gossip-based computation of aggregate information（gossip-based计算的聚合信息）.pdf

Gossip-Based Computation of Aggregate Information David Kempe, Alin Dobra, and Johannes Gehrke Department of Computer Science, Cornell University Ithaca, NY 14853, USA kempe,dobra,johannes@ Abstract our physical environment with sensor networks consisting of hundreds of thousands of small sensor nodes [24, 28, 35]. Over the last decade, we have seen a revolution in con- Applications for such large-scale distributed systems have nectivity between computers, and a resulting paradigm shift three salient properties that distinguish them from tradi- from centralized to highly distributed systems. With massive tional centralized or small-scale distributed systems. scale also comes massive instability, as node and link fail- First, the dynamics of large-scale distributed systems are ures become the norm rather than the exception. For such often significantly different. For example, in P2P networks, highly volatile systems, decentralized gossip-based proto- individual machines are often under the control of a large cols are emerging as an approach to maintaining simplicity number of heterogeneous users who may join or leave the and scalability while achieving fault-tolerant information network at any time. Sensor networks often involve the de- dissemination. ployment in inhospitable or inaccessible areas that are natu- In this paper, we study the problem of computi