Desktop version

Home arrow Computer Science

  • Increase font
  • Decrease font


<<   CONTENTS   >>

BACKGROUND INFORMATION

In this section, a brief overview of clustering is described in detail with objectives and characteristics [11].

Clustering Objective

A node is clustered in WSN with several objectives that depend on function necessity [12-14]. The energy maintenance and removing hot spot issue are the most general objectives of clustering. Several of the further objectives are described in the following sections.

4.2.1.1 Scalability

Sensor nodes are used in massive numbers ranging from hundreds to thousands, depending on the function necessity in the current situation. A pattern of routing methods has the capability to work with this massive count of sensor nodes [15]. If node in the cluster requires for transmitting data to a node in one more cluster, the nodes must identify the aspects of the connected obtaining CH. The hierarchical structural design gives scalability in huge scale WSN by separating the sensing domain into several layers, and all layers are again separated into count of clusters. These lead to enhanced scalability and diminish the size of the routing table.

4.2.1.2 Fault-Tolerant

For many purposes, sensors are used in harsh environment (e.g., sensors are dropped from helicopter). These nodes contain enhanced danger of physical damage and failure of nodes. Fault-tolerant nodes are significant in vital function where the loss of any sensor data results in catastrophe [16]. Clustering is the method capable of making a fault-tolerant and protected WSN. A self-organized WSN controls the fault by the procedure of reclustering the network. The reclustering procedure not only enhances the resource burden but also interrupts the present process. Fault tolerance is attained by reclustering, allocating backup CH, depute CH, or rotating CH.

4.2.1.3 Data Aggregation/Fusion

Because the huge counts of sensors sense the similar data in the physical surroundings, there is much possibility of needless data. A data aggregation is an efficient method to keep away useless data broadcast, and it also diminishes a count of broadcasts [17-19]. These methods are a signal procedure technique that aggregates each obtained packet into a resultant packet. These methods amplify the general data and suppress the redundant noise. In WSN, CH carries out data aggregation on data that is obtained from its CMs and forwards the aggregated data to BS using single hop or multihop. So, the count of broadcasts and the entire load of the network are also extensively decreased.

4.2.1.4 Load Balancing

Load balancing plays an essential role in extending the network duration. The load balancing is a critical problem where CHs are chosen from accessible nodes in the network. Uniform load distribution between the CHs is important for avoiding hot spot issue. In a clustering security regular load distribution, each of the CH utilizes around similar count of energy. As an outcome, further energy-capable network is simply obtained [20].

4.2.1.5 Stabilized Network Topology

A node is managed into clusters and CH is chosen from all clusters; CH is dependable to some topology modifies at the cluster level. The CH has the data of its CMs such as node id, location and energy level. Controlling the network topology in hierarchical structural design is optimal than flat design. If a node expires or goes to other cluster, this alters are directly registered and informed by CH to BS and reclustering is completed for maintaining the network topology efficiently.

4.2.1.6 Increased Lifetime

An important function of clustering is to enhance the network duration as a long as feasible. Because the sensors are energy-limited, maximizing the network duration is very essential to concurrent functions. Intracluster communication is diminished by choosing nodes as CHs with further neighbor nodes. A clustering and routing procedure can also be joined to maximize duration. A clustering extends the duration of WSN by rotating CHs suitably between the CMs, sleep modes and cluster conservation methods are accurately employed for enhancing the network duration.

Clustering Characteristics

For classifying several clustering manners, many clustering features are utilized. In this section, three characteristics of clustering are described in detail.

  • • Cluster properties
  • • CH properties
  • • Clustering procedure properties
  • 4.2.2.1 Cluster Properties

The design of the clusters is determined as cluster properties that contain cluster count, cluster size, intracluster communication, and intercluster communication [21, 22].

  • 4.2.2.1.1 Cluster Count The count of cluster created is fixed or variable depends on the function necessity. In few cases, the count of cluster is 5% of entire count of the nodes utilized. In several functions, the count of clusters is variable if the CHs are arbitrarily chosen.
  • 4.2.2.1.2 Cluster Size A cluster size is classified into equal and unequal size cluster. In equal clustering, the size of the cluster is similar throughout the network. In unequal clustering, the cluster size is defined depending on the distance to BS. A cluster size is lesser if the distance to BS is small and the size enhances as the distance to BS enhances.
  • 4.2.2.1.3 Intracluster Communication Intracluster communication contains the data broadcast among CH as well as CMs in a cluster. According to the clustering manners, the communication is direct or multihop communication. In huge-scale WSN, multihop communication is required to data broadcast in a cluster.
  • 4.2.2.1.4 Intercluster Communication Intercluster communication is direct or multihop communication. Generally, multihop method is selected for energy-capable data broadcasts from CHs to BS with in-between CHs in huge-scale WSN. A few functions of tiny-scale WSN, such as the communication between CHs and BS, are single-hop broadcast.

4.2.22 CH Properties

The CH carries out the subsequent processes: gathering data from CMs, data aggregating and forwarding to BS with direct or multihop communication. The CH obtains data from its CMs, execution data aggregation of the gathered sensor data transmits the combined data to BS.

  • 4.2.2.2.1 Clustering Process A characteristic of clustering procedure is listed under.
  • 4.2.2.2.2 Clustering Methods There are two techniques of clustering: centralized and distributed. In centralized methods, a central authority such as BS or supernodes manages the whole function (cluster formation, CH selection, and so on.) as distributed manners have no central influence and are extremely used in huge-scale WSN.
  • 4.2.2.2.3 Objective of Node Grouping Many objectives of nodes grouping have already been explained in this study, for example, fault tolerance, load balancing, and so on.
  • 4.2.2.2.4 Nature The clustering procedure is proactive, reactive, or hybrid in nature. A node always senses the data and forwards it to CH. In proactive form, the CH always broadcasts the data to BS. In reactive form, CH broadcasts the data every time the sensed value crosses the existing threshold. In hybrid cases, CH broadcasts the data to BS at longer usual time intervals and also if the value crosses the threshold value.
  • 4.2.2.2.5 CH Selection There are three methods for selecting CH in WSN: probabilistic techniques are attributed to dependent technique and preset form. In probabilistic manner, the CHs are chosen arbitrarily without some preceding consideration. In attribute-based technique, several metrics are utilized for selecting CHs such as remaining energy, node degree, node centrality, usual remaining energy, distance to BS, and so on. In preset form, CHs are predetermined before placing the sensors in the sensing domain.
 
<<   CONTENTS   >>

Related topics