Study on real time data collection and distribution method in sensor networks

在大规模传感器网络中，为了覆盖较大的监测区域，传感器节点之间需要进行多跳通信。此外，传感器节点应该分组成簇，以增强可扩展性和健壮性。研究了大规模传感器网络簇间多跳通信的特点，并与其他路由方法进行了比较。由于只需要相邻传感器节点的信息，所以我们也采用了CSMA/CA。结果表明，与基于所有传感器节点位置信息的时分多址方式相比，CSMA/CA方式的功耗提高了12%，报文收集时间延长了约4倍。此外，为了在实际环境中验证我们的传感器网络自主室内定位系统的有效性，我们在基于ZigBee标准的无线传感器网络中实现了该系统。该系统通过在适当数量的传感器节点上测量RSSI(接收信号强度指示器)来自动估计传感器节点之间的距离。通过实验，验证了本文提出的数据采集和位置估计技术的有效性。结果表明，该算法的位置估计误差降低到1.5-2m。此外，受群体智能的启发，提出了一种多汇点结构的传感器网络数据采集方案。在该方案中，每个传感器节点通过重复交互和来自邻居和环境的反馈来确定其下一步行动。通过这些操作，在网络级别出现了有利的集群和路由。仿真结果表明，该方案能够可靠地将事件信息传递给汇聚节点，在质量极差的无线信道中具有较强的鲁棒性，并且对传感器节点甚至汇聚节点的故障具有自恢复能力。

In large-scale sensor networks, multi-hop communication between sensor nodes is necessary to cover a large monitoring region. Moreover, sensor nodes should be grouped into clusters to enhance scalability and robustness. We examine the characteristics of multi-hop communication between clusters in large-scale sensor networks and compare them with other routing methods. We also apply CSMA/CA since the information of only neighboring sensor nodes is necessary. As a result, using CSMA/CA, power consumption increases by 12% and the packet collection time becomes about four times longer in comparison to using TDMA based on location information of all sensor nodes.Furthermore, to verify the validity of our autonomous indoor localization system for sensor networks in an actual environment, we implemented it in a wireless sensor network based on the ZigBee standard. The system automatically estimates the distance between sensor nodes by measuring the RSSI (received signal strength indicator) at an appropriate number of sensor nodes. Through experiments, we clarified the validity of our data collection and position estimation techniques. The results show that the position estimation error was reduced to 1.5-2 m.Moreover, we propose a new scheme of gathering data from sensor networks with multi-sink configurations inspired by the swarm intelligence. Each sensor node determines its next action through repeated interaction and feedback from its neighbors and environments in this scheme. Advantageous clustering and routing emerge in network level from these actions. Our simulation results revealed that the proposed scheme can reliably deliver event information to the sink nodes, is robust over very-poor-quality wireless channels, and has self-recovery capability to deal with sensor-node failure and even that in sink nodes.