Zoltan Juhasz

JM: A Jini Framework for Global Computing

The number of Internet users and internet-connected computers are estimated to be in the order of hundred millions. Users are restricted to performmainly information retrieval (document search) and e-commerce related (shopping, online banking) tasks, while the majority of the computing resources are under-utilised. With suitable technology, it is possible to create a service-centred Internet, where users can perform more sophisticated (e.g. computing) tasks. We argue that grid research should explore the issues of creating very large systems of services available to many users, as opposed to HPC programs available to relatively few scientific users. This paper describes the architecture of a Jini based global computing environment, whose aim is to support the execution of sequential and parallel programs on a large scale. We describe the architecture, supported programming models of the system, as well as our current, ongoing work and future development plans.

MOST-NNG: an accessible GPS navigation application integrated into the mobile slate talker (MOST) for the blind

Over the recent years, GPS navigation has been attracting a growing attention among the visually impaired. This is because assistive technologies can obviously be based on commercially available solutions, as the GPS capable hand-held devices entered the size range of the ordinary mobile phones, and are available at an ever more affordable price, now providing a real choice for a wider audience. For many, an accessible GPS navigator has even become an indispensable tool, an integral part of their every-day life. Since the most appropriate (or at least the most favored) device type for GPS navigation is the so-called PDA, whose user interface is dominated by a touch screen and usually lacks any keyboard, accessibility for the blind remains an issue. This issue has successfully been tackled by the MOST-NNG project in which, the MObile Slate Talkers blind-friendly user interface has been combined with Hungarys leading iGO navigator.

Usability evaluation of the MOST mobile assistant (slattalker)

The goal of the MOST project is to develop a novel, inexpensive, easy-to-use digital talking device for blind and visually impaired users based on off-the-shelf handheld computers (Personal Digital Assistant). The device provides a novel user interface based on a simple menu system and Braille text input, and a range of application programs to support everyday tasks, including clock, notepad, phone and short messaging, email. This paper reports on the usability evaluation of the device, its strategy and implementation, and shows that our approach results in an easy to learn and use system with input speed comparable to sighted users

How Can Java Be Made Blind-Friendly

The widely used and highly popular Java programming language is proved to be a great tool for developing platform independent applications. Everyday users mostly encounter them when using portable devices (mobile phones, PDAs, etc). However, the ordinary Java applications are inaccessible for the blind in general. Even the most used screen readers can only be enabled to handle GUI elements of a Java application by an additional adaptation package (e.g. access bridge for Jaws). Even with this, only a portion of existing Java programs that use swing classes may be made partially accessible for the blind. The solution offered eliminates the need of any screen reader.

Teaching Morse Language to a Deaf-Blind Person for Reading and Writing SMS on an Ordinary Vibrating Smartphone

Deaf-blind people have a very small window to the world. New technology can help, but portable Braille lines are expensive. We developed and tested a very low cost method for reading and writing SMS messages with a Hungarian deaf-blind person using Android smartphone with vibrating motor built in. Words and characters were converted to vibrating Braille dots and Morse words. Morse was taught as code for recognizing characters and also as language for recognizing words.

Parallel program execution support in the JGrid system

Service-oriented grid systems will need to support a wide variety of sequential and parallel applications relying on interactive or batch execution in a dynamic environment. In this paper we describe the execution support that the JGrid system, a Jini-based grid infrastructure, provides for parallel programs.

The benefits of Java and Jini in the JGrid system

The Java language and platform have been considered by many as natural candidate for creating grid systems. The platform-independent runtime environment, safe and high-level language and its built-in support for networking and security are very valuable features. Despite its potential and the many proof-of-concept systems developed, the grid community is turning to Web services technology as its implementation base. In this paper, we show that Java, by joining forces with Jini technology can provide a very appealing technology base for highly dynamic grid systems. The key properties of Java and Jini technology are examined with reference to their role in grids. Then, the JGrid Jini-based service-oriented grid system is overviewed describing its key concepts, services and how it extends Jini to address some of the unique requirements of grid systems.

Supporting interactive computational science applications within the JGrid infrastructure

Future computational grid systems must support interactive and collaborative applications to help computational scientists in using the grid effectively. This paper shortly introduces how the JGrid system, a service-oriented dynamic grid framework, supports interactive grid applications. Via providing a high level service view of grid resources JGrid enables easy access to grid services and provides an effective programming model that allows developers to create dynamic grid applications effortlessly.

Highly parallel online bioelectrical signal processing on GPU architecture

Signal processing is of central importance in biomedical systems, in which pre-processing steps are unavoidable in order to reduce noise, remove unwanted artefacts, segment time series into smaller epochs, or extract statistical and other descriptive features that can be used in consecutive classification stages. The high sampling rates and electrode counts used e.g. in advanced EEG or body-surface potential mapping ECG systems, result in very large data sets, which require considerable time to process. While long computation times can be tolerated in research settings, in many application areas, e.g. brain computer interfaces, online feature detection (e.g. in epilepsy or heart monitoring) or clinical diagnostics, fast - often real-time - execution speed is required. This paper examines whether and how GPUs can be used efficiently to implement parallel signal processing algorithms. The performance critical factors of GPU programs used for real-time processing are identified then we show that high performance is achievable only if great attention is given to the hardware execution characteristics of the GPU. Overlapped data transfer and computation, optimized use of the GPU memory hierarchy, using enough threads to hide memory transactions are used to obtain better results than sequential CPU implementations.

A GPU-based simultaneous real-time EEG processing and visualization system for brain imaging applications

A data-driven prototype software is presented for EEG processing and visualization. The system relies on the GPU architecture for providing simultaneous processing and visualization of the EEG data. Two example brain imaging algorithms, the surface Laplacian and the spherical forward solution are used for illustrating the effective use of the massively parallel GPU hardware in speeding up computations. The paper describes the architecture of our system, the key design decisions, and the performance optimization of the parallel implementation. Using the CUDA-OpenGL interoperability, the computing subsystem can directly modify potential data in the OpenGL vertex memory, avoiding unnecessary GPU-Host data transfers. The system and our parallel implementations demonstrate that real-time processing and visualization is possible for a range of algorithms during EEG processing. We are confident that these results can pave the way for supercomputing-class implementations and open up new opportunities in the clinical practice and neuroscience research.