Estimating Proper Communication Distance for Epidemic Routing in Japanese Urban Areas Using SNS-Based People Flow Data
Estimating Proper Communication Distance for Epidemic Routing in Japanese Urban Areas Using SNS-Based People Flow Data
复制标题
DOI：
10.1109/incos.2016.22
复制
发表时间：
2016-09
期刊：
影响因子：
--
通讯作者：
A. Fujihara
中科院分区：
文献类型：
--
作者：
A. Fujihara
近年来，基于P2P的网状网络和基于邻近的服务作为一种不依赖任何集中式基站设备的信息通信网络而受到关注。蓝牙、Wi-Fi、Wi-SUN和WiGig等通信协议已经作为服务的上下文应用于日常生活中。然而，这些协议的传输范围在它们之间变化。它的覆盖范围越大，一个节点可以传输的信息就越多，但它消耗的电池电量也就越多。因此，为了将这些协议应用于延迟容忍网络（DTN），我们需要估计适当的通信距离，以优化网络效率。例如，DTN中最基本的路由协议Epidemic Routing能否在一天之内将一条消息转发到城市中几乎所有的移动的节点，如汽车和人？这可能是确定适当通信距离的基本措施之一。本文调查了日本主要城市地区（包括东京都、中京和关西地区）流行病路由的适当通信距离。为了模拟人类的流动模式，我们使用基于SNS的人流数据，这是一个开源的数据集，由SNS签到帖子生成的伪人类流动。由于节点之间的网络连接由于它们的移动性和邻近性条件而随机改变，因此在一天结束时具有转发消息的节点的最终大小表现为相对于通信距离的简化型转变。因此，使用逻辑回归分析，我们首先估计适当的通信距离，以找到最终大小急剧增加的拐点。此外，通过比较所有的估计距离，我们证明了一个非线性的缩放关系与用户ID的总数。该关系意味着对于当今流行的无线通信技术，即，蓝牙和Wi-Fi，它们的通信距离太短，无法成功地使用流行路由转发消息。
In recent years, P2P-based mesh networks and Proximity-based Services have been attracted attention as an information-communication networking without relying onany centralized base-station equipment. Some communication protocols, such as Bluetooth, Wi-Fi, Wi-SUN, and WiGig havebeen already in use as the context of services in daily life. However, transmission ranges of these protocols vary among them. The longer its range becomes, the more a node canwidely transfer a message, but also the more it consumesits battery power. Therefore, for applying these protocols to Delay Tolerant Network (DTN), we need to estimate a proper communication distance to optimize network efficiency. For example, can Epidemic Routing, which is the most basic routing protocol in DTN, forward a message to almost all the mobile nodes like cars and humans in urban area within one day? This might be one of basic measures to determine a proper communication distance. In this paper, we investigate the proper communication distance for epidemic routing in major Japanese urban areas including Tokyo metropolitan, Chukyo, and Kansai areas. To mimic human mobility patterns, we use SNS-based people flow data which is an open-source dataset of pseudo human mobility generated by SNS check-in posts. Since the network connection between nodes stochastically changes because of their mobility and proximity conditions, the final size of nodes having the forwarded message by the end of one day behaves like a percolation-type transition with respect to the communication distance. Therefore, using logistic regression analysis, we first estimate the proper communication distance to find the inflection point where the final size drastically increases. Furthermore, by comparing all the estimated distances, we demonstrate a nonlinear scaling relation with the total number of user IDs. The relation implies that for today's prevailing wireless communication technologies, i.e., Bluetoothand Wi-Fi, their communication distances are too short tosuccessfully forward a message using epidemic routing.