FuturICT 2009
Hungarian-Japanese Joint Conference on
Future Information and Communication Technologies
29-30 June, 2009, Budapest, Hungary

Abstracts

Keynotes

Title: Physical Impairment Constrained Routing in WDM Networks
(Keynote)
   
Abstract: There are various linear and non-liniar physical effects that impair the optical signal. The longer the distances are the worse the signal quality becomes. All-optical re-amplification is feasible today, however, all-optical re-shaping and re-timing is comercially not yet available. Therefore, O/E/O conversion and electronic regeneration is required. In this talk we present two low-cost methods for handling impaired signals to increase the network throughput.
The first one uses grooming as the implicit means of O/E/O regeneration. The interesting result is that the number of grooming actions has increased negligibly, while the throughput growth is significant.
The second one tunes power level of certain wavelength channels to different levels in order to improve the quality of long wavelength paths, while maintaining the power budget of the links. This way the long wavelengths have better quality, while the short wavelengths have still satisfactory quality. This approach also increasis the throughput of a network of that size where the distances start to become critical.
We present the problems and our solution methods as well as simulation results.
   
Speaker: Tibor CINKLER
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Towards New-Generation Networks: Japan’s Approach
(Keynote)
   
Abstract: As a future- or post-Internet, R&D activities for the new-generation network have been evolving world-wide. Since 2008, the speaker has been involved in the activities of R&D efforts at NICT (National Institute of Information and Communication Technology), where all strategic plans towards the new-generation network have been drawn up. In this talk he will introduce recent activities towards the new-generation network in Japan and especially at NICT. In October 2008, NICT published its future visions for the new-generation network. NICT emphasizes the socio-economical aspects of the new-generation network to minimize the negatives, to maximize the potential, and inclusion - including everything i.e., individuals, society, and earth. More recently, in March 2009, the technological strategy including the technology roadmap for the new-generation network has been published. The technology roadmap consists of five network targets, which explore technological aspects of the future networks, but still features the socio-economical aspects of our future vision. In this talk, the speaker will introduce the above topics and also discuss possible directions of the new-generation network.
   
Speaker: Masayuki MURATA
Professor, Advanced Network Architecture Laboratory, Graduate School of Information Science and Technology, Osaka University, Japan
  Back to the conference program


Session 1: Network of the future


Title: Building the Measurement Infrastructure and Virtual Data Observatory in the European Future Internet Experimental Facility
(Invited)
   
Abstract: The future Internet research requires enhanced experimental techniques that aim for both large scale overview and deep insight of the network traffic characteristics. Novel methods have to be developed to handle the uprising protocols, router virtualization and the requirements of new kinds of applications.
In my presentation I will introduce the OneLab2 experimentation facility of the EU FIRE – Future Internet Research and Experimentation program, which is a GENI like initiative for future Internet research in Europe. The software and hardware solutions of the OneLab2 measurement facility are based on the techniques used in the ETOMIC Measurement Infrastructure including a special active measurement card. I will present the current and final measurement capabilities of the European Future Internet Testbed.
   
Speaker: Péter HÁGA
Department of Physics of Complex Systems, Eötvös Loránd University (ELTE), Hungary
  Back to the conference program

Title: New Generation Network Architecture Design and Optical Grid over Wavelength Switched Optical Network
(Short talk)
   
Abstract: In this talk, the speaker will address on-going activities of the AKARI Architecture Design Project, where the new generation network is designed to support future information society. This project is a five-year project and will produce a blueprint of the new generation network in 2011. The speaker will present design principles and enabling technologies under the design principles.  Then, the speaker will also present optical grid over wavelength switched optical network for providing parallel computing users multiple end-to-end lightpaths on demand basis simultaneously. He will show the overview of this system and experimental results using a real testbed over JGN2plus optical fibers in the Tokyo metropolitan area.
   
Speaker: Hiroaki HARAI
Group Leader, Network Architecture Group, New Generation Network Research Center, National Institute of Information and Communications Technology (NICT), Tokyo, Japan
  Back to the conference program

Title: High Speed Network Monitoring and Traffic Analysis
(Short talk)
   
Abstract: A network monitoring and traffic analyzing system can be of great help to service providers both in everyday operation and maintenance work as well as in future dimensioning of the network. The company AITIA in close cooperation with the Budapest University of Technology and Economics (BME) has developed several systems and tools for such tasks to capture, filter, store and analyze traffic from thousands of 64 kbps SS7 signaling links to high speed Ethernet traffic. (It is interesting to note that we have developed the lower layers, too, for NTT countrywide system to monitor ATM traffic on Japanese gigabit network.) Present day research is concentrated on sophisticated analysis for deep inspections, i.e. bottlenecks, burst- or overload-analysis, application identification or traffic mix determination.
   
Speaker: Péter TATAI
President, AITIA International Inc., Hungary
  Back to the conference program

Title: NTT R&D Challenges for Future Carrier Networks
(Short talk)
   
Abstract: In this talk, we present our research activities for future carrier networks. First, we clarify the service domains then describe our research and development targets and approach to the future network infrastructure for carriers. We think that the network should be capable of providing three kinds of services to meet customer requirements: huge-bandwidth services, broadband packet services, and tiny-bandwidth mass services. An effective way to achieve flexible deployment of service applications will be to integrate resource virtualization into network service management.
   
Speaker: Atsushi TAKAHARA
Executive Manager, Media Innovation Laboratory, NTT Network Innovation Laboratories, Japan
  Back to the conference program

Title: Mobility Architecture Based on ID/Locator Split and Cross-Layer Interaction
(Invited)
   
Abstract: The paradigm of ID/Locator separation has attracted considerable attention for support of mobility, multihoming, security, and scalable routing. With keeping compatibility to IPv6, we developed a network layer protocol named LIN6 based on the ID/Locator separation paradigm. LIN6 achieved optimal routing to mobile nodes without header overhead. In AKARI, one of the real network models of the New Generation Network developed at NICT, we redesigned the architecture based on the ID/Locator separation paradigm. The proposed architecture takes communication privacy into account and permits variety of L3 protocols. For communication privacy protection, we introduced two kinds of IDs: the primary ID and the active ID. The primary ID is the unique identifier of the node but is not contained in the header. The active ID is a temporary identifier and is contained in the header. The active ID changes per communication session. It is assumed that the future network employs several kinds of L3 protocols, for example, a network composed of non-powerful nodes employs a simple L3 protocol while that of powerful nodes employs a sophisticated L3 protocol. Even in such an environment, our architecture allows a node to communicate with the target node by specifying the ID even if the target node is connected to a network, which uses other L3 protocol.
The cross-layer paradigm has also attracted considerable attention for a node to adapt to a dynamic environment. One of the purposes of the cross-layer paradigm is fast handover. We designed a cross-layer architecture named CEAL, which focuses on control information exchange between layers in a node. We achieved fast handover in LIN6 by exchanging control information between L2 and L3. We also achieved fast failover and fast handover in SCTP by exchanging control information among L2, L3, and L4. In AKARI, we expanded the cross-layer paradigm so as to exchange control information between nodes in the network by introducing control plane.
   
Speaker: Fumio TERAOKA
Professor, Department of Information and Computer Science, Faculty of Science and Technology, Keio University, Japan
  Back to the conference program

Title: Entanglement Measures in Quantum Computing
(Short talk)
   
Abstract: Entanglement lies in the heart of quantum computing”, but giving good entanglement measures is a rather hard task. Quantum bits are treated as distinguishable particles, but in real life applications indistinguishable particles also appear. Usually the entanglement can be well described by entropy, Schmidt rank and concurrence but in case of three or more, indistinguishable particles the generalization is not trivial, more types of entanglement classes can be found and be described with the aid of two or more quantities. In the present work, the speaker – after a short introduction to qubit-entanglement and an overview of two-fermion entanglement measures and their properties – gives some recently studied three-particle entanglement measures with a generalization possibility to any fermionic system. Some relations between the measures of different sources will be studied as well.
   
Speaker: Szilvia NAGY
Senior lecturer, Dept. of Telecommunications, Széchenyi István University, Hungary
  Back to the conference program

Title: The Evolution of Optical Packet Switching
(Short talk)
   
Abstract: In "new" generation optical networks, high scalability and fine granularity will be essential, in addition to increased network capacity. Although an approach using Internet protocol (IP) over generalized multi-protocol label switching (GMPLS) with WDM provides fine granularity, its slower electronic processing will be a bottleneck in the network. To avoid such bottlenecks in commercial high-end IP routers, electronic parallel processing technologies are often used. However, such large-scale parallel processing leads to serious power consumption problems. Despite the relative immaturity of optical technologies, many optical packet switching (OPS) systems have been developed to exploit the merits of OPS systems, such as high capacity, ultra-high-speed hopping, and fine physical granularity. Recently, 160 to 640 Gbit/s/port colored optical packet switch prototypes are demonstrated. Arrayed burst-mode Tx./Rx., all-optical multiple label processor, high-speed polarization independent optical switch, optical buffer with FDL, parallel and pipelined buffer managere, and burst-mode EDFA are introduced as key technologies. Effects of polarization rotation due to the field transmission on the wideband optical spectra are shown. Furthermore, 1.28-Tbit/s/port ultra-wideband optical packet switching with high-speed polarization independent optical switches is demonstrated. In addition, 640 Gbit/s throughput granularity flexible, 3 node optical network using packet selective ROADM prototypes are also demonstrated.
   
Speaker: Naoya WADA
Group Leader, Photonic Network Group, New Generation Network Research Center, National Institute of Information and Communications Technology (NICT), Tokyo, Japan
  Back to the conference program

Title: Monitoring Network Bias
(Short talk)
   
Abstract: A fundamental question in the net neutrality debate is the extent to which network operators should be allowed to discriminate among Internet packet streams. Unfortunately, the state of affairs in reality has dramatically surpassed academic debating at various levels. Rough competition among Internet service and content providers, as well as huge monopoly potentials, are radically increasing the scale of discriminatory acts in the Internet. We propose an end-host based monitoring service (based on PlanetLab and M-Lab) capable of detecting such biased network behavior. The primary goal here is not to take sides in the emerging net neutrality debate, but rather to design a system capable of making the Internet more transparent. This project is joint work with Prof. Aleksandar Kuzmanovic at Northwestern University.
   
Speaker: Gergely BICZÓK
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: The Impact of IPTV Service on NGN
(Short talk)
   
Abstract: "New wine calls for new wineskins." Our next generation IPTV service will call for the Next Generation Network (NGN). This service has got more than 500,000 subscribers within one year. There are several features on this service, which use the global open standard technology H.721 to realize a STB-free watching environment. This means that TV set will have the IPTV function and the user can choose any service provider and network provider based on the standard. High quality signal for the DTT retransmit service can be achieved by the NGN. We are thinking of several new services for this brand new media which is based on the metadata technologies. Some of them could be the essential function only by the IPTV service. The IPTV service will launch from the same service with CATV or satellite broadcasting service to more attractive and personalized world soon.
   
Speaker: Jay KISHIGAMI
Executive Director, NTT Cyber Solutions Laboratories NTT, Vice President, Japan
  Back to the conference program

Session 2: Ubiquitous computing, services and applications

Title: 3D Internet for Cognitive Info-communication
(Invited)
   
Abstract: The paper presents a high-speed internet-based network capable of merging 3D virtual and real Intelligent Space environments into augmented intelligent reality environments, as well as transmitting them to remote locations. The network also enables human interaction with Intelligent Space located another continent within these augmented intelligent reality environments. The Intelligent Space surrounds the operator and it can understand the commands of the operator at the highest level of abstraction possible. For example, while a time consuming robot programming process is necessary in the conventional case, the intelligent space with a motion capture data suite can achieve similar results through the sensing of the operator's arm and hand movements.
The Intelligent Space is capable of not only interpreting these commands but also automatically handling problems of dynamics and precision. The space is able to track and recognize the movements of the robot, of the materials to be manipulated, as well as understand as many aspects of the environment as possible.
An experiment is presented where a simplified 3D virtual model of a laboratory at Tokyo University has been created and projected in Budapest, which can be used to control a virtual 3D robot. The real-world counterpart of the virtual robot can be found at Institute of Industrial Science at Tokyo University. The two robots were commanded by the gesture of the operator. The command was recognized by the Intelligent Space. The most important technological difficulty in the implementation was posed by the time delays inherent to internet-based communications. However, in the end, we were able to synchronize the movements of the virtual robot in Budapest and real-world robot in Tokyo using internet communications. This was achieved only through communications between the virtual 3D laboratory and Intelligent Space, and there was no direct communication between the two robots.
   
Speaker: Péter KORONDI
SZTAKI, Hungary
  Back to the conference program

Title: Glasses-Free Large 3D HD Display
(Short talk)
   
Abstract: NICT has been working on a projector array based 3D display. Without having "3D glasses" on, this display gives viewers both remarkable and suitable image experience, and such observed images vary according to viewpoints. The display’s resolution greatly surpasses 100 million pixels with exceptional screen optics to create true high definition 3D images in a 16:9 70" screen. This presentation reports in brief the development status of the display and also our discussed requirements and target applications.
   
Speaker: Shoichiro IWASAWA
Researcher, Multimodal Communications Group, Universal Media Research Center, National Institute of Information and Communications Technology (NICT), Keihanna Science City, Kyoto, Japan
  Back to the conference program

Title: Research Trends in Multimedia Content Services
(Short talk)
   
Abstract: The Web has revolutionized our relationship with information, creating new opportunities, markets and became the main medium through which people access and release information. With the recent trends towards the convergence of Web, television and mobile devices, the focus on content has switched from text to multimedia.
Multimedia information processing has the potential of revolutionizing human-machine relationship by the ability to process, categorize and retrieve media content and to provide clients with a new personalized look and feel via a more intuitive, user friendly and intelligent services. Current multimedia search systems are based on available annotation and the visual content. Visual content can for example be segmented or its definitive regions or points of interest can be selected and descriptive features can be generated based on color, texture and shape.
Recent research trends in information processing and retrieval focus on intelligent data mining, recommendation and multimedia search technologies that create the basis for the so-called Web 3.0 applications envisioned for the next 10-year period of Web development. Eric Schmidt, Google CEO claimed that the driving force of the Web in the near future will be fast and personalized applications accessible from any device, PC or mobile. These applications will be aided by data mining solutions handling distributed data by cloud computing. The basis of these systems is provided by recommendation systems based on community content and collaborative filtering.
In the talk we will present selected results in the areas of multimedia information retrieval and recommendation as well as user behavior profiling and analysis with emphasis on the research at the Computer and Automation Research Institute of the Hungarian Academy of Sciences. Some highlights include our KDD Cup 2007 winner recommender solution, our multimedia retrieval engine and our filtering technologies capable of removing manipulative content with the sole purpose of achieving high search engine ranking.
   
Speaker: András A. BENCZÚR
H
ead, Informatics Laboratory, Computer and Automation Research Institute of the Hungarian Academy of Sciences (MTA SZTAKI) , Hungary
  Back to the conference program

Title: Internet of Things and Places: Ubiquitous Computing Applications and Services in Japan
(Invited)
   
Abstract: Ubiquitous Computing has become very important in the computer industry and beyond. It has attracted much attention from wide audience lately. Ubiquitous Computing is often featured by several novel concepts or keywords such as “context-awareness”, “Internet of things”, “bridging the gap between real and virtual worlds”, and “information services triggered by tag-based identification”. RFID and/or sensor network are considered as key technologies which will realize these concepts. On the basis of the concepts of ubiquitous computing, industries and governments in the world are trying to provide various innovative information services such as efficient supply-chain management, product traceability systems, and location-aware information services.
In order to deploy these services in the real-world, we need not only to develop advanced information and communication technologies, but also to establish information and communication infrastructure based on open standardized technologies.
To answer these requirements, we have established Ubiquitous ID Center in 2003, in cooperation with more than 300 organizations such as industrial sponsors, universities, research institutes and public sectors. Ubiquitous ID Center is developing a fundamental architecture of ubiquitous computing called Ubiquitous ID Architecture, and a set of open technology standards based on the architecture. In this talk, we would like to present the architecture and technologies of Ubiquitous ID Center, and several practical applications and services based on Ubiquitous ID Technologies. Lastly, we will mention about the international standardization of ubiquitous computing technologies.
   
Speaker: Noboru KOSHIZUKA
Vice Director, YRP Ubiquitous Networking Laboratory, YRP, Japan
Associate Professor, The University of Tokyo, Japan
  Back to the conference program

Title: Towards Artificial Speech Functions Appropriate for Communication Situations
(Short talk)
   
Abstract: Artificial speech functions have been an area of intensive study for several years. Although practically applicable solutions have been developed there seems to be a gap between the R&D and the application community. The paper will demonstrate our results achieved in practical applications and our research ideas for increasing the appropriateness of speech technology for the requirements of real-world interactive communication situations.
Most application developers of currently available speech recognition and synthesis engines aim at solutions that can cover arbitrary communication domains or at least require minimal modifications. That approach supposes that the artificial components are used in simulating a human-human communication context. Although generic solutions perform basic functionality (e.g. in automatic e-mail reading) it is still not convenient for many people. They clearly recognize that the communication partner is a machine that tries to imitate human functions. The interface to the speech communication component is usually very simple (text input to speech synthesis, text output from speech recognition). That expresses just one (usually the most important) component of human-human communication and lacks other important elements such as sex, age, emotional state, etc.
As a response to this situation speech technology developers started to implement domain-specific solutions (e.g. automatic name- and address reader, railway and airline timetables, weather forecast, etc.) where computerized solutions come close to human performance. Additional information is included in the context itself. The disadvantage of this approach is the relatively high cost.
In a project just starting in June 2009 an alternative approach to human-computer interaction is under study that we call eto-informatics. The study is based on the results of etology. Humans have a long history of interaction with animals/pets. We suppose that after building a model for e.g. human-dog interactions it can be extended and mapped to human-computer interactions, i.e. computer functions could be modeled as a kind of artificial species. User requirements could probably be better fulfilled by artificial speech functions in this context.
   
Speaker: Géza NÉMETH
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Towards Web 3.0 as Collaboration Media in Real World
(Short talk)
   
Abstract: The reach of the Internet expands opportunities for involvement of social activities of people in the real world. The next generation of World Wide Web or “Web 3.0” will play the role of “collaboration media”, where collaborative data are treated by location-aware services with social interactions. Technologies for capturing real world situations or events have emerged, such as GeoRSS, mobile blogs like Twitter, GPS camera etc. Our voice micro-blog (vMblog) provides alternative way for annotating the real world by “people voices”, which allows more intuitive and impressive annotations. Those real world annotations are collected and managed by our moving phenomenon DBMS, in which both spatiotemporally- and thematically-correlated events are aggregated as individual “phenomenon”.  It allows “glocalization” of the annotations, which means gathering of local events for a global interest or scattering a global interest into local events. Finally, as an integrated infrastructure for coordinating knowledge sharing and problem solving in distributed environments, we also introduce “global knowledge grid”, which uses the basic functions of a grid and defines a set of additional layers to implement the functions of distributed knowledge discovery, analysis and provision.
   
Speaker: Koji ZETTSU
Senior Researcher, Knowledge Clustered Group, Knowledge Creating Communication Research Center, National Institute of Information and Communications Technology (NICT), Kyoto, Japan
  Back to the conference program

Session 3: Wireless communications

Title: Research and Innovation on Mobile Infocommunications at BME
(Invited)
   
Abstract: The talk is concerned with the methods and practice of the research and innovation activities at the Budapest University of Technology and Economics on the field of mobile infocommunication technologies and systems. The author deals with the research organization, the role of the staff members and the students, the research areas in mobile communication systems, and introduces the research areas of the Mobile Communication and Computing Laboratory. For illustrating the theoretical research results some examples are presented as: multi-user detection, interference analysis in FH channels, interference analysis in multi-user channels and call admission control. At the end of the paper the general activity of the Mobile Innovation Center is introduced and the industrial cooperation is described with some local and international industrial partners.
   
Speaker: László PAP
Budapest University of Technology and Economics (BME), Dept. of Telecommunications (HIT), Hungary
  Back to the conference program

Title: Research and Development on Cognitive Radio/Dynamic Spectrum Access Network
(Short talk)
   
Abstract: The National Institute of Information and Communications Technology, Japan (NICT) has researched and developed the Cognitive Radio/Dynamic Spectrum Access Network (CR/DySPAN) for ten years in order to improve frequency utilization efficiency by using spectrum sharing among several users (operators) in coming 5 to 10 years. Especially, from 2005 to 2008, our researches were supported by the Ministry of Internal Affairs and Communications (MIC). After 2008, the MIC is continuously enforcing new R&D projects to realize CR/DySPAN. This presentation introduces our research, development, and standardization activities on CR/DySPAN that NICT has achieved, and an overview of the new R&D projects starting this year.
   
Speaker: Hiroshi HARADA
Group Leader, Ubiquitous Mobile Communications Group, New Generation Wireless Communications Research Center, National Institute of Information and Communications Technology (NICT), Yokosuka, Japan
  Back to the conference program

Title: General Distributed Economic Framework for Dynamic Spectrum Allocation
(Short talk)
   
Abstract: The emergence of novel radio techniques enables the application of advantageous revolutionary spectrum policies. An important body of research has appeared about possible frequency management schemes, but none of them proposes solutions that meet every related criteria. In this talk I present our work on dynamic spectrum allocation and pricing that offers a distributed mechanism design, well-suited to practical employment issues. Our model handles interference effects without any restricting assumptions, provides universal scalable and incentive-compatible allocation and pricing mechanisms. Both analytical and numerical evaluation of the proposed framework will be shown, and in either case I prove this latter to be a suitable approach to efficient and flexible spectrum utilization.
   
Speaker: Attila VIDÁCS
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: R&D Activities for Future Mobile Communications
(Short talk)
   
Abstract: The presentation first introduces successful deployment of the 3G mobile communications in Japan, and then focuses on the expected evolution of the 3G technologies such as HSDPA/LTE/IMT-Advanced from a mobile operator’s point of views. The presentation also describes DOCOMO’s mid-term vision on ‘change and challenge to achieve new growth,’ that includes new and evolutional mobile services, then, conclude with a view on the research direction for the future mobile communications
   
Speaker: Tomoyuki OHYA
Director, Wireless Networking Research Group, Research Labs., NTT DoCoMo Inc., Japan
  Back to the conference program

Title: Fast-response inter-vehicle communications
(Invited)
   
Abstract: Safe driving support is one of the most attractive and important applications of an inter-vehicle communications system. Real-time and reliable exchange of information on data such as vehicle location, speed, sudden braking etc., among vehicles is a key to offering promptly warnings to drivers, in order to avoid fatal traffic accidents. For this purpose, the conventional media access control (MAC), CSMA/CA, used in wireless LAN (IEEE802.11) has a limit in transmission delay and packet delivery ratio, due to its control scheme based on the carrier sensing. For satisfying challenging requirements on supporting safe driving, we have proposed a new scheme called “Multi-carrier Multi-code Spread ALOHA (MM-SA)”, which is based on code division multiple access (CDMA) technology, and significantly reduces the control latency to millisecond order and improves the packet delivery ratio among vehicles. It also has inherent robustness to the increase of vehicle density as well as the hidden terminal problem.
In this presentation, I give an overview of the MM–SA scheme, including protocol design, performance evaluation by simulation and prototyping, which are developed for avoiding collisions at an intersection. I also mention our future study.
   
Speaker: Sadao OBANA
Director, Adaptive Communications Research Laboratories, Advanced Telecommunications Research Institute International (ATR), Japan
  Back to the conference program

Title: Fading-aware Reliable and Energy Efficient Routing in Wireless Sensor Networks
(Short talk)
   
Abstract: In this paper two fading-aware reliability based routing algorithms are proposed for Wireless Sensor Networks (WSNs) with lossy radio links which can find optimal paths in polynomial complexity. The lossy links are modeled by using standard fading models (e.g. Rayleigh and Rice fading). The proposed schemes minimize the energy consumption and ensure reliable packet transmission to the Base Station (BS) at the same time. Reliability is guaranteed by selecting the path over which the probability of correct packet reception at the BS will exceed a predefined threshold. The first algorithm minimizes the total energy consumption sending a packet over the selected path to the BS. The second algorithm selects a path which maximizes the minimum remaining energy and, as a result, balances the energy consumption among the nodes yielding high longevity. In both cases, reliable and energy efficient packet forwarding in WSN can be reduced to a constrained optimization problem. By using a specific link metrics, these problems can then be mapped into shortest path problems solved in polynomial time. Thus the obtained results ensure reliable path selections with minimum energy consumption in real time.
   
Speaker: János LEVENDOVSZKY
Budapest University of Technology and Economics (BME), Dept. of Telecommunications (HIT), Hungary
  Back to the conference program

Title: Introduction to the Medical ICT Projects at NICT
(Short talk)
   
Abstract: By the arrival of aged society in our countries, the information and communications technology (ICT) is expected to contribute to medical and healthcare fields. For example, medical radio equipment senses our vital information (blood pressure, SpO2, body temperature, cardiac beats, body motion etc.) and sends the data to a doctor to get medical treatment. In our medical ICT group at NICT, the possibility and the profit of wireless ICT in the medical and healthcare fields, especially a wireless body area network (BAN) system, are being examined. Recent activities of medical ICT project are briefly introduced.
   
Speaker: photo Kiyoshi HAMAGUCHI
Group Leader, Medical ICT Group, New Generation Wireless Communications Research Center, The National Institute of Information and Communications Technology (NICT), Yokosuka, Japan
  Back to the conference program

Title: High Precision In-door Localisation by Non-Linear Noise Reduction
(Short talk)
   
Abstract: Our work was aimed at creating high-precision indoor localisation (≤1.5m) methods based on signal strength of wi-fi access points in a knowledge poor and unknown in-door environment. We proposed a lightweight intuitive technique for signal strength based localisation problems. In-door localisation is a key problem for indoor logistic, asset management, home automation, and robot based applications. There are two main classes of available techniques for locating an object in an indoor environment; the one strongly involves the knowledge and help of an operator, while the other requires cross-validations on locations of the surroundings world. In our work, we were focusing on the latter one. The algorithm was tested and evaluated in an environment where no map information was given, limited number (<4) training samples with no trace information were available for the most of 1m × 1m location cells, and at most three access point were viewed from a certain location. Evaluations indicated our methodology were proven ground in knowledge poor environments, it has a 62% precision (within 1,5m) with a 3,1m root mean square error rate.
   
Speaker: photo Zsolt T. KARDKOVÁCS
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Posters

Title: Secure-TinyLUNAR: A Secure Label-switching Routing Protocol for Wireless Sensor Networks
(Poster)
   
Abstract: Although a multitude of routing protocols exist for wireless sensor networks developed for various application domains, secure sensor network routing protocols do not exhibit such variety. In addition, those few sensor network routing protocols that were developed with security in mind still lack a formal proof of their security properties. In order to remedy this situation, we propose a novel secure routing protocol, called Secure-TinyLUNAR, for wireless sensor networks, and we formally prove its security. In our model, security is defined in terms of the correctness of the routing table entries of the honest nodes. Besides its provable security, Secure-TinyLUNAR has three further advantages. First, similar to TinyLUNAR, it uses label-switching routing, which results in reduced addressing overhead during data packet forwarding. Second, it employs symmetric key cryptography exclusively to authenticate control messages which incurs substantially less computational overhead than digital signatures. Finally, due to label-switching routing, intermediate nodes do not need to check the authenticity of the message origin that results in further energy savings.
   
Author: Gergely ÁCS
Budapest University of Technology and Economics (BME), Dept. of Telecommunications (HIT), Hungary
  Back to the conference program

Title: Link Failure Localization with M-trails in All-Optical Mesh Networks
(Poster)
   
Abstract: Failure localization serves as one of the important functional modules for control and management in the modern communication networks. Failure localization and monitoring becomes particularly critical in an all-optical mesh network due to its high date rate along each fiber and transparency in the data plane, where a single fiber cut may cause serious damage on the network service continuity and availability. One possible solution is to monitor each link with a single monitor, called link monitoring, which requires an excessive number of relatively expensive devices. The goal of this research is to unambiguously localize all single or dual link failures of the network while drastically lower the number of monitors used by link monitoring. Our framework provides a new  monitoring concept by allocating supervisory light-paths for monitoring purposes, called m-trails, where the m-trail shapes a simple path with some possible loops. The proposed methods are presented for localizing single and adjacent dual link failures. Possible solutions for arbitrary dual failures are presented by adopting the recent results of Combinatorial Group Testing theory.
   
Author: Péter BABARCZI
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Large Vocabulary Continuous Speech Recognition of Morphologically Rich Languages
(Poster)
   
Abstract: Speech retrieval is a special case of information retrieval in which the information is in spoken form. The size and type of accessible audiovisual archives are growing contiunuosly, therefore speech retrieval systems and the underlying large vocabulary speech recognition (LVCSR) technologies are getting higher emphasis. Recognizing spontaneous databases is a challenging work, especially for morphologically rich, agglutinative and inflective languages, like Hungarian. Our toolkit is based on state of the art acoustical and language modeling methods, including phoneme- and grapheme-based acoustic modeling techniques, morpheme based statistical language modeling and speaker adaptation. Depending on the task 60-80% recognition accuracy can be achieved, which is generally adequate for multimedia database indexing.
   
Author: Tibor FEGYÓ
AITIA International Inc., Hungary
  Back to the conference program

Title: Cognitive Radio/Dynamic Spectrum Access Network
(Poster)
   
Abstract: TBA
   
Author: Hiroshi HARADA
Group Leader, Ubiquitous Mobile Communications Group, New Generation Wireless Communications Research Center, National Institute of Information and Communications Technology (NICT), Yokosuka, Japan
  Back to the conference program

Title: Cognitive Engineering Behind a Webcasting Application: A Case Study
(Poster)
   
Abstract: Today web-technologies enable the Web to be a full-fledged broadcasting medium. Such live webcasts are common for sport events of different audiences. Large-scale events are webcasted to deliver additional content beside traditional media coverage and smaller exotic sports also utilize this channel as the main medium for cost advantages. Textual commentaries and live results form a unique feature completing the dynamic video and audio streams. There is a class of endurance sports where athletes do not start with a mass-start but they have predefined individual start times. Such sports are cross-country skiing, orienteering, and multi-sports just to name some. The competition is not easy to follow. Extensive calculations and book keeping of times are required. The situation is complicated further by the longer extent of the competition in time, and heats or classes competing parallel. At high profile events computer supported electronic timekeeping is used. These systems provide split times from the course/track and can carry out the necessary calculations and comparisons. An appropriately designed web-application can form a live result system, which is capable of delivering insight into and excitement of the competition. In this presentation the special design issues of a novel live results webcasting application will be discussed.
Considering the commonly deployed live result and split time systems for individual start competitions a lot of deficiencies can be identified. The novelty of our proposed solution is the highly effective utilization of screen real estate and supporting the user via attention guidance. These goals are achieved by the application of elastic focus+context tabular displays in a nested control attention guidance frame. Current web-technologies such as AJAX can provide the necessary user interface tools to build these interfaces.
   
Author: Tamás HECKENAST
Senior lecturer, Dept. of Informatics, Széchenyi István University, Hungary
  Back to the conference program

Title: Pedestrian Recognition
(Poster)
   
Abstract: Our work was aimed at creating a framework for image content based indexing, retrieval, and pattern recognition. We proposed new methodology in order to fasten image content retrieval by mixing visualisations, thinning techniques, and well-known classification algorithms. The basic idea was first tested in DaimlerChrysler-NiSIS pedestrian recognition challenge 2007 where the joint group of our department were awarded by the best overall performance (98% precision) special prize.
Image content retrieval is a hard task since there is no commonly accepted model for human perceptible visual information representation. On the other hand, computers may enhance human capabilities in multi-dimensional modelling and computational intelligence. Our framework combines the human and computer abilities in a single framework. The framework consists of a visual mining part in which wide range of data type can be visually represented. It supports commonly used, and star map, bat, hierarchical tree charts as well. Key patterns are identified by data miners who select the most appropriate classification algorithm for the task. If represented data have more than two dimensions thinning techniques are used. It is clearly shown that the use of our framework increases the overall performance and radically decreases the development time.
Since colouring clearly identifies different classes we currently running projects on how to automate selection and tuning of classification algorithms.
   
Author: Zsolt KARDKOVÁCS
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: WISDOM: A Web Information Credibility Analysis System
(Poster)
   
Abstract: We demonstrate an information credibility analysis system called WISDOM. The purpose of WISDOM is to evaluate the credibility of information available on the Web from multiple viewpoints. WISDOM considers the following to be the source of information credibility: information contents, information senders, and information appearances. We aim at analyzing and organizing these measures on the basis of semantics-oriented natural language processing (NLP) techniques.
   
Author: Daisuke Kawahara Daisuke KAWAHARA
Senior Researcher, Knowledge Clustered Group, Knowledge Creating Communication Research Center, National Institute of Information and Communications Technology (NICT), Kyoto, Japan
  Back to the conference program

Title: Autogenerated Web Search Directory Torishiki-kai
(Poster)
   
Abstract: The goal of this research is to present, in an intuitive directory structure, valuable additional search terms relevant to a user's topic of interest. Selecting one of these search keywords displays a list of Web pages containing information about the search term and the selected keyword. For this purpose we have autogenerated a Web search directory called Torishiki-kai from a large collection of Web documents using state of the art knowledge acquisition methods. Torishiki-kai maps out the context of use of the terms input by the user, and classifies topically related search terms according to semantic categories such as potential troubles, methods or tools to help the user find potentially relevant ``unknown unknowns''.  Underlying this notion of context of use is the assumption that what people want to know about a given search topic is, in a broad sense, related to its intended use (e.g. ``reading'' a novel), or preparation for this use (e.g. ``buying'' the novel). Restricting the semantic search space to the context of use of whatever the user inputs lets us delineate the scope of relevant keywords in a more meaningful way, allowing the user to discover meaningful but previously unknown information through predefined semantic categories without drowning in a sea of related but irrelevant information. For ease of navigation, the distance to the category center indicates the co-occurrence frequency of the keyword, and the angle between keywords reflects their semantic similarity, which is computed using the EM clustering method. We believe that the technology to pro-actively find and organize relevant information vital to our day to day goals will become increasingly important as the Internet continues to permeate further aspects of our lives, and we see Torishiki-kai as an important step in that direction.
   
Author: Junichi KAZAMA
Researcher, Language Infrastructure Group, Knowledge Creating Communication Research Center, National Institute of Information and Communications Technology (NICT), Japan
  Back to the conference program

Title: Wireless Technologies toward Medical ICT System
(Poster)
   
Abstract: NICT has developed, in cooperation with Yokohama City University and other research institutions, a portable electrocardiogram device that a person can wear on his/her body. The data collected by the device can be transferred to remote areas in a timely manner. Researchers are now seeking help from hospitals in order to examine the technology in more practical settings.
This portable electrocardiogram device is as big as one’s palm and weighs only 20 grams. Using a wireless network, it sends real-time data to surrounding computers and cell phones. The method of networking processes is modified in such a way that saves energy for the device to operate: With one button cell, it runs for 24 hours. As its main functions, the device can figure out a wearer’s heart rate from collected electrocardiographic data, and it can also be used in combination with a thermo sensor.
The device constantly provides an up-to-date electrocardiographic data of a user. The data is also transferable to remote locations. These functions are expected to aid in the self-maintenance of health by the elderly and in monitoring medical conditions of patients.
Researchers are planning to perform tests to verify the applicability of the device in real-life situations, such as at hospital. Their goal is to put it to practical use in a couple of years.
   
Author: Hachihei KUREMATSU
Technical Expert, Project Promotion Office, New Generation Wireless Communications Research Center, National Institute of Information and Communications Technology (NICT), Japan
  Back to the conference program

Title: Efficient Heuristic Topology Optimization and CapEx Estimation of FTTx Access Networks
(Poster)
   
Abstract: Several optical access network technologies are available for network operators providing broadband services (Fiber-to-the-X solutions, such as PONs, VDSL and active Ethernet or point-to-point fiber networks). These are mature and standardized technologies now in deployment phase; therefore network and topology design issues play an increasingly important role. Theory has to be put into practice, considering as much information about technology and geography as possible to design optimal topology that ensures minimal investment cost and maximal profitability. In addition to the task of topology planning, a framework is provided for comparison of these technologies on the same area.
We address broadband optical access network design minimizing deployment costs, taking operation issues into account, using detailed cost and network models of the above listed FTTx technologies that suit best to actual networks due to detailed financial metrics used instead of just minimizing fiber lengths.
Our proposed methods use real geographical information (maps), supporting network designers to find the optimal technology solution and to assess investment- and operational costs of given development area.
Generally known optimization algorithms cannot handle practical network sizes of ten thousands of customers on city-scale network areas due to complexity issues, therefore properly chosen heuristics are sought. Our graph theory based solutions work fast even for large problem instances (within minutes), providing results with a difference less than approximately 10-20% from the computed ILP (Integer Linear Programming) optimum.
Our FTTxDesigner Framework provides a complete workflow for map processing, topology planning, business analysis and techno-economic comparison. Several case studies were carried out for real-life network and service requirement instances (number of customers ranging from 400 to 20.000).
   
Author: Attila MITCSENKOV
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Multi-User MIMO in OFDMA Systems with Adaptive Modulation and Power Control
(Poster)
   
Abstract: OFDM technique is the one of fundamentals of next generation (4G) mobile communication networks, which is able to adapt to the challenges induced by the time-variant radio channel – such as multi path propagation – providing broadband communication for users at the same time. Evolution of computational capacity permits the integration of multi-user extension of OFDM into telecommunication standards like IEEE 802.16e (WiMAX) or 3GPP LTE. The mentioned multi-user variant is OFDMA, which is the most effective solution among OFDM based multiple access schemes exploiting multi-user diversity. Application of spatial multiplexing (SDM or MIMO) enables – in principle – the multiplication of system’s spectral efficiency by increasing the number of transmit and receiver antennas. In MIMO-OFDMA, radio resource management – especially subcarrier and transmit power allocation – is essential question. The optimization task is extended with QoS and fairness constraints, which are commonly conflicting goals with spectral efficiency maximization. Present work includes proportional fair OFDMA-MIMO (time-frequency) scheduling, which considers actual radio channel state information. Subcarrier allocation based on eigenvalue-product criterion of proper MIMO channels is realized. Adaptive M-QAM modulation is executed on selected subcarriers calculating analytical capacity results of Rayleigh fading channel. Performance is increased by transmitter-side antenna selection and power control methods for antennas on each subcarrier.
   
Author: Albert MRÁZ
Budapest University of Technology and Economics (BME), Dept. of Telecommunications (HIT), Hungary
  Back to the conference program

Title: An Instance-Based Approach to Indoor Positioning
(Poster)
   
Abstract: With the extension of ubiquitous and pervasive computing there has been an increasing need to capture the local information of users both outdoor and indoor. We present an algorithm enabling localization and tracking mobile clients in indoor environment. Our method uses signal strength information at APs and the location of client is estimated by an instance-based learning method which uses the last signal strength values of clients. We present empirical evaluation that demonstrate the ability of our method to estimate client location an average accuracy of 2 m using a relatively low number of AP as compared experiment area.
   
Author: István NAGY
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Improving the Quality of Unit Selection and HMM based Speech Synthesis
(Poster)
   
Abstract: The quality of Text-To-Speech (TTS) systems is judged on the basis of how successfully the generated synthetic speech imitates human speech. The intelligibility of synthetic speech is close to that of human speech in state-of-the-art TTS systems. These systems use various techniques (e.g. speech segment concatenation, unit selection, statistical parametric approaches). However, there seems to be a lack of variability in most speech synthesizers: they produce deterministically the same speech output for the same textual input, when it is repeatedly given to the system. This contradicts the variability of human speech (e.g. if you say “Good morning” a few times, the sound will always be different). It is difficult to generate proper prosody (pitch, stress and rhythm) resembling to human speech, thus machine-generated fundamental frequency contours are less rich than natural prosody. One direction in improving the quality of a unit selection speech synthesizer is to introduce the variability of human speech. Other ways for synthetic speech improvement include statistical approaches. In the last decade Hidden Markov Models (HMM) are successfully applied in Text-To-Speech synthesis. The HMMs learn the spectral, excitation and duration parameters of a speech corpora from one (speaker dependent training) or from various (adaptive training) speakers. In the second case a small speech corpus is enough to adapt the voice characteristics to this target speaker. The parameters are stored in HMM databases and with a simple voice decoder method they are transformed into high quality human sounding artificial voice.
   
Author: Géza NÉMETH
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Adaptive Sampling Protocol (ASP) for Wireless Sensor Networks
(Poster and Demo)
   
Abstract: One of the most important issues in a wireless sensor network is energy efficiency, in order to extend the lifetime of the network. Sleep scheduling solutions proved to be exceptionally effective strategies to achieve this goal. Numerous such algorithms have been proposed and examined, but virtually without any considerable support for dynamic event-driven systems. Dynamically occurring events, the constantly changing environment, or the presence of mobile elements in the architecture make existing sleep scheduling techniques far too rigid. In such environments dynamic adaptation can be obtained through learning. In this work we address the problem of energy efficient sensing by adaptively predicting the measured samples within a user specified error bound, which enables sleep scheduling even in a dynamically changing environment. This functionality is incorporated in our Adaptive Sampling Protocol (ASP), and is called the Local Extrapolation function (ASP-LE). Experimental evaluations on traces of a medium size network with temperature and light sensors indicate that the ASP-LE function can extend the lifetime of the network by a factor of 2 or higher, even for a strict error target.
   
Author: Gergely ÖLLÖS
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Accelerometer Based Recognition of Everyday Activites
(Poster)
   
Abstract: We present an accelerometer based activity recognition system for mobile phones with a special application for personal time management. We describe the user interface which was designed to easily collect data about personal activities and compare several data mining algorithms for the automatic recognition task. The results show that SVM has the highest accuracy, but all methods have the recognition rate around 90%. It is an excellent result considering that there are 10 different activities to be recognized. As this system is mainly intended for personal time management, there is no need for perfect recognition, because its goal is only to track important trends and provide overall statistics. We believe that this application can give important insights to our personal lives and help us manage better the everyday tasks that we face.
   
Author: Zoltán PREKOPCSÁK
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Identity-based Security Architecture for VANETs
(Poster)
   
Abstract: Vehicular ad hoc networks have attracted extensive attention in recent years, which may be due to the fact that they have become the most global form of ad hoc networks. Nevertheless, security and privacy are critical factors of vehicular networks and remain a challenge nowadays. We propose a novel practical, scalable and certificateless identity-based security architecture, which is designed for VANET systems. We also describe a set of security protocols, which rely on the proposed architecture and include batch signature verification, authenticated key exchange and service request. In addition, it can preserve security and privacy properties: authentication, confidentiality, non-repudiation and message integrity, but excludes the need for certificate transmission and storage. We show that the proposed architecture is scalable in the way that arbitrary number of domains can be defined and enables vehicles and base stations to apply mutual authentication fast and securely, even if they have never communicated before.
We compare three schemes in terms of verification delay and transmission overhead: a signature algorithm described in the IEEE1609.2 standard, a non-hierarchical scheme and our proposed scheme. We show that the proposed scheme needs less computation time than the algorithm described in the standard and it is at least as efficient as the non-hierarchical scheme. It generates a bit more overhead than the non-hierarchical scheme, but it is still better than the standardized solution.
Securing vehicular networks is a hard problem nowadays, due to the numerous security conditions (including the balance between security and privacy) and the numerous and diverse applications. Designing secure VANETs is not only a challenge for network designers, but it includes legal, economical and social considerations as well, which will raise a unique mixture of security and administration requirements in the future.
   
Author: Attila SZENTGYÖRGYI
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: On the Impact of Per-Flow Throughput Using Fountain Code Based FEC in Lossy Environments
(Poster)
   
Abstract: One of the most important functions of today’s Internet is congestion control accomplished by mostly TCP. However the several problems of current TCP resulted in a number of new TCP proposals to cope with high speed and lossy environments to achieve efficient control. One of the possible alternatives in future Internet is to avoid any kind of congestion control. Moreover, the basic principle is that applications always send their data as fast as they can whenever they have something to transfer. If congestion does not occur in this scenario, then clearly it is the best solution. However, due to the maximal rate sending excessive packet losses are likely to occur. To cope with packet loss, efficient erasure coding based on Fountain Codes can be applied which seems to suit perfectly for this scenario. The present work shows the fundamentals of practical Fountain Codes and presents preliminary results investigating the effects of different lossy environments on the Fountain Code based forward error correction scheme. Results are presented for an independent and identically distributed Bernoulli error model and also for burst-loss environments according to the Gilbert model.
   
Author: András TEMESVÁRY
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Interactive blackboard and gesture recognition
(Demo)
   
Abstract: This demo presents a simple solution to extend the functionalities of ordinary projectors to serve as intelligent electronic blackboards that can recognize gestures made by the user and control various electronic devices like lamps, heaters, etc. in the office. The system allows the user, standing directly in front of the projected screen, with a light pen in his hands, to interactively communicate with the software, using the pen as the pointing device, similarly to a mouse. One of the main components of the architecture is the infra camera that can detect the two dimensional position on the projected screen of a pen equipped with an infra source. If the user turns on the infra pen and makes some gestures, the system can identify and act on them. Several actions, such as starting a presentation or turning on the lights, can be associated with a gesture. For this purpose, we have chosen to use a Wiimote as an infra camera, due to its low price and high availability. There are similar systems available on the market, but those systems usually do not support gesture detection outside of the plane of the blackboard. In our case the user can choose between two modes, namely mouse mode and gesture detection. In the mouse mode the user standing in front of the blackboard can interact with the graphical user interface provided by the operating system but cannot use gestures to control electronic devices. In the gesture detection mode the mouse function is not available but the user can control electronic devices or generate events for the operating system. The gestures can be personalized to meet the specific requirement of the user. The user can delete or add new gestures, associate new functions or change the associations of gestures to events, or just simply use the built in gestures and their associated functions. This application was developed as a project of the Mobile Innovation Center.
   
Author: Zsolt TIHANYI
Budapest University of Technology and Economics (BME), Dept. of Telecommunications and Media Informatics (TMIT), Hungary
  Back to the conference program

Title: Enhancement of GUIs: Audio Extensions for Blind Users for a Better Human-Computer Interaction
(Poster)
   
Abstract: Human-Computer interaction is usually based on a graphical user interface (GUI) and some auditory warnings. Blind users do not benefit from GUIs, therfore, a sound-based extension is being developed to enhance accessibility.
In a framework called GUIB (Graphical User Interface of Blind Persons) a collection of special sound samples - such as auditory icons, earcons, spearcons, auditory emoticons - were developed and based on the target group's opinion they were categorized, selected and evaluted.
he most frequently used programs, icons, functions and events of a computer screen were extended or replaced by auditory events. Sounds of everday life, specially compressed and speeded up speech samples are there to link visual representations to auditory events. Instead of speech, these signals are language independent, easy to learn, fast to use and sometimes even funny and amusing. On the other hand, spearcons (speech-based earcons) can be created easily for different languages using MATLAB codes (current database includes Hungarian, English and various German versions).
Results show that blind users welcome this idea and sighted users may also benefit from having auditory feedbacks and warnings. Future works includes implementation of the sound database under text-to-speech applications (e.g. J.A.W.S.) or the Windows OS.
   
Author: György WERSÉNYI
Széchenyi István University, Hungary
  Back to the conference program