Energy Efficient Hierarchical Clustering Approaches in Wireless Sensor Networks: A Survey

Table 2

Summary of grid-based protocols.

Grid-based hierarchical energy efficient protocols
Protocol	CH selection; approach	Type	Advantages	Disadvantages
GBDD [25]	Probabilistic	Distributed	(i) Guarantees continuous data transfer from source to destination	(i) It has communication overhead; (ii) Timestamp is used for grid validity and has to reconstruct it, which is an overhead
GHND [11]	Nonprobabilistic	Hybrid	(i) Ensures even distribution of nodes; (ii) Is energy efficient; (iii) Minimizes network overhead; (iv) Improves overall network lifetime	(i) Only suitable for static node; (ii) Not suitable for large scale networks
CBDAS [26]	Nonprobabilistic	Distributed	(i) Energy efficiency is achieved through cycle head, as only one is responsible for sending data of the entire network	(i) In a long chain, far away nodes might be selected as CH resulting in high energy consumption; (ii) It has chain breakage due to suboptimal CH; (iii) Cycle head selection is based on only residual energy
DUCA [27]	Random	Distributed	(i) Even distribution of cluster heads; (ii) Decreasing the differences in the cluster sizes though identifying overlapped regions	(i) CH selection being based on random number might lead to suboptimal CH; (ii) It is not suitable for mobile nodes and large scale networks
Grid and genetic algorithm [28]	Nonprobabilistic	Distributed	(i) Optimal CH selection; (ii) Energy efficient	(i) Periodic calculations and dynamic changes of clustering midpoints increase network overhead
PBDAS [29]	Random/nonprobabilistic	Distributed	(i) It improves network performance; (ii) Only chain head is responsible for sending data to BS; the rest of cell heads will be in sleep mode	(i) Suboptimal CH can cause chain breakage; (ii) The initial selection is random and thus can lead to suboptimal CH.; (iii) It is not suitable for large scale networks due to large chains
Grid sectoring [30]	Nonprobabilistic	Hybrid	(i) Energy efficient; (ii) Even distribution of load	(i) CH selection is based on only one parameter that is the distance from the centroid; (ii) It might have isolated nodes which can lead to network partitioning
GBRR [12]	Nonprobabilistic	Distributed	(i) Adaptive approach; (ii) Suitable for large scale randomly deployed sensor nodes	(i) Far away nodes might lead to suboptimal CH; (ii) It can have grids having no node
CH using ANP [31]	Nonprobabilistic	Hybrid	(i) Optimum CH selection; (ii) Achieves energy efficiency and extends overall network lifetime; (iii) Parameters prioritization	(i) The mobility of nodes is not considered, applied on static nodes; (ii) For large scale network, computational overhead will increase
Randomized grid-based approach [32]	Hybrid	Hybrid	(i) Energy efficient; (ii) Optimum number of CHs; (iii) Suitable for defined small scale networks	(i) There is computational overhead if active nodes do not satisfy the coverage area; (ii) The percentage is calculated again for nodes that are not redundant nor active, leading to extra energy consumption; (iii) It is not suitable for large scale networks