Adaptive Clustering with Dynamic Cluster Radius for Wireless Sensor Network

Luqiao Zhang, Qinxin Zhu, Juan WangInternational Journal of Control and Automation 2013 · DOI

Abstract

Topology control in Wireless Sensor Network (WSN) has recently been an area of intense investigation, especially the cluster-based topology control algorithms. In this study, we propose Adaptive Clustering with Dynamic Cluster Radius (ACDCR), a cluster-based topology control algorithm. This protocol proves to have several salient features, such as higher energy efficiency, and better QoS (Quality of Service). The design is accomplished by employing the hop-adaptive cluster radius, cluster head selection criteria with left energy threshold, and several other mechanisms. In ACDCR, only nodes with higher residual energy level than their neighbors are identified as cluster head candidates. Each cluster is assigned with a hop related radius for load balancing. In addition, nodes are allowed to setup direct links with the base station, rather than only their cluster heads. Simulation study shows that ACDCR outperforms HEED (Hybrid Energy-Efficient Distributed Clustering) in terms of network lifetime and QoS with fairly reasonable tradeoffs.

Introduction

In Wireless Sensor Networks (WSNs), battery-powered nodes are typically deployed to collect and relay data to a central Base Station (BS). A significant challenge in such networks is the "funneling effect": nodes closer to the BS must relay significantly more traffic than peripheral nodes, leading to rapid battery depletion. This imbalance causes the "inner circle" of nodes to fail prematurely, isolating the BS and rendering the entire network non-functional despite remaining energy in other nodes. Existing clustering algorithms often overlook this geographical energy disparity. The Adaptive Clustering with Dynamic Cluster Radius (ACDCR) algorithm is proposed to address this by balancing the energy load through topology control.

Methodology

The ACDCR algorithm introduces several innovations to manage network topology and energy consumption. Central to its design is the ability for nodes to transition through various functional states based on their local environment and energy levels.

Figure 1: State transition diagram for nodes in the ACDCR protocol.

Key Mechanisms

  1. Hop-Adaptive Radius: Cluster sizes are dynamically adjusted based on the node's hop distance from the BS. Nodes closer to the BS form smaller clusters to preserve energy for relaying tasks, while distant nodes form larger clusters.
  2. Energy-Aware Cluster Head (CH) Selection: A node only becomes a CH candidate if its residual energy exceeds the average energy of its one-hop neighbors, ensuring only the most capable nodes lead.
  3. Flexible Connectivity: Nodes can establish direct links to the BS if it is closer than any available CH.
  4. Orphan Handling: Isolated nodes can join clusters as Second Level Cluster Members (SLCMs), relaying data through neighbors instead of being forced into the energy-intensive CH role.

Experimental Results

The performance of ACDCR was evaluated using Matlab simulations across two scenarios: 200 nodes with 0.5J initial energy and 400 nodes with 1.0J initial energy. The algorithm was benchmarked against the HEED (Hybrid Energy-Efficient Distributed Clustering) protocol.

Figure 6: Comparison of network lifetime (number of dead nodes vs. rounds) between ACDCR and HEED.

As shown in Figure 6, ACDCR significantly extends the network lifetime. In the 200-node scenario, the first node failure occurred at round 2422 for ACDCR compared to 1954 for HEED. More impressively, the last node in ACDCR survived until round 18678, a 55% improvement over HEED's 11987 rounds. Furthermore, ACDCR maintained higher Quality of Service (QoS) by keeping a larger percentage of nodes active for more rounds, demonstrating a more stable and efficient energy distribution.

Conclusion

ACDCR effectively mitigates the funneling effect in WSNs by employing dynamic cluster radii and energy-aware selection criteria. The protocol achieves superior network longevity and QoS compared to standard algorithms like HEED. Future work will focus on extending the protocol to mobile environments and enhancing fault tolerance for inter-cluster communication.

Summary updated

Abstract

Topology control in Wireless Sensor Network (WSN) has recently been an area of intense investigation, especially the cluster-based topology control algorithms. In this study, we propose Adaptive Clustering with Dynamic Cluster Radius (ACDCR), a cluster-based topology control algorithm. This protocol proves to have several salient features, such as higher energy efficiency, and better QoS (Quality of Service). The design is accomplished by employing the hop-adaptive cluster radius, cluster head selection criteria with left energy threshold, and several other mechanisms. In ACDCR, only nodes with higher residual energy level than their neighbors are identified as cluster head candidates. Each cluster is assigned with a hop related radius for load balancing. In addition, nodes are allowed to setup direct links with the base station, rather than only their cluster heads. Simulation study shows that ACDCR outperforms HEED (Hybrid Energy-Efficient Distributed Clustering) in terms of network lifetime and QoS with fairly reasonable tradeoffs.
BibTeX
@article{zhang2013adaptive,
  author = {Luqiao Zhang and Qinxin Zhu and Juan Wang},
  title = {{Adaptive Clustering with Dynamic Cluster Radius for Wireless Sensor Network}},
  journal = {International Journal of Control and Automation},
  year = {2013},
  doi = {10.14257/ijca.2013.6.6.31},
}
Figure