Tom Mahieu

Self-adapting concurrency: the DMonA architecture

A major problem in todays Internet servers is that they suffer from extreme peak loads. Traditional (operating) systems are designed to perform extremely well under heavy load conditions. However, it is not feasible to over-provision resources only to support peak loads. A key factor to deal with such peak loads is internal concurrency control. We have developed a component based architecture (DMonA), which allows to adapt internal concurrency according to measured throughput. Performance tests show that DMonA outperforms traditional approaches, while it is still very manageable thanks to the underlying DiPS component architecture.

Connecting embedded devices using a component platform for adaptable protocol stacks

Research domains such as sensor networks, ad-hoc networks, and pervasive computing, clearly illustrate that computer networks have become more complex and dynamic. This complexity is mainly introduced by unpredictable and varying network link characteristics, heterogeneous capabilities of attached nodes, and the increasing user expectations regarding reliability and quality of service. In order to deal with this complexity and dynamism of computer networks, the system’s protocol stack must be able to adapt itself at runtime. Yet, to handle this complex challenge effectively and efficiently, we claim that it is essential for protocol stacks to be developed with run-time adaptability in mind. This chapter presents a software architecture tailored to build highly adaptable protocol stacks, along with a component platform that enforces this architecture. Although the presented software architecture focuses on protocol stacks in general, we zoom in on the application of its founding principles in the domain of embedded network devices.

A Session Layer Concept for Overlay Networks

Current day mobility solutions are characterized by a number of essential restrictions that prevent them from being successful in an overlay network environment. This paper proposes 4 properties that mobility solutions must have to address these restrictions: application cooperation, higher layer mobility awareness, general applicability and protocol heterogeneity. Subsequently, a mobility solution that realizes all these properties is described. The solution introduces a session layer in the protocol stack that consists of two subsystems: the connection abstraction system and the address management system. Other mobility solutions found in the literature, session layer and others, typically only realize a subset of the four properties.