[论文解读] An Energy Efficient Routing Protocol for Wireless Internet-of-Things Sensor Networks.

Anshuman Chhabra, Vidushi Vashishth+3 more|arXiv (Cornell University)|Aug 2, 2018

Energy Efficient Wireless Sensor Networks参考文献 30被引用 12

一句话总结

本文提出了一种简单且节能的无线物联网传感器网络路由协议，通过基于距离、数据长度和感知数据的节点聚类，构建带权边的有向无环图（DAG），以找到通往基站的最小成本路径。与现有方案相比，该协议提升了网络寿命、能量效率和覆盖范围，且可选的休眠调度机制可进一步节省能量。

ABSTRACT

Internet of Things (IoT) are increasingly being adopted into practical applications such as security systems, smart infrastructure, traffic management, weather systems, among others. While the scale of these applications is enormous, device capabilities, particularly in terms of battery life and energy efficiency are limited. Despite research being done to ameliorate these shortcomings, wireless IoT networks still cannot guarantee satisfactory network lifetimes and prolonged sensing coverage. Moreover, proposed schemes in literature are convoluted and cannot be easily implemented in real-world scenarios. This necessitates the development of a simple yet energy efficient routing scheme for wireless IoT sensor networks. This paper models the energy constraint problem of devices in IoT applications as an optimization problem. To conserve the energy of device nodes, the routing protocol first aggregates devices into clusters based on a number of different features such as distance from base station, data/message length and data sensed from the environment in the current epoch. Then, a cluster head is elected for each cluster and a directed acyclic graph (DAG) is generated with all the cluster heads as nodes. Edges represent communication intent from transmitter to receiver and the edge weights are computed using a formulated equation. The minimum cost path to the base station is computed to allow for efficient real-time routing. Sleep scheduling is also optionally used to further boost network energy efficiency. The proposed routing protocol has been simulated and outperforms existing routing protocols in terms of metrics such as number of active nodes, energy dynamics and network coverage.

研究动机与目标

• 解决物联网传感器设备电池寿命有限这一关键挑战，该挑战在实际部署中限制了网络寿命和感知覆盖范围。
• 克服现有路由协议过于复杂、难以在实际中部署的局限性。
• 设计一种简单但高效的路由方案，通过聚类和高效路径选择优化能量使用。
• 通过集成可选的休眠调度机制作为节能手段，提升网络寿命和能量效率。

提出的方法

• 基于多个特征对传感器节点进行聚类：与基站的距离、消息/数据长度以及当前周期内感知的环境数据。
• 为每个聚类选举一个簇头，负责管理簇内通信和数据聚合。
• 以簇头为节点构建有向无环图（DAG），其中边的权重通过公式化的成本函数计算。
• 利用DAG计算每个簇头到基站的最小成本路径，以实现高效的实时路由。
• 可选地应用休眠调度，以减少空闲时段的能量消耗。
• 将整体能量约束建模为优化问题，以指导路由决策。

实验结果

研究问题

• RQ1 基于多个节点特征的聚类在多大程度上能提升物联网传感器网络的能量效率？
• RQ2 带权边的DAG路由在多大程度上能减少能量消耗并延长网络寿命？
• RQ3 休眠调度的集成如何影响整体能量动态和网络覆盖？
• RQ4 在实际物联网场景中，一种简单且可实现的路由协议是否能优于复杂且现有的方案？

主要发现

• 所提出的协议在时间维度上表现出更优的活跃节点数量。
• 协议展现出更优的能量动态，表明节点能量资源得到更好保存。
• 由于活跃节点的持续运行，网络覆盖范围得以延长。
• 基于DAG的路由结合优化的边权重，实现了高效的实时数据转发。
• 应用休眠调度可带来额外且可量化的能量效率提升。
• 该协议凭借更简单的结构实现更优性能，显著提升了在实际部署中的可行性。