Traditionally, network and system management are manually controlled processes. It usually takes one or more human operators to manage all aspects of a dynamically evolving computing system. Several trends shaping the development of IT infrastructures have aggravated network and system management. This leads to the question of how we can administer systems of this complexity adequately. This question has stimulated a large interest in self-management technologies - technologies that help systems autonomously control themselves. It's not realistic for human operators to maintain control over a system that consists of thousands of networked computers, mobile clients, and numerous application servers and databases. We must redefine human operators' roles so that instead of being involved in the decision process in an interactive and tightly coupled fashion, operators define general goals and policies for system control. So self-management is the solution only if we can first solve several open problems.

The provisioning of ambient services is gaining importance as users become more and more embedded in environments that are saturated with electronic devices. In our previous work, we have proposed the ad hoc service grid (ASG) approach as a means for deliberately setting up an infrastructure for providing ambient services at medium-sized locations like shopping malls. In this paper, we introduce a self-repairing lookup service architecture that is able to handle dynamic services in an AGS. These services can autonomously replicate and migrate within an ASG to optimize resource usage and message latency. Our lookup service is able to cope with the inconsistencies caused by service migrations efficiently by applying a lazy, request-driven update protocol. We discuss the architecture and the protocols employed by the lookup service. Moreover, we provide experimental results that show the efficiency and effectiveness of our approach

Providing an up-to-date view on dynamically changing data presents difficulties. While several different data-dissemination platforms address the problem, client mobility is still a major challenge. Volatile wireless connections and resource limitations void the attempt to apply the same mechanisms that work in fixed-wired networks. The Rebeca publish-subscribe middleware implementation introduces several concepts for accommodating client mobility in publish-subscribe systems. We present experimental results that show how intelligent filtering mechanisms can save network bandwidth inside the infrastructure and at mobile clients while preserving the accuracy of client notifications.

The maturing domain of ad hoc networking still holds some major challenges on its way to commercial exploitation. The dynamic and decentralized nature of ad hoc networks calls for new paradigms in the design of distributed applications. Most notably, there is a need for self-organization mechanisms in the middleware layer to enable medium-scale and large-scale distributed applications. In this paper we present NESHMd1, a middleware platform for ad hoc networks that is built on mobile agents and tuple space technology. MESHMd1 provides basic building blocks for the implementation of self-organizing distributed applications for ad hoc networks. Unlike other middleware platforms, MESHMd1 puts a high emphasis on mobility awareness, decoupling and bio-inspired coordination primitives. We discuss the aspects of ad hoc networking and self-organization and derive the basic architecture of MESHMd1 before we present our prototype implementation.

Mobile ad hoc networks provide a communication environment that is characterized by dynamic changes in the topology and in the availability of resources. The limited transmission range of mobile devices and the mobility of the users make it difficult to transfer wellknown mechanisms for the organization of distributed systems to the ad hoc networking domain. Therefore, there is a need for self-organization mechanisms that help applications in coping with these effects. So far, the research on self-organization in ad hoc networks has not resulted in generally applicable mechanisms. In this paper we present our vision of socio-awareness that exploits recent advances in the research on complex social networks. We propose a system that is able to capture the patterns of interactions between mobile ad hoc users and uses them to organize, for example, spontaneous information exchange and service deployment in mobile ad hoc networks.

The mobile agent paradigm can be employed effectively for the decentralized management of complex networks. Nevertheless, legacy management protocols such as SNMP need to be supported for reasons of migration and efficiency. We show how the integration of the two paradigms leads to a flexibly adjustable decentralized management architecture. NetDoctor is a network management application framework that allows the monitoring of hosts in a heterogeneous network. Mobile agents may query local or remote SNMP agents, process the status data, and perform certain corrective actions. The system supports run-time adjustment and reconfiguration. A quantitative analysis underlines the benefits in respect to the network load reduction.