) are (1) innovative applications and services and (ii) multi-disciplinary cross-domain R&D. With the advance of broadband wireless mobile networks technology (3.5/4G and Wi Max) and powerful handheld devices (iPhone, Google Phone, HTC), R&Ders are investigating the feasibility of anywhere, any time, ubiquitous applications and services. Through the multi-disciplinary cross-domain R&D effort, one can chain theory, technology, application/service and content to layout his research to real development and then to the practical deployment and use. In this talk, a case study of multi-disciplinary cross-domain research, development, and deployment called Demodulating and Encoding Heritage (DEH) for U(biquitous)-Heritage over CSE and Human Literature Science is given. Instead of having virtual heritage exploration in some desktops, a user can use the DEH system to physically walk into some heritage to have real experience and, at the same time, the user’s handheld device that has 3.5G wireless connection and GPS can receive the corresponding heritage information from the DEH server. In DEH, CSE people are in charge of the aspects of technology and application/service system, which contain R&D of database, intelligent inference, location-based service and navigation, and web technology for U-Heritage; Human Literature Science people are in charge of the content creation, which contains Point Of Interests (POIs) and Line Of Interests (LOIs), for U-Heritage. Main technical issues and some experience of cross-domain research, development, and deployment based on DEH will be introduced.

Bio

Chung-Ming Huang received the B.S. degree in electrical engineering from National Taiwan University on 1984/6, and the M.S. and Ph.D. degrees in computer and information science from The Ohio State University on 1988/12 and 1991/6. Currently, he is a Distinguished Professor in Dept. of Computer Science and Information Engineering, National Cheng Kung University, Taiwan, R.O.C. He is also Director of the Promotion Center for the Telematics Consortium (PCTC), Ministry of Education (MOE), Taiwan, R.O.C. He was (i) Chair of Dept. of Computer Science and Information Engineering and Director of Institute of Medical Informatics, National Cheng Kung University, Taiwan, R.O.C., and (ii) Principal Project Reviewer of Industrial Development Bureau and Department of Industrial Technology, Ministry of Economic Affairs (MOEA), Taiwan, R.O.C. He got the Best Paper Award in the 20th IEEE International Conference on Advanced Information Networking and Applications (AINA 2008) on 2008 and 2008 International Computer Symposium (ISC2008). He edited world’s 1st Telematics book - “Telematics Communication Technologies and Vehicular Networks” that was published by Information Science Reference (IGI Global). He has published more than 200 referred journal and conference papers in wireless and mobile communication protocols, interactive multimedia systems, audio and video streaming and formal modeling of communication protocols. His research interests include wireless and mobile network protocol design and analysis, media processing and streaming, web technologies, and network applications and services.