AD-HOC NETWORKING

In the project, we will investigate architectures and technologies for mobile, broadband, ad-hoc networks as well as particular technical challenges in terms of self-structuring and quality-of-service (QoS) provisioning. The major innovative aspect related to this project is the investigation of the general applicability of the algorithms for network self-structuring and automatic partitioning, which is a feature missing from the existing proposals for practical networks, such as IP or ATM. To achieve this, a general, hybrid network architecture will be proposed, where mobile and stationary nodes are combined in a dynamic and more efficient fashion. We will investigate the effects of the active network partitioning on the particular requirements on QoS delivery and routing. More precisely, we will concentrate on the graph theoretical abstraction of the network state information and couple the problem of network organisation and optimal routing, the solution of which will be the original contribution of this work. Finally, we will evaluate the applicability of proposed algorithms by developing an extensive simulation environment.

The project aimed at defining control protocols and services that combine the features of mobile and stationary network architectures, and are used to build a hierarchical, self-structuring network. The main goal was to study the network architecture that can support automatic network configuration, with the ultimate vision to enable the next generation of flexible, heterogeneous networks. Particular emphasis was on clustering concepts and on its applicability to different problems in network self-configuration. To this aim, the research team developed control architecture for wireline transport networks, applied new methods of resource management for ad-hoc wireless networks, developed hybrid routing protocol unifying the features of the proactive and reactive routing for data aggregation in sensor networks. For the simulations and results, the widely adopted tool OMNET++ was used and specific modules have been added to address the network scenarios developed in this project. The OMNET++ based simulation framework was proven to be modular and extensible, and the new networking environments were integrated, implemented and tested. With help of the simulation tool, the research team created and analyzed two main networking environments: (i) self-configurable wireline transport networks and (ii) wireless ad-hoc/sensor networks. For self-configurable wireline transport networks, the research team used clustering to design the partitioned network control planes. For wireless ad-hoc networks, clustering was used for resource management and routing. In particular, the research team focused at stable generalized clustering techniques, i.e. the techniques that introduce the smallest amount of changes in clustering structure as a reaction to network state changes. The research team also developed smart clustering techniques in sensor network scenarios, where clustering was applied for different purposes, such as to exploit node redundancy, to achieve energy efficient network operation, and to perform data aggregation and load reduction in the network. By combining wireline and wireless networking scenarios, this research produced unique and significant results that can be used to design future heterogeneous networks, such as wireless infrastructure of sensors and devices supported by wireline infrastructure that connects sensor data processing centres. The results of the research in this project were presented at the recognized international conferences and they attracted significant attention.