A cooperative Hide and Seek discovery over In Network Management

Cooperative Learning for Distributed In-Network Traffic Classification

“…The issues of nodes discovery for interaction among nodes is presented in [9], network nodes discover one another in a network in other to cooperate with each other. In the same note, reference [4] proposed bootstrapping and discovery mechanisms to ensure the proper information dissemination in distributed network process. However, these methods are proposed for discovering neighbouring nodes, they did not consider the task of nodes interaction for information sharing.…”

“…Each DNM entity participates in a distributed management process, which requires cooperation among DNM entities to monitor, analyze and make decision to achieve global network objectives. Cooperative Learning (CL) enables network nodes to discover each other, exchange information, disseminate local decision, enhance self-adaptation of nodes, improve scalability and finally enforce management decision [4].…”

Traffic engineering supported by Inherent Network Management: analysis of resource efficiency and cost saving potential

“…Gossiping and P2P approaches can deal with dynamic networks subject to a high churn rate; aggregation and evaluation of information in management overlays, from simple functions, e.g. a sum or a minimum to more complex evaluations of network‐wide threshold crossings [16,22]; estimation of the size of networks or of groups with specific characteristics in the network at minimum messaging overhead [19], as a basic function to be embedded in higher‐layer evaluations and applications; efficient topology discovery using a new hide and seek principle for scalable information propagation and synchronization [29]; and search methods for highly dynamic networks using combinations of flooding and random walks, including partially available information about the path to a target [20]. …”

“…efficient topology discovery using a new hide and seek principle for scalable information propagation and synchronization [29]; and…”

On the investigation of the joint use of self-* properties and Peer-To-Peer for network management

“…The same way, MAS solutions enable a more distributed network management infrastructure, where agents (or intelligent agents) are spread along the network to be managed. Most of the proposals of MAS-based network management solutions claim that the agents cooperate by sub-tasking/splitting/replicating the management task among multiple agents, or by gathering information from other agents for individual deliberation of management actions (LI et al, 2001) (AKASHI et al, 2005) (TERAUCHI; AKASHI, 2009) (GUARDALBEN et al, 2010). This way, most of the P2P and MAS-based network management applications still present a master-slave behavior, which is a simple and tending to be centralized style of defining an application.…”