Article Preview
Top1. Introduction
In each and every domain, the wireless sensor network (WSN) and its applications are rising in a rapid manner. The WSN installed in a specific area comprises of numerous Sensor Nodes (SNs) that are battery power-driven. Although these SNs are small in size but are capable of sensing, computing and doing the communication. The WSN has its applications in almost every domain, be it in controlling air pollution, flood/ fire detection, tracking, environmental monitoring, battlefield monitoring, and security etc. (Akyildiz et al. 2002). However, there is an issue of power consumption for the networks as the battery that is used in the sensors can’t be replaced. The activity of communication carried out by the SN consumes the maximum battery power and the computation and sensing activities utilizes the rest of its energy. In SNs, battery, memory and processing power are quite limited, so to maximize the performance of a system, there is need to utilize these resources efficiently and judiciously (Jain and Kumar 2020). All techniques and protocols aim to be designed in an efficient manner to save the energy of SN in order to improve the overall network lifetime. The sensors sense their target area to collect data and then the gathered data is sent to the base station (BS) (Yick et al 2008). In the network, the nodes use the short-range radio signals to communicate with each other to accomplish their tasks. The lifetime network augmentation is also possible by using the cluster-based methodologies for routing (Gupta et al. 2015). In these approaches, the sensing region is divided into various clusters and for each cluster, the allocation of a cluster head (CH) is done (Kumaresan et al. 2020). The data that is sensed by the cluster member nodes is transferred by them to the CH. The CH then aggregates the collected data and conveys it to the BS (Jain et al. 2018; Dhand et al. 2016). Redundant data is produced as same data is sensed by the SNs sometimes due to close proximity or some other possible reasons. As a result, the lifespan of a network is reduced because to process this redundant data, a certain amount of network energy is wasted. The redundant information is filtered in the network using data aggregation methods (Logambigai et al 2016). Energy expenditure is reduced by minimizing the number of bits in transmission and reception of data by using data aggregation as an operative measure.
To improve the network lifetime, the packets count is reduced by using various approaches for combining data in the aggregation methods. CH is responsible for data aggregation in routing approaches that are cluster-based. Redundant data is eliminated and gathered data is handed over by the CH to the BS (Toor et al. 2019, Ramar et al. 2015). The energy usage in the network is minimized as the number of total packets of data transferred to base station is reduced because of the usage of the aggregation function at CH. A huge number of sensors are deployed in networks for measuring pressure, temperature, etc. which may result in similar or closely related readings by the nearby sensors depending upon the environmental conditions. In such possibilities, the BS requires only the SN aggregated data instead of all the data. Using data aggregation in such situations can improve the network lifespan in a very effective manner (Pradhan et al. 2016). The various neighboring SNs may transfer duplicate or redundant data. A lot of bandwidth and energy gets wasted in processing the duplicate data, so it also creates a load on the part of BS. So, for the network to work efficiently, the data aggregation or data filtering methods play an important role in these types of data handling (Sarkar et al 2017).
Another significant issue that has drawn less attention of the researchers is network availability and service without disruption. Generally, sensor networks are placed in a hostile environment, they may fail or dies due to various reasons like connectivity issues because of mobility, energy consumption, equipment failure and so forth (Heinzelman 2000). Similarly, a cluster head also faces many problems and does not work properly because of connectivity and inefficient routing that may lead to network partition. To manage such circumstance when a CH fails, we have inculcated a fault tolerance mechanism to address the issues of network partition due to the CHs failure.