A Framework for Information Propagation in Mobile Sensor Networks
A Framework for Information Propagation in Mobile Sensor Networks
复制标题
DOI：
10.1109/mass.2013.9
复制
发表时间：
2013-10
期刊：
影响因子：
--
通讯作者：
Jiajia Liu;Hiroki Nishiyama;N. Kato
中科院分区：
文献类型：
--
作者：
Jiajia Liu;Hiroki Nishiyama;N. Kato
在移动传感器网络中，如何有效地控制中继节点的转发行为，以节省节点的能量消耗和缓冲区使用，同时满足指定的传输性能要求，是路由问题中一个常见的难题。可用作品为每条消息分配生存期、最大副本数或序列号，或者在消息接收后在整个网络中刷新特殊反馈信息。在前一种情况下，中继节点不知道消息的接收状态，将消息携带并转发到目的地，而后一种情况可以有效地通知所有中继节点，但需要额外的通信资源。与以往的研究不同，本文考虑了中继节点的显式概率停止机制。在这种机制下，主动传播消息的中继节点在遇到另一个已经收到消息的节点后，会以一定的概率停止传播消息。我们首先建立了一个二维马尔可夫链框架来描述消息传播结束前的高度复杂动态，然后进行马尔可夫分析，得出相关的重要性能指标，包括完成消息传播所需的平均时间，最终接收消息的节点比例的期望和方差，以及给定数量的节点最终接收消息的概率等。最后，提供了广泛的数值结果来分析探讨网络参数设置如何影响这些性能指标。
A common complication for routing in mobile sensor networks is how to efficiently control the forwarding behaviors of relay nodes so as to save their energy consumption and buffer usage while simultaneously satisfy the specified delivery performance requirement. Available works either assign each message with a lifetime, a maximum number of copies, or a sequence number, or flush special feedback information among the whole network after the message reception. In the former case, a relay node has no idea of the message reception status and will carry and forward the message until meeting the destination, while the latter could efficiently notify all relay nodes but demands extra communication resources. Different from previous studies, we consider in this paper an explicit probabilistic stopping mechanism for relay nodes. Under such mechanism, a relay node that is actively disseminating a message will stop spreading the message with a certain probability, after meeting another node having already received the message. We first develop a two-dimensional Markov chain framework to characterize the highly complicated dynamics until the end of message propagation, then conduct Markovian analysis to derive the associated important performance metrics, including the average time required for the completion of message propagation, the expectation and variance of the fraction of nodes finally receiving the message, and the probability that a given number of nodes end up with the message, etc. Finally, extensive numerical results are provided to analytically explore how the network parameter settings may affect these performance metrics.