Rapid and Efficient Weighted Clustering Algorithm for High-Speed Flight Vehicle Ad Hoc Network

0
0
0

Rapid and Efficient Weighted Clustering Algorithm for High-Speed Flight Vehicle Ad Hoc Network

Introduction

High-speed flight vehicle Ad Hoc networks represent an emerging network architecture composed of nodes such as fighter jets, high-speed drones, and hypersonic missiles. These networks have become crucial tactical and strategic assets in modern warfare and surveillance. However, compared to traditional mobile Ad Hoc networks, high-speed flight vehicle networks exhibit unique characteristics that pose significant challenges.

First, the extreme mobility of high-speed flight vehicles, reaching speeds of hundreds or even thousands of kilometers per hour, introduces substantial difficulties in maintaining network topology and ensuring reliable data transmission. Second, these networks often involve large-scale deployments with rapidly changing topologies, complicating network management and coordination. Additionally, the high-speed flight environment imposes stringent requirements on communication latency and rapid recovery from disruptions.

To address these challenges, clustering architectures are commonly employed to enhance scalability, manageability, and resource allocation. Clustering improves coordination among nodes, optimizes routing, and ensures efficient bandwidth utilization. However, existing clustering algorithms face several limitations, including prolonged network setup times, high maintenance overhead, and slow recovery from node failures.

This paper introduces a Power-aware Weighted Clustering Algorithm (PWCA) designed specifically for high-speed flight vehicle Ad Hoc networks. PWCA improves upon existing weighted clustering algorithms by refining key metrics, optimizing the network setup process, and introducing efficient cluster maintenance mechanisms. The algorithm reduces setup time, minimizes control overhead, and accelerates network recovery, making it highly suitable for dynamic and complex flight environments.

Network Model and Problem Description

Network Model

The network model adopts a hierarchical clustering structure. The lower layer consists of centralized single-hop clusters where nodes communicate via cluster heads using the same transmission frequency. The upper layer forms a distributed multi-hop network composed of cluster heads. This architecture offers advantages such as low latency, structural stability, and rapid recovery from disruptions, making it ideal for high-speed flight vehicle networks.

Problem Description

Existing clustering algorithms face several critical issues:

  1. Dependence on Location Information – Many algorithms rely on satellite-based positioning to calculate metrics such as inter-node distance and mobility. However, in adversarial environments where GPS signals are jammed or unreliable, these algorithms fail to function effectively. PWCA addresses this by using received signal strength to estimate distances, reducing dependency on external positioning systems.

  2. Gradient Suppression in Network Setup – In distributed networks, gradient suppression occurs when nodes with lower weights are inhibited from becoming cluster heads due to neighboring nodes with higher weights. Traditional algorithms require multiple election rounds, increasing setup time. PWCA introduces an optimized setup process that mitigates gradient suppression, allowing more nodes to join the network in a single round.

  3. High Maintenance Overhead – Periodic backup cluster head elections and notifications generate significant control overhead. If the election interval is too long, backup cluster heads may become outdated, leading to inefficient recovery. If too short, excessive control messages degrade network performance. PWCA introduces an efficient cross-layer notification mechanism to reduce overhead while ensuring timely updates.

  4. Slow Recovery from Cluster Head Failures – When a cluster head fails abruptly, existing algorithms exhibit delayed responses from cluster members, prolonging network recovery. PWCA implements a rapid response switching mechanism that minimizes downtime and ensures seamless transitions to backup cluster heads.

Cluster Head Election Metrics

PWCA employs three key metrics for cluster head election:

  1. Average Inter-Node Distance

Instead of relying on GPS coordinates, PWCA estimates distances using received signal power. Under the free-space path loss model, the distance between two nodes can be approximated based on transmission power, receiver sensitivity, and signal wavelength. The algorithm normalizes the average distance to all neighboring nodes, ensuring a balanced distribution of cluster heads.

  1. Average Mobility Similarity

Mobility similarity measures how consistently two nodes move relative to each other. PWCA calculates the coefficient of variation (CV) for distance changes over time. A low CV indicates stable movement patterns, while a high CV suggests erratic mobility. Nodes with higher mobility similarity to their neighbors are preferred as cluster heads to maintain stable clusters.

  1. Node Degree with Reliability Correction

Traditional node degree metrics count the number of neighboring nodes but ignore link quality. PWCA refines this by incorporating connection reliability, measured as the ratio of successfully received packets to total transmissions. The algorithm also considers an ideal cluster size derived from network density to avoid overly large or small clusters.

These metrics are combined into a weighted score, where coefficients can be adjusted based on network requirements. For example, in highly mobile scenarios, mobility similarity may be weighted more heavily, whereas in dense networks, node degree could take precedence.

Clustering Algorithm Implementation

Network Setup Process

The network setup process in PWCA consists of the following steps:

  1. Node Initialization – Upon activation, nodes enter an “undecided” state and begin broadcasting hello messages to discover neighbors.
  2. Weight Calculation – Each node computes its weight based on the three election metrics and broadcasts this information.
  3. Cluster Head Election – Nodes compare their weights with neighbors. If a node has the highest weight, it declares itself a cluster head and broadcasts a cluster formation message.
  4. Cluster Member Joining – Nodes that cannot become cluster heads listen for formation messages and send join requests to the nearest cluster head.
  5. Gradient Suppression Mitigation – Nodes that neither become cluster heads nor receive formation messages broadcast additional hello messages to update neighbor weights, reducing gradient suppression effects.
  6. Multi-Round Elections – Nodes failing to join in the current round reattempt in subsequent rounds until all nodes are integrated.

This optimized setup process significantly reduces the number of election rounds required, accelerating network formation.

Cluster Maintenance Mechanisms

Backup Cluster Head Election

Cluster heads periodically evaluate member nodes to select the most suitable backup. The selection criteria include mobility similarity, connection reliability, and distance stability. Unlike traditional periodic elections, PWCA triggers backup updates only when membership changes or a predefined interval expires, reducing overhead.

Cross-Layer Notification Mechanism

Instead of using separate control messages, PWCA embeds backup cluster head information in MAC-layer beacon frames. The first time slot in each frame is reserved for the cluster head, while the second slot indicates the backup. This approach eliminates the need for explicit backup notifications, improving efficiency and reliability.

Rapid Response Switching Mechanism

When a cluster head fails, the backup cluster head detects the absence of beacon messages and listens for activity from other cluster members. If no activity is detected, the backup assumes leadership and broadcasts a new cluster formation message in the next contention slot. Cluster members that miss beacon frames check their stored backup information and attempt to rejoin the network promptly.

This mechanism ensures fast recovery while minimizing false positives (where a backup mistakenly assumes leadership due to packet loss).

Performance Evaluation

Simulations using OPNET 14.5 compared PWCA with existing algorithms (WCA, WSCA, DSWCA, FSWCA) under varying conditions:

  1. Network Setup Time – PWCA required fewer election rounds than WCA and DSWCA, thanks to its gradient suppression mitigation. Although the additional hello messages increased per-round overhead, the overall setup time was significantly reduced.
  2. Cluster Stability – PWCA produced fewer isolated clusters than WCA and maintained better stability than FSWCA, which occasionally suffered from false leadership transitions.
  3. Recovery Time – After cluster head failures, PWCA restored connectivity faster than WSCA and DSWCA, which relied on periodic heartbeat messages. The cross-layer notification mechanism ensured timely updates.
  4. Control Overhead – By eliminating explicit backup notifications, PWCA reduced control traffic by up to 30% compared to WSCA and FSWCA.

These results demonstrate that PWCA is well-suited for high-speed flight vehicle networks, offering rapid setup, low overhead, and robust recovery.

Conclusion

The Power-aware Weighted Clustering Algorithm (PWCA) addresses critical challenges in high-speed flight vehicle Ad Hoc networks. By refining election metrics, optimizing the setup process, and introducing efficient maintenance mechanisms, PWCA reduces network formation time, minimizes control overhead, and accelerates recovery from failures.

Future research will focus on enhancing topology convergence during large-scale disruptions, such as cluster mergers or splits, to further improve network resilience in dynamic environments.

For more details, refer to the original paper: https://doi.org/10.19734/j.issn.1001-3695.2024.07.0257

Was this helpful?

0 / 0