Survey of Connectivity Restoration in 3D Wireless Ad Hoc/Sensor Networks

Table 5

Summarization of the recovery method, contribution, and limitation of some literature in 2D WASN.

Literature	Recovery method	Contribution	Limitation
[37]	Passive, node movement recovery based on clustering and considering the secondary damage.	This proposed DBCE connectivity establishment scheme, which includes segment evaluation/selection approaches to reduce connectivity cost, enhance network robustness, and improve longevity by considering segment shapes and local network features.	This method cannot be fully applied in the 3D environment.
[38]	Active passive hybrid, node movement k-connectivity recovery based on noncritical nodes.	The proposed PINC algorithm efficiently restores movement-based k-connectivity by categorizing nodes into critical and noncritical groups, moving noncritical nodes to replace failed critical nodes with minimal movement cost, outperforming competitors in terms of speed and effectiveness based on real motes and robot testbed data.	This algorithm reduces coverage and cannot be fully applied in the 3D environment.
[39]	Passive, node movement recovery based on clustering and using machine learning.	The proposed CRrbf, a machine learning-based connectivity restoration strategy using RBFNN and UKF to optimize aggregation ratio, reduce energy cost, and outperform distance and terrain-based strategies in terms of aggregation ratio, network latency, throughput, and energy efficiency based on theoretical analysis and simulation results.	This method is not considered to restore the connectivity through a limited number of relay nodes and mobile data collectors. At the same time, it is not considered to use deep reinforcement learning (i.e., DDPG) for efficient path optimization of mobile data collectors.
[40]	Active passive hybrid, node movement k-connectivity recovery based on two-hop neighbor nodes and considering the locations, moving costs, and obstacles.	The proposed CMH algorithm for k-connectivity restoration in heterogeneous WSANs, where nodes determine local subgraphs and minimum moving costs, check failure impact on k using disjoint paths, and utilize actuators to restore connectivity.	It does not consider how to realize k-connectivity restoration in the 3D environment.
[41]	Passive, node deployment recovery based on clustering and sleep scheduling using the Steiner tree and convex polygons.	The proposed method utilizes the Steiner tree and convex polygons to enhance fault tolerance in wireless sensor networks by creating dual connectivity through relay nodes. By simplifying the network structure and reducing data communication delay, the approach minimizes relay node deployment and extends the network lifetime compared to existing algorithms.	There is no simulation experiment in the actual scene, and the security of data transmission is not considered. And it also cannot be fully applied in the 3D environment.
[46]	Active, node movement recovery based on clustering using irregular cellular automata.	The proposed irregular cellular automata (ICA) model in WSN for fault node identification through clustering and cluster-head selection, minimizing computational overhead and bandwidth usage. The system utilizes node data structures to select cluster heads based on specific rules within ICA.	The number of clusters is not optimal, and they also dot not consider the security, QoS, and load balancing. This method cannot be fully applied in the 3D environment.
[57]	Active, critical node (C-N) detection only using the neighbor’s received signal strength indicator (RSSI) information.	The proposed ABCND algorithm for critical node detection in IWSN, consisting of a 2D critical node detection algorithm in phase I and a correlation-based reliable RSSI approach in phase II. The algorithm achieves efficient convergence and critical node detection with reduced energy consumption compared to state-of-the-art methods, demonstrating 90% to 95% accuracy in detecting critical nodes while consuming 50% less energy.	The ABCND algorithm is not simulated in the large-scale IWSN and on real hardware. Moreover, the algorithm has some false positivity when implemented in 3D topology.
[60]	Passive, node deployment recovery based on multiobjective evolutionary method using a hop count bound as a delay constraint.	The proposed GPrim algorithm to minimize relay nodes and maximize network lifetime with local heuristics aiding initialization, crossover, and mutations in wireless sensor networks. Extensive experiments demonstrate the method’s effectiveness in improving network metrics with reasonable computational time tradeoff compared to standard encoding methods.	The model cannot cope with mobile sensors and is not suitable in heterogeneous wireless sensor networks.