Energy consumption and transmission reliability are the most common issues in wireless sensor networks (WSNs). By studying the broadcast nature of data transmission in WSNs, the mechanism of guaranteeing reliable transmission is abstracted as propagation of responsibility and availability. The responsibility and availability represent the accumulated evidence of nodes to support reliable transmission. Based on the developed mechanism, an evidence-efficient cluster head rotation strategy and algorithm are presented. Furthermore, backbone construction algorithm is studied to generate the minimum aggregation tree inside the candidate cluster heads. A minimum aggregation tree-based multihop routing scheme is also investigated, which allows the elected cluster heads to choose the optimally main path to forward data locally and dynamically. As a hybridization of the above, an evidence-efficient multihop clustering routing (EEMCR) method is proposed. The EEMCR method is simulated, validated, and compared with some previous algorithms. The experimental results show that EEMCR outperforms them in terms of prolonging network lifetime, improving transmission reliability, postponing emergence of death nodes, enhancing coverage preservation, and degrading energy consumption.
Decreasing energy consumption, improving energy efficiency, and enhancing transmission reliability are still main challenges of wireless sensor networks (WSNs). The related technique-efficient issues, such as clustering routing, topology control, and multihop transmission, are widely used to improve energy efficiency for WSNs [
Hierarchical topology control, in which nodes are grouped into clusters and cluster heads (CHs) are elected for each cluster to form a backbone construction, can effectively utilize the limited resources of sensor nodes. Hierarchical topology benefits maximizing the network lifetime and optimizing the data delivery ratio on each link. Clustering technique has been proven energy-efficient in WSNs [
Multihop routing is one of two main communication modes for WSNs. In comparison with multipath routing, its transmission has generally been considered an efficient energy-saving approach, especially for large-scale sensor networks [
In this paper, we also address the LEACH-improved scheme. By studying the broadcast nature of data transmission in WSNs, it is abstracted as a propagation of responsibility and availability of nodes. The responsibility and availability integrate sufficient considerations on various network factors including the residual energy of nodes, distance between nodes, distance between the CHs and the base station, and energy loss on the node-joint links. Essentially, the responsibility and availability express the accumulated evidence of nodes to support high-quality communication. Namely, the responsibility and availability represent the comprehensive capability of nodes against node failures and link losses. Based on the above development, an evidence-efficient CH rotation scheme and algorithm are proposed. Furthermore, in order to improve the data transmission reliability between the candidate CHs to the base station, the well-known Kruskal algorithm [
A demo of the principles employed in the proposed scheme.
As far as we know, the main contribution of this paper has at least the following points:
The rest of this paper is organized as follows. The related works are introduced in Section
Grouping nodes into clusters has been the most popular approach for supporting scalability in WSNs [
Usually, clustering is typically based on the energy reserve of sensor nodes and node’s proximity to the CH [
Additionally, some nonclustering routing methods or models [
The LEACH protocol, one of the first clustering routing protocols proposed for WSNs, is an adaptive, distributed algorithm that forms clusters of sensors based on the received signal strength and uses local CHs as routers to the sink node [
Based on the LEACH algorithm, two factors, the energy and distance, were cast into modifying the threshold function of LEACH protocol; a cluster head multihop routing improved algorithm (CMRAOL) based on LEACH [
Additionally, some of the existing clustering routing algorithms [
The differences covered between the proposed EEMCR scheme and the previous researches include the following: the nature of data transmission in WSNs is abstracted as propagation of responsibility and availability, and the responsibility and availability are taken as the accumulated evidence of nodes for guaranteeing transmission reliability; the responsibility and availability integrate comprehensive considerations on several network factors; an evidence-efficient cluster head rotation mechanism and algorithm are presented; backbone construction algorithm is developed to generate a minimum aggregation tree; essentially, the branches inside the generated minimum aggregation tree construct the multihop communication paths. The experimental results show that there exists a remarkable improvement on prolonging the network lifetime, postponing the death of nodes, saving energy, and conducting coverage preservation.
Assuming that
In this paper, a well-known common first-order wireless energy consumption model [
Wireless transceiver circuit energy consumption model.
With respect to the above model, the energy consumption of transmitting
The energy consumption by the node receiving
In this section, we discuss the issue of CH rotation. First, we think that the essence of CH election in clustering topology is equivalent to determining the cluster centroids of cluster algorithms in data mining. The Affinity Propagation (AP) [
Additionally, since the energy efficiency is one of the critical factors that influence the network lifetime, in particular, several new definitions with respect to the measurement or estimation of energy consumption are presented as follows.
Node-Energy-Level (NEL) is used to measure the energy consumption of nodes in the network. NEL is calculated as Formula (
The Energy-Cost
Based on the above, this paper integrates inspirations on the broadcast nature of data transmission in WSNs and the idea in AP algorithm. Furthermore, we give improvement recognition of the responsibility and availability, respectively.
The eligibility evidence of a node as a cluster head is quantified using the responsibility and availability. They are defined as follows:
Formulas (
Usually, link losses and node failures are the primary reasons to influence transmission reliability in WSNs; unfortunately, they are negatively affected by many network factors. In this paper, we take the availability and responsibility as the comprehensive evidence against node failures and link losses, which aims to improve the total performance including transmission reliability and other properties. To this point, the availability and responsibility provide a logical metric for nodes to guarantee transmission reliability.
Furthermore, in order to ensure the rationality of CH rotation and correctness of CH election mechanism, we give the following theorem and proof.
The greater the sum of responsibility and availability, the greater the possibility of a node to become a cluster head.
The essence of a “cluster” in the clustering routing is similar to the “cluster analyze” in the associated cluster algorithms of data mining, thus the CH election likes the procedure of determining the cluster centroids in the algorithms of data mining. Moreover, the well-known AP algorithm is an essential cluster algorithm, and its cluster procedure is carried out by the propagation of responsibility and availability, in which the first critical business is to adaptively determine the different cluster centroids according to the changeable orientation parameter. The cluster procedure is iteratively ongoing until all of the data items are clustered. This above idea is equivalent to the mechanism of the CH rotation in WSNs and equivalently applied in selecting the CHs; that is, there exists a set of dynamic candidate CHs composed of a large amount of nodes, which correspond to the cluster centroids and are traversed by the propagation of responsibility and availability. In each cycle of the clustering, the node-joint links build the interpath within a single cluster and the cluster head-joint links set up the intrapath between different clusters. During the propagation processing, the availability and responsibility denote the capability of resisting nodes failures and link losses, as well as their propagation direction which guides the data transmission paths. Obviously, the greater the sum of responsibility and availability is, the greater the probability that a node becomes a CH would be. The theorem is proved.
Consequently, in a real clustering phase, each cycle includes two phases: cluster head election and clustering. The two phases are periodically ongoing until any node runs out of its energy. The responsibility and availability transmit themselves within nodes-joint links in the clustering procedure. The sum of “
The propagation direction of responsibility and availability, as the multihop transmission path, effectively avoid the node failures.
Use reduction to absurdity to prove it. From the perspective of the definition of responsibility and availability, as long as a node fails in the network, its availability automatically gets “0” and the associated responsibility of it also becomes “0”; that is, the responsibility or availability does not continuously spread along such a path including these nodes. Thus the lemma is proved.
The termination condition for the propagation of responsibility and availability is the energy exhaustion of any node in the network.
As it can be seen from the aforementioned Definitions
Evidently, the aforementioned description is a global election scheme for CH rotation. Namely, all of the running nodes also have the same opportunity to become candidate CHs regardless whether they used to be CHs or not. Undoubtedly, this scheme is helpful to achieve balance in energy consumption of each node and prevent the premature death of any node, thereby guaranteeing coverage preservation and prolonging the network lifetime.
In this section, we talk about the proposed EEMCR routing scheme in detail. Exactly, EEMCR scheme inherits the basic framework of LEACH algorithm, which hybrids the basic processing of clustering and data transmission within each cycle. In fact, EEMCR scheme integrates several subalgorithms on CH rotation and backbone construction; such hybridization subalgorithms are ongoing in an iteration mode and they finally complete multihop clustering routing. The details of each subalgorithm are discussed in the following parts, respectively.
The EEMCR conducts the clustering in the mechanism of iteration cycle by cycle, which is similar to that in LEACH algorithm. However, the LEACH protocol randomly replaces the CHs in each cycle of iteration; in contrast, the EEMCR method adaptively updates the CHs according to the accumulated evidence of each node. In the initial stage of each cycle for the proposed EEMCR, the base station calculates the orientation-tended matrix of the nodes in terms of the Formula (
Summarizing the above steps describes the clustering algorithm, as shown in Algorithm
The time complexity of clustering algorithm is
The termination condition of Algorithm
Backbone construction essentially consists of generating minimum aggregation tree and minimum aggregation tree-based multihop communication paths according to graph theory. On the other hand, the multihop transmission path selection problem is very complex. The following theorem is given firstly.
Multihop routing with respect to the cluster heads in WSNs is an NP-complete problem.
At the stage of building the minimum aggregation tree, the energy consumption of each CH depends on its EC (e.g., Definition
Additionally, the well-known Kruskal algorithm in the field of graph theory is suitable for solving the NP-complete problems [
Furthermore, the construction of multihop communication paths within the candidate CHs is as follows: first, using the proposed Algorithm
In summary, backbone construction algorithm is presented, as shown in Algorithm
The generated tree, conducted by the Kruskal-based method, is a minimum aggregation tree.
Since the typical Kruskal algorithm targets to find an undistorted minimum aggregation tree, thus we use the reduction to absurdity to prove the Theorem
Assuming that there exists a really minimum aggregation tree
Based on the above analysis and proof, under the consideration of building a new tree denoted as
Hybridizing the aforementioned Algorithm
The time complexity of EEMCR scheme is
Assuming the connected graph
In summary, since the time complexity of clustering is
In this section, several experiments are arranged to validate the effectiveness and efficiency of the proposed method from different aspects. The parameters configured in the simulation experiments are shown in Table
Parameters in the experiments.
Parameter type | Parameter value |
---|---|
Node distribution area | 100 m × 100 m |
Sink location | (150, 50) |
Node numbers | 100~850 |
Initial energy of sensor node |
0.3 J |
Circuit processing data consumption energy |
50 nJ/bit |
Packet length | 2000 bits |
Control packet length | 32 bits |
|
100 pJ/bit/m2 |
|
0.0013 pJ/bit/m4 |
Data aggregation energy consumption | 5 nJ/bit/signal |
Data aggregation rate | 0.6 |
|
0 |
|
1000 |
|
0.9 |
Figures
The clustering results.
The clustering results using EEMCR
The clustering results using LEACH
Furthermore, to validate the influence of the proposed clustering and backbone construction algorithm on the performance, we fundamentally implement the LEACH [
Comparisons of the five algorithms.
Name | FirstDeath | LifeTime | MaxEC | MinEC |
---|---|---|---|---|
LEACH | 24 | 301 | 0.0167 | 0.0011 |
LEACH-CS | 65 | 377 | 0.0074 | 0.0015 |
LEACH-C | 43 | 345 | 0.0141 | 0.0007 |
LEACH-MT | 36 | 326 | 0.0117 | 0.0009 |
CMRAOL | 33 | 313 | 0.0150 | 0.0021 |
Regarding the first emergence of death nodes in Table
The network lifetime mainly reflects the capability of different algorithms to allocate the energy and schedule the task of data transmission. As shown in Table
The maximum average energy consumption and minimum average energy consumption of the nodes is analyzed. The difference between the maximum average energy consumption and minimum average energy consumption for the nodes is regarded as the index, a larger difference indicates the more energy consumption is centralized on a few nodes with a large amount of workload; in contrast, a smaller difference indicates the energy consumption and its distribution is more reasonable and uniform, which benefits the coverage preservation. From Table
In summary, the hybridization LEACH-CS method shows the best performance in terms of the indices LifeTime, FirstDeath, MaxEC, and MinEC. The experimental results indicate that the proposed cluster head rotation mechanism effectively limits the transmission range and set of neighbors of nodes.
In this part, experimental comparisons and analyses are performed between the EEMCR method and other methods, such as the LEACH-developed algorithm including BM-ELBC [
Usually, the emergence of death nodes is later and number of death nodes is less, which indicates higher coverage preservation, less communication holes, better energy efficiency, and higher QoS.
The death nodes versus the cycle of clustering.
Additionally, EBAPC algorithm is not the algorithm whose emergence of death nodes is the earliest one, but its uptrend gradient is the largest one; that is, its ratio of death nodes is the highest. The primary reasons which caused this phenomenon are that the CHs have to undertake more data forwarding tasks after the cluster formation, which spur a fast energy consumption on some nodes; thus it is not even good as that of LEACH algorithm.
The residual energy versus the cycle of clustering.
The average energy consumption of different nodes.
The influence of base station location on the survival time.
The emergence of the first death node versus the changing location of the sink
The survival time versus the changing location of the sink node
The influence of the scale on algorithms.
Usually, as the scale of the network increases, that is, the number of nodes in the network becomes larger, the workload also increases accordingly, and thus the CH election gets more critical. Figure
Additionally, besides the LEACH algorithm has a little fluctuation and ELBC has an abrupt change, the lifetime of the network generated by the remainder three algorithms shows more or less uptrend. The reason causing the changings in LEACH algorithm is the unreasonable CHs election mechanism. For ELBC algorithm, it has a relatively longer lifetime when the number of nodes is less than 200, its lifetime even reaches 621 within the size of 50 of nodes, and then the network lifetime performs stably and remains at around 390, which indicates that the ELBC algorithm only has a better adaptability to the small-scale network. The reason covered is primarily that the ELBC algorithm delivers data directly to the base station after the cluster formation, which makes it easy to encounter the bottleneck problem of data transmission such as data collision.
Based on the above experimental analyses and comparisons, the effectiveness of the proposed EEMCR scheme is convincing, and its efficiency is promising.
In the light of network lifetime, coverage preservation, transmission reliability, and energy consumption, a novel EEMCR scheme has been proposed. In EEMCR scheme, the guaranteeing capability of a link or a node of data transmission is comprehensively considered. The support of nodes themselves and node-joint links to data transmission are abstracted as a propagation of responsibility and availability, that is, the accumulated evidence. The responsibility and availability integrate considerations on the residual energy of nodes, CH location, energy consumption on the selected communication path and distance between nodes. The presented cluster head rotation mechanism makes a globally reasonable CH distribution, proper size of clustering, and uniform allocation of energy consumption. On the other hand, backbone construction algorithm is employed to build the minimum aggregation tree within the candidate CHs, and the generated minimum aggregation tree provides an energy-efficient multihop routing to guarantee the optimal and shortest path between the CHs to the base station. This scheme can effectively improve coverage preservation, save energy, and even degrade the link redundancy. Although the EEMCR scheme inherits the infrastructure of LEACH algorithm, the theoretical analysis and empirical results demonstrate the promising performance over the LEACH algorithms. Regarding emergence of the first death node, equilibrium of energy assignment, rationality of CH distribution, and convergence rate of the proposed algorithm, our scheme is promising. The proposed scheme is applicable for the large-scale WSNs. In the future works, the performance of our scheme could be evaluated when embedded in the design of new routing protocols. Beyond that, various kinds of protocols design and topology construction methods based on the presented innovations for different layer of WSNs would be studied.
The authors declare that they have no conflicts of interest.
This work is supported by the Future Research Projects Funds for the Science and Technology Department of Jiangsu Province (Grant no. BY2013015-23) and the Fundamental Research Funds for the Ministry of Education (Grant no. JUSRP211A 41).