Computing Frequent Elements Using Gossip
نویسندگان
چکیده
We present algorithms for identifying frequently occurring elements in a large distributed data set using gossip. Our algorithms do not rely on any central control, or on an underlying network structure, such as a spanning tree. Instead, nodes repeatedly select a random partner and exchange data with the partner – if this process continues for a (short) period of time, the desired results are computed, with probabilistic guarantees on the accuracy. Our algorithm for frequent elements is built by layering a novel small space “sketch” of data over a gossip-based data dissemination mechanism. We prove that the algorithm converges to the approximate frequent elements with high probability, and provide bounds on the time till convergence. To our knowledge, this is the first work on computing frequent elements using gossip.
منابع مشابه
DisTriB: Distributed Trust Management Model Based on Gossip Learning and Bayesian Networks in Collaborative Computing Systems
The interactions among peers in Peer-to-Peer systems as a distributed collaborative system are based on asynchronous and unreliable communications. Trust is an essential and facilitating component in these interactions specially in such uncertain environments. Various attacks are possible due to large-scale nature and openness of these systems that affects the trust. Peers has not enough inform...
متن کاملDisTriB: Distributed Trust Management Model Based on Gossip Learning and Bayesian Networks in Collaborative Computing Systems
The interactions among peers in Peer-to-Peer systems as a distributed collaborative system are based on asynchronous and unreliable communications. Trust is an essential and facilitating component in these interactions specially in such uncertain environments. Various attacks are possible due to large-scale nature and openness of these systems that affects the trust. Peers has not enough inform...
متن کاملIdentifying frequent items in a network using gossip
We present algorithms for identifying frequently occurring items in a large distributed data set. Our algorithms use gossip as the underlying communication mechanism, and do not rely on any central control, nor on an underlying network structure, such as a spanning tree. Instead, nodes repeatedly select a random partner and exchange data with that partner. If this process continues for a (short...
متن کاملEfficient Decentralized Nonlinear Approximation via Selective Gossip
Recently, gossip algorithms have received much attention from the wireless sensor network community due to their simplicity, scalability and robustness. Motivated by applications such as compression and distributed transform coding, we propose a new gossip algorithm called Selective Gossip. Unlike the traditional randomized gossip which computes the average of scalar values, we run gossip algor...
متن کاملDynamic Resource Management in Large Cloud Environments Using Distributed Gossip Protocol
Resource management poses particular challenges in large-scale systems, such as server clusters that simultaneously process requests from a large number of clients. We mainly focus on the dynamic resource management in large scale cloud environment. Our core contribution centers around outlining a distributed middleware architecture and presenting one of its key elements, a gossip protocol P* t...
متن کامل