^{1}

^{2}

^{★}

^{2}

^{3}

^{3}

This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).

We propose a stable backbone tree construction algorithm using multi-hop clusters for wireless sensor networks (WSNs). The hierarchical cluster structure has advantages in data fusion and aggregation. Energy consumption can be decreased by managing nodes with cluster heads. Backbone nodes, which are responsible for performing and managing multi-hop communication, can reduce the communication overhead such as control traffic and minimize the number of active nodes. Previous backbone construction algorithms, such as Hierarchical Cluster-based Data Dissemination (HCDD) and Multicluster, Mobile, Multimedia radio network (MMM), consume energy quickly. They are designed without regard to appropriate factors such as residual energy and degree (the number of connections or edges to other nodes) of a node for WSNs. Thus, the network is quickly disconnected or has to reconstruct a backbone. We propose a distributed algorithm to create a stable backbone by selecting the nodes with higher energy or degree as the cluster heads. This increases the overall network lifetime. Moreover, the proposed method balances energy consumption by distributing the traffic load among nodes around the cluster head. In the simulation, the proposed scheme outperforms previous clustering schemes in terms of the average and the standard deviation of residual energy or degree of backbone nodes, the average residual energy of backbone nodes after disseminating the sensed data, and the network lifetime.

Wireless sensor networks (WSNs) have a wide range of potential applications, including environment monitoring, military surveillance, and remote medical systems [

Much of the research on energy efficient routing for WSNs has several drawbacks. A single-hop clustering algorithm [

Hierarchical Cluster-based Data Dissemination (HCDD) [

The Max-Min

This paper is organized as follows: in Section 2, the major routing techniques are introduced and discussed. Section 3 presents our stable backbone formation scheme for WSNs. The performance of MCBT is evaluated in Section 4. Finally, we conclude with the main findings and contributions of our research in the last section.

In the single-hop communication model, each sensor node communicates with the cluster head within a single-hop, and the cluster head transfers the sensed data directly to the sink (base station). Low Energy Adaptive Clustering Hierarchy (LEACH) [

The multi-hop communication model [

Routing through a backbone is restricted in a set of particular nodes to reduce the communication overhead for route discovery and the number of active nodes for WSNs. A typical approach for backbone formation is to partition the network into clusters consisted of cluster heads and ordinary nodes. The cluster heads are then linked to form the connected backbone. Several approaches have been presented to select the cluster heads and construct the clusters. The optimal selection of cluster heads is an NP-hard problem [

Such protocols include (i) highest connectivity cluster algorithm and (ii) highest-ID or lowest-ID cluster algorithm. The highest connectivity cluster algorithm of MMM [

HCDD applies the backbone construction algorithm for wireless ad hoc networks, such as Max-Min

Backbone nodes consume more energy than ordinary nodes. It is important to form a stable backbone by taking into account the node’s residual energy and degree in order to prolong the network lifetime. MCBT constructs a stable backbone by selecting nodes with higher energy or degree as the cluster heads and distributes the role of packet forwarding among nodes around the cluster heads to enhance the network lifetime. In this paper, we consider a new factor: the flooding value. Using the flooding value, we can simultaneously consider the residual energy and degree, which are the core factors to evaluate the stability of the sensor node. If the node has high residual energy and degree, it has a high flooding value.

Each node exchanges the flooding values with its neighbors for 2_{i}

Let _{ini}_{res}_{i}_{j∈Si}_{j}_{ini}_{res}

The proposed cluster formation consists of three steps. First, each node calculates its own flooding value in round 0, as mentioned in the previous subsection, and maintains the (2^{st}^{th}^{th}^{th}

^{th}

^{th}

In Floodmax, the nodes with the higher flooding value propagate their flooding value in the 0^{th}

^{th}^{st}^{th}^{th}

After Floodmax and Floodmin, the node declares itself a cluster head or selects another node as a cluster head using three rules of cluster head selection.

^{th}

^{th}

After cluster head selection, each node broadcasts its elected cluster head’s ID to all of its neighbors. If a node receives a different cluster head ID, this node becomes a cluster gateway. Finally, in the case of a cluster head that is on the path between an ordinary node and its elected cluster head, the ordinary node chooses the cluster head with the minimum hops.

Applying the first rule, if the flooding value selected in Floodmax-^{th}^{th}^{th}^{th}

Other nodes select their cluster head by the second and the third rule. Applying the second rule, the node selects one node among cluster heads within ^{th}^{th}^{th}

MCBT Clustering Algorithm

_{res}

// FloodingArray_{i}_{i} | |

_{res}, d | |

| |

where _{i} | |

FloodingArray_{i} | |

// Floodmax Phase | |

| |

_{i} | |

// A node broadcasts its (^{th} | |

// all neighbors’ (^{th} | |

| |

_{i}_{j∈Si}_{j} | |

// Floodmin Phase | |

| |

_{i} | |

| |

_{i}_{j∈Si} (FloodingArray_{j} | |

// Cluster head Selection | |

| |

_{i}_{i} | |

_{i} | |

| |

_{i} | |

_{i}_{i} | |

| |

_{id}_{id} | |

_{id} | |

_{i} | |

| |

The pseudo code of MCBT clustering algorithm is shown above. Lines 1–4 show that each node computes its flooding value using the residual energy and degree, and makes a (2^{th}^{th}

If the flooding value selected in Floodmin-2^{th}

Consider node ^{th}^{th}

We simulate our algorithm to confirm that the backbone is composed of nodes with higher energy and degree. The parameters of the simulation are shown in

This weight policy can be implemented to optimize in various environments of WSNs by changing the weight factor,

Via the ns-2 simulator, we implement and compare the performance of MCBT with other backbone construction schemes, i.e., HCDD and MMM. Their information is listed in

The main objective of our scheme is to create a stable backbone with the node’s residual energy and degree taken into account, and each node transmits packets through this stable backbone. Thus, we evaluate the performance of the backbone construction scheme in terms of the average residual energy or degree of backbone nodes, the standard deviation of the residual energy or degree of backbone nodes, the average residual energy of backbone nodes after event occurrence, and the network lifetime.

The residual energy or degree of the backbone nodes shows the stability of the backbone. The demands on the backbone nodes are larger than those of ordinary nodes and nodes around the cluster head consume more energy in order to forward packets. However, by selecting a node with higher energy or degree, the high energy consumption of specific nodes should be alleviated.

Assume that the number of backbone nodes in the network is

We can evaluate the network lifetime by measuring the operational time until the backbone is disconnected or by counting the maximum number of transmitted messages from source to sink. This indirectly indicates how long a backbone is maintained before its reconstruction. Reconstruction overhead places a heavy load on the network, and it is important to ensure the backbone is stable and maintained for a long time. The time can be affected by variable factors, such as routing paths and data transfer collisions. Thus, in this paper, we compare the network lifetime with other algorithms by evaluating the maximum number of transmitted messages.

The main parameters of our simulation are listed in _{tx}_{11} + _{2}^{n}, E_{rx}_{12}, where _{tx}_{rx}_{11} is the energy/bit consumed by the transmitter electronics. _{2} is the energy dissipated in the transmit op-amp and _{12} is the energy/bit consumed by the receiver electronics.

In this paper, we propose a stable backbone tree construction algorithm using multi-hop clusters for WSNs. MCBT constructs the backbone by considering the node’s residual energy and degree. Since the backbone nodes have extra functionality and consume more energy compared with other nodes in the network, the construction of a stable backbone by selecting nodes with higher energy as the cluster heads could prolong the network lifetime. Nodes around the cluster head also consume more energy to forward packets, thus selecting the node with higher degree could distribute the load for packet forwarding among the nodes near the cluster head and balance the energy consumption. Moreover, MCBT could create an appropriate backbone for the system by adjusting the weight factor. Simulations show that MCBT performs better than existing schemes, in terms of the average and the standard deviation of residual energy or degree of backbone nodes, the average residual energy of backbone nodes after event occurrence, and the network lifetime.

This research was supported by MKE, Korea under ITRC IITA-2009-(C1090-0902-0046), IITA-2009-(C1090-0902-0005); and, by the MEST, Korea under the WCU Program supervised by the KOSEF (No. R31-2008-000-10062-0).

2-hop clustering - Floodmax and Floodmin phase.

2-hop clustering - Resulting network topology.

Average and standard deviation of residual energy.

Average and standard deviation of degree.

The effect of weight factor (

Residual energy of backbone nodes.

Degree of backbone nodes.

Average residual energy after event occurrence.

Remaining energy of nodes after event occurrence.

Maximum number of transmission messages.

Three clustering schemes in simulation.

Schemes | Cluster head selection criteria | Time complexity |
---|---|---|

HCDD | ID | O ( |

MMM | Degree | O ( |

MCBT | Energy&Degree | O ( |

Simulation parameters.

Network size | 500 m × 500 m |

Transmission range | 28 m |

Initial energy | 2.5 J |

Data packet size | 500 bytes |

Control packet size | 15 bytes |

Energy consumption model _{tx} |
_{11} + _{2}^{n} |

Energy consumption model _{rx} |
_{12} |

_{11}_{12} |
80 nJ/bit |

_{2} |
1 pJ/bit/m^{2} |

2 |