RESEARCH BACKGROUND

Swarm intelligence defines cooperative intelligence [12]. Biologists and natural scientists are continually examining the nature of social insects because of their effectiveness in solving the difficult problems like determining the shortest route from nest to food source or constructing nests [13,14]. Despite the fact that the behaviors of these insects are unrefined separately, they make wonders as a swarm by interaction with each other and their environment. Recently, the nature of different swarms, which are utilized in the identification of prey or mating partner, is mimicked to numerical optimization technique [15]. In this study, unequal clustering problem has been resolved by the use of swarm intelligence techniques.

Energy-balanced unequal clustering (EBUC) [16] makes use of particle swarm optimization (PSO) algorithm to produce clusters of uneven sizes. It partitions the network into various parts of different sizes where the clusters near to BS be supposed to be of smaller sizes. The CHs near to BS save high energy which can be useful for data transmission between clustering and thereby hot spot problem will be eliminated. Genetic Algorithm based Energy-Efficient Adaptive Clustering Hierarchical Protocol (GAEEP) [17] is presented to increase network lifetime as well as network stability by electing optimum sum of CHs and its location by the use of GA. The working of GA is divided to setup as well as steady state phases. The effectiveness of GAEEP is verified by comparing its results with LEACH [18], SEP, ERP, LEACH-GA along with DEU in both homogeneous as well as heterogeneous networks in terms of lifetime, average remaining energy, and throughput.

Unequal clustering by improved particle swarm optimization (IPSO) [19] is developed to eliminate hot spot issue and also to overcome the usual limitations of PSO. The existing modified PSO algorithm may provide better performance but it suffers from high algorithmic complexity or high computational cost. It usually operates in numerous rounds where every round begins amid a setup phase continued by a steady state phase. The simulation results verified that the limitations of PSO algorithm are overcome by IPSO algorithm and is proved in terms of the number of alive nodes in WSN.

Sink mobility-based EBUC (SMEBUC) [20] is introduced for the attainment of balanced energy utilization by the use of shuffled frog leaping algorithm (SFLA). SFLA is used for the election of CHs and to organize clusters of varying sizes by the consideration of remaining energy level in the sensor node. To minimize the rotation of CHs often, CHs work continuously to identify the exchange time of CHs as well as node weights. The greedy algorithm is employed to choose the finest relay node stuck between CHs and BS. In addition, mobile sinks are presented to conquer the hindrance of hot spot concern. The highlight of the SMEBUC algorithm is verified by the performance comparison with LEACH and EBUCP in terms of energy dissipation as well alive nodes.

Novel chemical reaction optimization-based unequal clustering and routing algorithm (nCROUCRA) is proposed. A new chemical reaction optimization (nCRO) is proposed for asymmetrical clustering as well as routing algorithms and is called as nCRO-UCRA. To obtain unequal clustering, an nCRO paradigm-based CH selection is done using a derived cost function. In addition, a routing strategy is also developed using nCRO algorithm. They are designed with the effective models of molecular structure encoding as well as novel potential energy functions. This algorithm [21] is implemented in different scenarios based on the sensor count and CHs count.

LITERATURE SURVEY

At present, diverse models have been introduced using optimization techniques to address the hot spot issue. Yuan et al. [22] developed a genetic algorithm-based clustering model which determines the CH count and their location for lessening the utilization of energy in WSN. The functioning of this technique takes place in diverse iterations where every iteration comprises a setup and steady stage. At the former stage, the BS finds out the CHs and position. In the latter stage, a path is derived from the node to BS. It enables a node to transmit data in a straightforward manner to BS when the distance to BS is lower compared to the distance to CHs.

Fan and Du [23] generated different-sized clusters with respect to remaining energy level. Is also chooses the CHs using SFLA. It operates in two levels namely cluster construction and communication. The choice of CHs takes place in the earlier level and greedy method based path identification takes place in the latter level. Gajjar et al. [24] presented a clustering model which involves three levels, namely setup, neighborhood discovery, and steady state levels. At the initial two levels, node classification takes place in various layers and packet broadcasting takes place for neighboring node identification. It makes use of nonpersistent carrier sense multiple access (CSMA) for accessing the channel. At the last level, the process of CH election, construction of the cluster, and communication takes place. It makes use of fuzzy logic for selecting the CHs and optimum path selection takes place using ant colony optimization (ACO) algorithm.

Sabor et al. [25] developed an unequal clustering model for determining the cluster size and multi-objective immune technique for producing a routing tree. The size of the clusters will be identified using remaining energy level and distance to BS. Therefore, various optimization algorithms are applied for the identification of CHs and determination the cluster sizes. Once the CHs are chosen, the cluster will be constructed. Some of the hybridizations of unequal clustering techniques are presented in WSN. An energy balancing method for each cluster is introduced in Ref. [26]. To achieve this, the clusters are formed and then CHs are allocated to it. So, three steps of clustering take place, namely construction of clusters, election of CHs, and communication. At the first step, Sierpinski triangle is applied for creating small-sized clusters for the nodes located closer to the BS. On the selection of the CHs, it assumes the distance, remaining energy level, and node degree. A voting mechanism [27] to select the CHs takes place using remaining energy level, topology, and communication power. But, the selection of CHs takes place in a distributed way. It suffers from the drawback that this process extents the lifetime of a WSN. A Social Spider-based Unequal Clustering Protocol (SSUCP) for WSN presented in Ref. [28] is used to select the CHs and cluster size effectively.

 
Source
< Prev   CONTENTS   Source   Next >