Overview

Research Topics

Research Projects

Infrastructure

Scientific Contributions

Research Prototypes

Autonomic & Grid Computing Group (AGC)

Research Infrastructure

Autonomic and Intelligent Systems Laboratory (Smart Room)
A unique state of the art laboratory has been set up, supporting its perceptual interfaces, multimodal interactions and autonomic systems middleware research. The goal of this laboratory is to promote understanding, education and research in the areas of Autonomic and Intelligent Systems combining: self-management regardless of device, multimedia and multimodal human-machine interaction, context aware middleware, novel applications, and intelligent, self-managed networks. The laboratory aims to produce results constituting steps towards what the European Commission calls “Ambient intelligence” as an extension and improvement of “Ubiquitous Computing”. This technological direction puts humans at the centre of a self-managed, multimedia and multimodal computing infrastructure that aims at satisfying human needs for work, pleasure and improved quality of living, without requiring special training or knowledge.

The laboratory consists of the following elements:

• 6 video cameras (SONY) – one being Pan-Zoom-Tilt
•  4 high resolution (1600x1200) video cameras (ids-imaging)
•  One panoramic (fish-eye) camera
•  MarkIII Microphone Array (64 microphones)
•  Projection devices
• Small computer mounted cameras
• Microphone clusters
•  6 LINUX twin processor INTEL HT computers.
• 4 LINUX Intel Core 2 Duo computers
• One Targeted Audio Device

• Used for transmitting personal messages, directly to someone in the room.
• Only the recipient is able to hear the message transmitted from the Targeted Audio Device
• It is created with the collaboration of Daimler-Chrysler

The topology of the previous components is shown graphically in the next top view of our Smart Room.
 
System software that is currently installed, configured and used in this laboratory includes distributed multimedia streaming (e.g., NIST SmartFlow), camera and sensor control and a host of state of the art perceptual technologies. Examples of such technologies are audio-visual speaker recognition, face recognition, gesture recognition, automatic classification of documents, and context aware working spaces.

RFID Testbed
The Groups operates an RFID testbed comprising an RFID dock-door portal with multiple readers and antennas. Specifically, the dock-door portal is built based on Intermec and Impinj readers. Furthermore, 5 antennas are available to enable a variety of physical readers configurations. The group has developed fully fledged RFID middleware platforms (compliant to the EPC global architecture) for building and testing RFID solutions end-to-end (i.e. from the physical readers to the enterprise applications such as ERPs (Enterprise Resource Planning), WMS (Warehouse Management Systems) and corporate databases).
The RFID testbed includes the following components:

• Two RFID readers (Impinj Speedway & Intermec IF5)
• Three Intermec, two Impinj far-field and one Impinj near-field RFID antennas

One RFID gate loaded with the above components

Grid Testbed
The Group operates a small Grid Programming Testbed based on the Open Grid Services Architecture (OGSA). The testbed comprises 5 Sun Solaris (version 8) and 1 Linux machines  (within the AIT’s intranet), each one hosting the Globus toolkit (version 3.2.1 (GT3),)  Experiments based on Grid programming are carried out on the testbed.

Software Laboratory
The AGC group also makes extensive use of AIT’s Software Laboratory, which is a state-of-the-art, powerful computational, network and laboratory infrastructure. Indicatively, is comprise computational power of a SUN cluster consisting of 4 SUN Twin-Processor servers (upgradeable up to 8CPU’s per server), 9 SUN Blades and 25 SUN Ray terminals.  The network supporting the SUN cluster is 100Mb full duplex forming a powerful distributed computational system.
A variety of applications and educational programs, as well as software development tools, simulation tools and telecommunication system design tools are available. Indicatively, we list some of the available software packages:

• Matlab: Software package for the development, design, and simulation of Digital Signal Processing and Communication Systems (12 licenses)
• OPNETQ Software package for the simulation of data networks. (16 licences)
• NS: Open Software for data network simulation – a standard academic tool
• Rational software: Analysis and design of software systems incorporating the latest technologies and methodologies (e.g. object oriented programming).
• Oracle RDBMS: The full edition of the popular Database software (version 9) (10 licenses)
• Software simulators: Open Source for a variety of complex applications like for example Operating System design and experimentation
• Complete software development environment for Java, J2EE (e.g. SUN NetBeans, Eclipse), C, C++, versioning.
• Open source web servers and application servers.
   

Business and Research Collaborations - Consulting Agreements
On-line Interactive TV program Guide – MovieTalk.
In order to demonstrate the improvement in user experience, when multi-modal interfaces are used, IBM and RESIT/INTRACOM have collaborated on the development of a speech-enabled interactive TV system. More specifically, IBM has provided its embedded speech recognition technology (in addition to typical remote controls and keyboards) as an interface to INTRACOM’s end-to-end digital content distribution platform. In other words the user can simply use his voice to search for multimedia content from the provider's database. By investing on the force of the standalone systems we have implemented the first parts towards the total integration of these separate technologies. The main advantage of the system is its ability to deal with speech commands that require the syntax of complicated queries. The system thus copes not only with simple serial browsing but also with dynamic requests. This allows for efficient searching through vast amount of multimedia content, by simply specifying a title, an artist or some other similar property. Incorporating the same logic to all the components of the existing Electronic Programming Guide (EPG) could ideally allow for production of STBs that will take the entertainment interfaces to a new level of usability while in parallel minimizing the need for the typical type of input devices like keyboards and remote controllers.

Daimler – Chrysler collaboration on the Targeted Audio device.
The AGC group of AIT is collaborating with Daimler-Chrysler on improving the directionality of the Targeted Audio device using Digital Signal Processing techniques.

The Daimler-Chrysler directed audio donated to AIT is shown in the picture. It is steerable up-down and left-right, thus allowing the arrays of loudspeakers to turn to the desired direction. The basic principle of operation is that an ultrasound wave carrier (40KHz t 80KHz) is amplitude modulated with the audible signal that we would like to “project” to a particular direction. The air operates as a demodulator because of its non-linear nature, dissipating the high frequency faster than the lower- audible frequencies. Thus an audible lobe is created at a specific distance which also dissipates fast afterwards. The effect to the listener is quite impressive: as if the voice-of-God is heard coming from a particular direction in thin air. The system works best with absorbing surfaces on the walls and on the floor in order to minimize reflections.
AIT’s research work with Daimler-Chrysler will focus on the adaptation of the signal processing of the emitted signals, including recalculation of the phases and amplification/attenuation of the low-frequency audible signal in order to improve directionality when turning the device to various directions. The following figures show the directivity of the device based on analysis by Daimler-Chrysler.