Large-scale decentralized systems organized in overlay networks are complex to manage. Such systems integrate organizational complexity in the application-level resulting in low abstraction and modularity in their services. This thesis introduces a multi-level conceptual architecture for overlay services. An overlay service is a large-scale decentralized system organized in overlay networks that provides generic application capabilities. Two overlay services introduced in this thesis provide a proof-of-concept for the higher abstraction and modularity that can be achieved with this architecture. The first overlay service builds and maintains overlay networks that are self-organized in different tree topologies according to different application criteria. The second overlay service performs a function-indepencent and routing-independent decentralized aggregation of values that are distributed in a network and continuously changing. Experimental evaluation studies the performance trade-offs imposed by this generic design. This thesis studies overlay services in the application domain of the Smart Power Grid. More specifically, overlay services are used in demand-side energy management to perform (i) load-shifting and (ii) demand-adjustment in a fully decentralized fashion. The evaluation of this approach is grounded in the current reality and practice of the Smart Power Grid.

Multi-level Reconfigurable Self-organization in Overlay Services

Distributed management of complex large-scale infrastructures, such as power distribution systems, is challenging. Sustainability of these systems can be achieved by enabling stabilisation in global resource utilisation. This paper proposes the Energy Plan Overlay Self-stabilisation system (EPOS), for this purpose. EPOS is an agent-based approach that performs self-stabilisation over a tree overlay, as an instance of a hierarchical virtual organisation. The global goal of stabilisation emerges through local knowledge, local decisions and local interactions among software agents organised in a tree. Two fitness functions are proposed to stabilise global resource utilisation. The first proactively keeps deviations minimised and the second reactively reverses deviations. Extensive experimentation reveals that EPOS outperforms a system that utilises resources in a greedy manner. Finally, this paper also investigates and evaluates factors that influence the effectiveness of EPOS.

/pdf/local-agent-based-self-stabilisation-in-global-resource-3xew63aths.pdf

Local agent-based self-stabilisation in global resource utilisation

Query driven Broadcast through wireless sensor nodes also leads to the domain of event driven converge cast. A query-response based application in Wireless Sensor Networks(WSN) demands the correct delivery of data message at each sensor node. A Breadth-First Search(BFS) tree rooted at the base station offers shortest path traversal for each data message which utilizes the sensor resources efficiently. Resource constrained sensor nodes are highly prone to sudden crash. So the application demands a quick and smart approach to repair the tree when a node dies. In this paper a novel scheme has been proposed to locally repair the tree with constant round of message transmissions. Each node piggybacks a few bytes of extra information along with each query and response messages. Based on these piggybacked values each node calculates its alternate parent. When a parent node fails, its children can contact their respective alternate parents immediately to establish an alternate path to the root. Reduced communication cost in terms of extra message transmissions saves battery power at each node. Efficient query-response message handler ensures the correct delivery of messages. Fast repairing offers good Quality of Service(QoS). Simulation result shows that no message is lost except the one holding by the crashed node.

A Tree-Based Local Repairing Approach for Increasing Lifetime of Query Driven WSN

Tree topologies are often deployed in large-scale distributed systems to structure a hierarchical communication. Building and maintaining overlay networks self-organized in tree topologies is challenging to achieve in dynamic environments. Performance trade-offs between resilience to failures and message overhead need to be considered. This paper introduces eight adaptation strategies that provide a higher abstraction, modularity and reconfigurability in the tree self- organization process. Performance can be further enhanced by dynamically changing strate- gies during system runtime. Experimental evaluation illustrates the performance trade-offs and properties of adaptation strategies.

Adaptive Self-Organization in Distributed Tree Topologies

MatchTree: Flexible, scalable, and fault-tolerant wide-area resource discovery with distributed matchmaking and aggregation

Overlay multicast constructs a multicast delivery tree among end hosts. An important problem for making overlay multicast more dependable is how to recover from node departures in order to minimize the disruption of service to those affected nodes. In this paper, we propose a proactive tree recovery mechanism to make the overlay multicast resilient to these failures and unexpected events. The salient feature of the approach is that rescue plans for multiple non-leaf nodes can work together for their respective children when they fail or leave at the same time. Extensive simulations demonstrate that our proactive approach can recover from node departures much faster than reactive methods, while the quality of trees restored and the cost of recovery are reasonable.

Supriya Chaluvadi

Papers

A Proactive Tree Recovery Mechanism for Resilient Overlay Multicast