Research projects
Algorithms
Software Platform and Component Environment for yoU (Space4U)
Period: 7/2003 – 6/2005
Researchers: Patrik Floréen, Michael Przybilski, Teemu Kurppa.
Funding: Nokia Research Center
The EUREKA ITEA project Space4U (Software Platform and Component Environment for yoU) continues the work of the ITEA project ROBOCOP. ROBOCOP resulted in a component-based architecture for middleware for embedded appliances. The aim of Space4U is to enhance this architecture with power management, fault management and remote terminal management. HIIT/BRU acts as a subcontractor to Nokia Research Center in this project. The work is focused on the parts of the Space4U project connected with terminal management and context-dependent configuration. This part of the project develops the selection, transferral and execution of context-based software components in terminals with limited resources.
The project was divided into two phases; t t he first one, “Terminal Software Management System Design/Development,” started in July 2003 and ended in June 2004. The second phase, “Terminal Management Demonstrator Development,” started in July 2004 and ended in June 2005. The The project activities in 2005 included implementation of the new version of the Robocop Runtime Environment (RRE), participation in the final specification of the extended software framework and description of extended methods and technologies, development of demonstrators and demonstration at the ITEA Symposium in October 2005.
Trust4All
Period: 10/2005-09/2006
Researchers: Patrik Floréen, Michael Przybilski
Funding: Nokia Research Center
In the ITEA project Trust4All, we work on context-aware adaptations of trustworthy systems, in particular with regard to its dependability and security properties. The project develops a middleware software architecture for embedded systems that require a defined level of trust, due to the nature of the services they provide.
To enable this adaptation we use the frameworks and mechanisms developed in the preceding projects ROBOCOP and Space4U . The ROBOCOP project defined and implemented a middleware architecture that is especially suited for mobile devices and PDAs. The Space4U project extended this middleware with fault prevention, power management and terminal management, including secure downloading.
The project is a joint effort of several European universities and companies. The research project at HIIT/BRU is a subcontract from Nokia Research Center .
The project started with studying architecture requirements and a trust framework, as well as demonstrator use cases and requirements.
Mobile Life – MobiLife
Period: 9/2004 – 12/2006
Researchers: Patrik Floréen, Petteri Nurmi, Michael Przybilski, Jukka Suomela, Cilla Björkqvist, Fredrik Boström, Eemil Lagerspetz
Funding: EU
The objective of the MobiLife Integrated Project is to bring advances in mobile applications and services within the reach of users in their everyday life. The MobiLife project, with 22 partners in nine countries, is coordinated by Nokia. It is part of a larger set of EU projects called Wireless World Initiative (WWI). The group at HIIT / BRU focuses on the context-awareness aspects of the project. We WeWe research context reasoning methods, as well as context-aware software architectures. HIIT/BRU is task leader of the task Context Management.
In 2005, the group worked on the design and specification of the Context Management Framework. The group's focus is on context reasoning. We have designed and implemented a geneic reasoning component that allows the use of different reasoning mechanisms. For this end, we have implemented a reasoning mechanism that uses Bayesian classifiers. We have also worked on building cross-platform data gathering tools, studied the automatic recognition of socially meaningful groups from context data, and promoted the integration of the MobiLife architecture into the general WWI system architecture.
Project: Networking and Architecture for Proactive Systems - Algorithmics (NAPS)
Period: 1/2003-12/2005
Researchers: Patrik Floréen, Petteri Nurmi, Jukka Suomela, Jukka Kohonen, Fredrik Boström, Marja Hassinen, Joonas Kukkonen, Jouni Sirén
Funding: The Academy of Finland, the PROACT research programme
This HIIT/BRU project is part of the NAPS consortium (Networking and Architecture for Proactive Systems), to which belongs also the research groups of Professor Pekka Orponen (Helsinki University of Technology, Laboratory for Theoretical Computer Science) and Professor Jorma Virtamo (Helsinki University of Technology, Networking Laboratory). The consortium is part of the research programme Proactive Computing (PROACT) of the Academy of Finland . The network computing and communication models underlying proactive applications give rise to new opportunities and challenges in the fields of algorithm design and analysis.
During 2005, the project continued research on energy-efficient designs in sensor networks. We have studied balanced data gathering, where the task is to maximise the total amount of data received at the sink, while simultaneously ensuring that a minimum amount of data is forwarded from all sensors. We have presented a new approximation algorithm for finding an optimal routing in this setting [1] . We have also studied the possibility of further improving balanced data gathering by adding a number of new relay nodes [2]. Our earlier research on lifetime maximisation in ad-hoc networks was presented in a journal article [3].
References: [1] P. Floréen, P. Kaski, J. Kohonen, P. Orponen: "Lifetime maximization for multicasting in energy-constrained wireless networks". IEEE Journal on Selected Areas in Communications 23 (2005), 117-126. [2] J. Suomela: "Computational Complexity of Relay Placement in Sensor Networks." Proc. 32nd Conference on Current Trends in Theory and Practice of Computer Science (SOFSEM 2006, Merin , Czech Republic , Jan 2006), Lecture Notes in Computer Science 3831. Springer-Verlag , Berlin , 2006, 521-529. [3] P. Floréen, P. Kaski, J. Kohonen, P. Orponen: "Exact and approximate balanced data gathering in energy-constrained sensor networks." Theoretical Computer Science 344 (2005), 30-46.
Generic Library of Algorithms on Strings (GLAS)
Period: 1/2004-12/2007
Researchers: Juha Kärkkäinen, Jarkko Toivonen. Esko Ukkonen
Funding: The Academy of Finland
The aim of this project is to design and implement a generic program library of string algorithms and data structures. Implementations can be used in fields like computational biology and information retrieval that process string data.
In 2005, the focus of the project was on the design of the basic architecture for the library and the implementation of core features. {0> Arkkitehtuuria käsittelevä artikkeli ilmestyi vuoden aikana. <}0{> An article on the architecture was published during this year. A corresponding library project has been carried out at Freie Universität in Berlin , and the group started a cooperation with them to combine the two libraries.
Approximation and learning algorithms (ALEA)
Period: 1/2005-12/2008
Researchers: Ilkka Autio, Jyrki Kivinen
Funding: The Academy of Finland
Approximation algorithms are methods that search for approximate, non-optimal solutions to computational problems. The goal is to save computation resources in problems that are computationally hard to solve exactly.
The basic method of machine learning is to form a hypothesis on the basis of given sample data. The hypothesis explains the examples and is as simple as possible. It often seems natural to use approximation algorithms for this, since forming a simple hypothesis can be difficult computationally, and there is often no need to explain all examples in detail; the examples may contain errors that the learning algorithm is not even supposed to track.
The project studies approximation algorithms, especially for the needs of machine learning. One sub-goal is to modify traditional approximation algorithms based on worst-case scenarios so that they may better utilise easy cases that often occur in reality. Another approach is the development of models and methods of machine learning where an approximation problem that is known to be hard is avoided either by changing the representation of a hypothesis or by formalising the learning problem in a completely different way.