[论文解读] Adaptive Probabilistic Model for Energy-Efficient Distance-based Clustering in WSNs (Adapt-P): A LEACH-based Analytical Study

Husam Suleiman, Mohammad Hamdan|arXiv (Cornell University)|Jan 1, 2021

Energy Efficient Wireless Sensor Networks被引用 5

一句话总结

该论文提出了一种基于LEACH的自适应概率聚类算法Adapt-P，通过利用剩余能量和节点间距离动态调整簇头选择概率，以最小化能量消耗并延长网络寿命。通过根据存活节点数和动态更新的最大簇头数量调整概率，Adapt-P在稳定性周期和能量效率方面优于LEACH和SEP，在仿真中将早期节点死亡率降低了最多20%。

ABSTRACT

Network lifetime and energy consumption of data transmission have been primary Quality of Service (QoS) obligations in Wireless Sensor Networks (WSNs). The environment of a WSN is often organized into clusters to mitigate the management complexity of such obligations. However, the distance between Sensor Nodes (SNs) and the number of clusters per round are vital factors that affect QoS performance of a WSN. A designer's conundrum resolves around the desire to sustain a balance between the limited residual energy of SNs and the demand for prolonged network lifetime. Any imbalance in controlling such objectives results in either QoS penalties due to draining SN energies, or an over-cost environment that is significantly difficult to distribute and operate. Low-Energy Adaptive Clustering Hierarchy (LEACH) is a distributed algorithm proposed to tackle such difficulties. Proposed LEACH-based algorithms focus on residual energies of SNs to compute a probability function that selects cluster-heads and an optimal energy-efficient path toward a destination SN. Nevertheless, these algorithms do not consider variations in network's state at run-time. Such a state changes in an adaptive manner according to existing network structures and conditions. Thus, cluster-heads per round are not elected adaptively depending on the state and distances between SNs. This paper proposes an energy-efficient adaptive distance-based clustering called Adapt-P, in which an adaptive probability function is developed to formulate clusters. A near-optimal distance between each cluster-head and its cluster-members is formulated so that energy consumption of the network is mitigated and network lifetime is maximized. The cluster-head selection probability is adapted at the end of each round based on the maximum number of cluster-heads permitted per round found a priori and the number of alive SNs in the network.

研究动机与目标

• 解决LEACH中静态簇头选择的局限性，该方法忽略了实时网络状态和距离变化。
• 通过将剩余能量和节点间距离纳入簇头选择，提升网络寿命和能量效率。
• 设计一种自适应概率函数，使其基于存活传感器节点数和每轮动态更新的最大簇头数量进行调整。
• 通过基于网络状况优化簇形成，降低能量消耗并延迟首个节点死亡。

提出的方法

• 基于剩余能量和节点间距离，构建自适应概率函数 $ P_{\text{adp}} $，以指导簇头选择。
• 根据网络密度和能量约束，预先计算每轮的最大簇头数量 $ \kappa_{\text{max}} $。
• 在每轮结束时，根据存活传感器节点数 $ \zeta $ 更新 $ \kappa_{\text{max}} $，实现对网络状态的动态适应。
• 采用距离感知的聚类模型，确保簇头与成员之间的距离接近最优，从而最小化传输能量。
• 一种基于学习的变体在各轮中动态演化 $ \kappa_{\text{max}} $ 和 $ P_{\text{adp}} $，提升稳定性。
• 该方法整合了剩余能量与距离的比值 $ \digamma $，以平衡能量负载和簇分布。

实验结果

研究问题

• RQ1 基于剩余能量和节点间距离的自适应簇头选择在无线传感器网络中如何影响网络寿命？
• RQ2 与静态的LEACH和SEP相比，动态调整 $ \kappa_{\text{max}} $ 和 $ P_{\text{adp}} $ 在多大程度上提升了稳定性和能量效率？
• RQ3 引入 $ \digamma $ 比值（即剩余能量与簇头到成员距离的比较）对簇负载均衡有何影响？
• RQ4 $ \kappa_{\text{max}} $ 的基于学习的演化如何影响首个节点死亡的时间和整体网络稳定性？
• RQ5 与原始LEACH和SEP相比，使用Adapt-P在网络寿命和能量消耗方面实现了多大程度的定量提升？

主要发现

• 基于学习的Adapt-P变体相比原始LEACH和SEP，将稳定性周期延长，使首个节点死亡时间延迟最多20%。
• 所提出的自适应概率 $ P_{\text{adp}} $ 显著减少了每轮的死亡传感器节点数量，尤其在800至2000轮窗口内效果明显。
• 引入 $ \kappa_{\text{max}} $ 和 $ \digamma $ 改进了负载均衡，减少了能量热点，实现了更均匀的能量消耗。
• 基于学习的算法在稳定性与能量效率方面均优于原始LEACH和SEP的原始版本及自适应版本。
• 每轮的最大簇头数量根据存活节点数量动态调整，增强了对网络状态变化的适应能力。
• 该算法表现出更优的QoS性能，包括总能量消耗减少和网络运行时间延长。