2.2. Related Work
A myriad of energy efficient schemes have been proposed for WSNs. The one with the most hype around is low energy adaptive clustering hierarchy (LEACH) [16
]. LEACH is one of the pioneer routing protocols proposed for energy optimization in WSNs. The basic idea of LEACH is to balance the energy consumption by rotating the CH role among the nodes. Each node generates a random number between 0 and 1. If the random number is less than a certain threshold, the node declares itself as a CH. In LEACH, each node in the network has an equal probability to be elected as CH irrespective of its residual energy. In this equal probability CH selection mechanism, there is a possibility of electing CH having lower residual energy compared to the one with a higher energy level. LEACH is then enhanced to R-LEACH in [17
], which provides a CH selection mechanism based on the node’s initial energy, residual energy and the optimal number of CHs in the network. In [18
], low energy adaptive clustering hierarchy centralized (LEACH-C) mechanism was proposed. LEACH-C modifies the existing LEACH protocol. In LEACH-C, each node in the network forwards location and residual energy information to the sink node. Based on the received information, the sink node performs clustering and determines the optimal CH for each cluster. Due to the centralized nature of LEACH-C, communication overhead increases in the re-clustering process.
], a sleep-wake energy-efficient distributed clustering algorithm (SEED) for wireless sensor networks was proposed. SEED divides the sensing network into three regions based on energy level (e.g., low energy region, advance energy region, high energy region) to achieve even energy distribution. In SEED, nodes which have high residual energy in each region can only become CHs. However, SEED results in additional control messages overhead that highly effects the lifetime of the network. Moreover, SEED does not take into account the node(s) storage capacity, and computational capability during CH declaration process which may degrade the overall performance of network in high multimedia traffic conditions.
], an Energy Centric Cluster-Based Routing Protocol (ECCR) for WSNs was proposed. In ECCR, clusters are predetermined and static, and CHs are selected based on the node’s rank where rank comprises of node’s residual energy and average distance from member nodes. The nodes with high residual energy and low average distance have a high rank. ECCR also resulted in control messages overhead which increases the power consumption of the nodes.
A dual and static CH selection mechanism was proposed in [21
] that split the entire network into equal sized static clusters. In this scheme, two CHs are selected in each cluster e.g., one for data aggregation and another for data transmission. The CHs are selected based on residual energy and distance from child nodes and sink node. The CH which has low distance from member nodes is used for data aggregation while the other CH which has low distance from sink node is responsible for data transmission. The selection of two CHs generates additional control overhead in the network. Furthermore, this scheme only considers residual energy and distance for CH selection which may result in inefficient CH selection in a heterogeneous network environment.
An energy-driven unequal clustering protocol (EDUC) [22
] was proposed to minimize the energy consumption in re-clustering (i.e., CH node rotation). In EDUC, a node can act as a CH only once during the entire network lifetime. Another scheme named as energy and proximity based unequal clustering algorithm (EPUC) [18
] was proposed to address the energy imbalance issue of the nodes closer to the base station (BS) due to excessive relaying of data. In this scheme, nodes are selected as CHs based on their residual energy and BS proximity. In [23
], an extension of LEACH protocol named as stable election protocol (SEP) was proposed. SEP ensure the uniform energy consumption of all the nodes in the network to prevent the early depletion of nodes.
], an energy efficient multi level and distance aware clustering (EEMDC) was proposed. EEMDC splits the entire network into three levels based the number of hops from the base station. These levels comprised of first level clusters which have hop counts of one to two, the second level clusters having hop counts of three to five and third level clusters which have hop counts of six or more. The different levels determine the distance from base station (e.g., the nodes in level 1 are closer to BS than that of nodes in level 3). In EEMDC, CH declaration is based on residual energy and hop-count. EEMDC ensures the smallest route towards the BS in order to optimize the energy consumption and enhance the overall network lifetime.
An energy-aware routing algorithm (ERA) has been proposed in [25
]. In ERA, the CH declaration mechanism is based on residual energy of nodes. Meaning that, the nodes which have high residual energy are elected as CH. ERA generates directed virtual backbone (DVB) of CHs to relay the data from CHs towards the sink. Sink initiate process of DVB generation by broadcasting a route request message towards the CHs. On recieving route request message, the CH(s) increment its level by one, higher than that of sink (e.g., consider the level (L) of the sink node is zero e.g., L(sink) = 0, then L(CH) = L(sink) + 1) and then re-broadcast the message to other CH nodes in their vicinity and creates a DVB. A CH selects the relay nodes based on the ratios of the average residual energy of the CHs in different levels and distributes all the aggregated data packets towards the sink node sequentially.
Energy and coverage aware distributed clustering (ECDC) has been proposed in [26
]. In ECDC, the node declare itself as a CH based on its residual energy and coverage. Each node share its residual energy and coverage information with its neighboring nodes. The node which has high residual energy and coverage is selected as a CH. ECDC achieves a lower energy consumption of nodes and better in-coverage performance compared to other protocols.
A decentralized energy-efficient hierarchical cluster-based routing algorithm (DHCRA) was proposed in [27
]. In DHCRA, CHs are selected at tree edges based on residual energy and distance from BS. DHCRA minimizes control overhead and optimizes the energy consumption of nodes. In [28
], an adaptive clustering algorithm (TCAC) has been proposed to enhance the lifetime of the network. TCAC enables the CHs to adjust their power level to achieve the desired network connectivity. TCAC generate the balanced clusters across the whole network and improves the network lifetime. However, in TCAC, the periodic transmission of range updates and competition-based CH selection increases the network complexity.
Link-aware clustering mechanism (LCM) [29
] for WSNs has been proposed. In LCM, the CHs are elected based on the condition of the links and status of sensor nodes. In LCM, predicted transmission count (PTX) has been introduced to evaluate the condition of the node(s). PTX is calculated based on the residual energy, transmission power, and the link quality with a specific neighbor. The node which has a large PTX value has significant chance to become a CH.
Cluster chain weighted metrics (CCWM) has been proposed in [30
] to optimize the energy consumption and enhance the performance of WSNs based on weighted metrics. In CCWM, the CHs are elected based on weighted metrics. In this scheme, the child nodes forward the sensed data directly towards the CH whereas the CH transfer the aggregated data towards the neighbor CH until the data reach the BS. CCWM improves the network lifetime however, the direct intra-cluster communication may leads to unfair energy distribution. A non-probabilistic multi-criteria based CH selection mechanism was proposed in [31
]. In this scheme, analytical network process decision tool is used for CH is selection. A collection of various parameters have been collected and the best parameters among the collected ones are selected for CH selection.
How does the proposed scheme differ from prior works? In most of the existing cluster-based schemes in WSNs, CH declaration is based on residual energy, distance from member nodes, and distance from the sink node. However, to the best of our knowledge, all such schemes do not consider storage capacity and computational capability for best CH declaration. Besides, most of the existing proposals do not take into account one or more of the following CH and sensor node parameters in new node association process: (i) traffic load on CH; (ii) residual energy; (iii) computational capability of a CH; (iv) distance between CH and sink node; and (v) less distance between a sensor node and a CH. In contrast to the existing schemes, our scheme proposes an efficient mechanism for the new node association, and considers all of the aforementioned parameters for optimal CH selection. Furthermore, the proposed scheme also provides CH-acquaintanceship and CH-friendship mechanism in order to prolong the network lifetime.