Autonomic network management for wireless mesh and MANETs

Abstract

The realization of the envisioned ubiquitous computing paragon has resulted into new breeds of hybrid networks, e.g., u-Zone [1]. In u-Zones, a high-speed wireless mesh is used as the backbone-allowing MANETs nodes to be connected as a stub network. Hybrid in their nature, these networks belong to the generation of networks that deal with high levels of heterogeneity, mobility, and variability. The existing network management architectures address either MANETs [2] or mesh [3] in isolation, and do not meet the dynamic paradigm and fluctuating requirements for u-Zone networks. Highly dynamic node and network configurations, management and network tasks excessively overlapping, mobility, unpredictable environment, and lack of centralized control make continuous human-supervised management almost impossible. The catering of such challenges, plus the continuous growth factor, demand an autonomous management architecture that facilitates all selfmanagement aspects. The challenges of dynamic configuration, on-the-fly resource allocation, and reliable service provisioning can hardly be met without embedding a robust context-aware mechanism within network management framework. Our contribution is to provide an autonomie, policy-based, context-aware and hierarchical manager-agent framework that is adaptable and robust to network variations, node failures, and dynamic user requirements. The policy-based framework provides flexibility and scalability by enabling the network operators to specify the high level networking requirements, in terms of configuration, healing, accounting, optimization and general policies. The context information is gathered, analyzed and synthesized to predict the current and future network behavior. Correspondingly, required management functions are identified, invoked, and executed through the mobile agents, on the managed nodes. The system parameters optimize themselves by estimating new transitions and reacting accordingly, with the least human intervention. © Springer-Verlag Berlin Heidelberg 2006.