W-Grid: A scalable and efficient self-organizing infrastructure for multi-dimensional data management, querying and routing in wireless data-centric sensor networks

摘要

Data-centric sensor networks are advanced ad hoc networks that act like a distributed database managing and indexing sensed data in order to efficiently perform advanced in-network tasks, such as routings, searches, data processing, fusion and analysis. The supplied distributed services, such as routing, content location and information sharing should be provided anywhere and at any time optimizing energy consumptions, computational resources, memory occupation and radio transmissions. Moreover, the network traffic should be equally balanced among participants in order to avoid premature discharge of some devices that may partition the network. This work describes a fully decentralized infrastructure able to self-organize nodes in ad hoc networks by exploiting local interactions and topology learning among devices. In this solution all nodes are peers and nothing prevent the approach to be used in wireless mesh networks as well. Differently from existing solutions, our proposal does not require global information or external help, such as the Global Positioning System, which works only outdoor with a precision and an efficacy both limited by weather conditions and obstacles. The infrastructure natively enables devices to perform routing and data management without using message broadcast/flooding operations. The work introduces also a feature, called full learning, that improves routing performances while balancing the traffic among devices. We report an extensive number of simulations comparing the new solution results with four existing proposals, two of which deriving from preceding versions of the infrastructure.