HCABS: The hierarchical clustering algorithm ... - Semantic Scholar

2 downloads 0 Views 680KB Size Report
IEICE Electronics Express, Vol.9, No.7, 685–690. HCABS: The hierarchical ... hierarchy with deterministic cluster head selection,” Proc. IEEE MWCN,. 2002.
IEICE Electronics Express, Vol.9, No.7, 685–690

HCABS: The hierarchical clustering algorithm based on soft threshold and cluster member bounds for wireless sensor networks Mehdi Golsorkhtabaramiri1a) , Mehdi Hosseinzadeh1 , Ali Golsorkhtabaramiri2 , and Saeed Rasouli Heikalabad1 1

Department of Technical and Engineering, Science and Research Branch, Islamic

Azad University, Tehran, Iran 2

Mazandaran University of Science and Technology, Babol, Iran

a) [email protected]

Abstract: Wireless Sensor Networks (WSNs) for reducing energy consumption and increasing sensors lifetime can use the clustering algorithms. We propose a new energy-efficient hierarchical clustering algorithm based on soft threshold cluster-head election and cluster member bounds for WSNs which called HCABS. Our simulation studies suggest that HCABS achieves longer lifespan and reduce energy consumption in WSNs as well as low latency and moderate overhead across the network. Keywords: sensor networks, clustering algorithm, energy-efficient, member bounds Classification: Wireless circuits and devices References

c 

IEICE 2012

DOI: 10.1587/elex.9.685 Received February 23, 2012 Accepted March 26, 2012 Published April 11, 2012

[1] M. Golsorkhtabar, M. Hosinzadeh, M. Heydari, and S. Rasouli, “New Power Aware Energy Adaptive protocol with Hierarchical Clustering for WSN,” (IJCNS) International Journal of Computer and Network Security, vol. 2, no. 4, April 2010. [2] S. R. Heikalabad, A. Navin, M. Mirnia, S. Ebadi, and M. Golesorkhtabar, “EBDHR: Energy Balancing and Dynamic Hierarchical Routing algorithm for wireless sensor networks,” IEICE Electron. Express, vol. 7, no. 15, pp. 1112–1118, 2010. [3] M. G. Amiri, N. Rahmani, and S. Ebadi, “FTEAP: A Fault Tolerant Energy Adaptive and Power Aware Clustering Protocol for Wireless Sensor Networks,” (JGRCS) J. Global Research in Computer Science, vol. 1, no. 1, Aug. 2010. [4] M. Golsorkhtabar, F. K. Nia, M. Hosseinzadeh, and Y. Vejdanparast, “The Novel Energy Adaptive protocol for Heterogeneous Wireless Sensor Networks,” Proc. IEEE ICCSIT 2010, Chengdu, China, July 2010. [5] M. J. Handy, M. Haase, and D. Timmermann, “Low energy clustering hierarchy with deterministic cluster head selection,” Proc. IEEE MWCN, 2002. [6] W. R. Heinzelman, A. P. Chandrakasan, and H. Balakrishnan, “An

685

IEICE Electronics Express, Vol.9, No.7, 685–690

application-specific protocol architecture for wireless microsensor networks,” IEEE Trans. Wireless Commun., vol. 1, no. 4, pp. 660–670, 2002. [7] R. Ding, B. Yang, L. Yang, and J. Wang, “Soft Threshold Based Clusterhead Selection Algorithm for Wireless Sensor Networks,” 2009 Third International Conference on Sensor Technologies and Applications. [8] D. Li, K.Wong, Y. Hu, and A. Sayeed, “Detection, classification, tracking of targets in micro-sensor net-works,” IEEE Signal Process. Mag., pp. 17– 29, March 2002. [9] W. Ye and J. Heidemann, “Medium access control in wireless sensor networks,” Wireless sensor networks, pp. 73–91, 2004.

1

Introduction

Smart sensor nodes are devices supplied with a micro processor, memory, power supply, transition unit and one or more sensors. In order to obtain an objective scalability of network, gathering sensor nodes into clusters has been greatly pursued by the research community [1, 2]. Each cluster would have a head, in many cases referred to the clusterhead (CH). A CH may be elected by the sensors in a cluster or pre-assigned by the network designer [3, 4]. In this paper, we study the performance of clustering algorithm based on soft threshold and cluster member bounds for WSNs. The rest of the paper is organized as follow. In Section 2, related work is described. Section 3 presents the details of HCABS protocol. In Section 4 we compare our clustering approach with the LEACH [6] clustering protocol as one representative from a group of low-energy adaptive clustering and two other protocols. Finally, in Section 5, we conclude this work and provide directions for our future work.

2

Related works

LEACH-E [5] is proposed to elect the CHs according to the energy remaining in each node. This protocol extend LEACH’s stochastic cluster selection algorithm by a deterministic component. In STCS against the LEACH, once a node after has been selected as a CH, its threshold will not be set to 0, and thus it will not lose the chance to participate CH selection [7]. In the rest of this section, we review LEACH algorithm and discuss its limitations, due LEACH very popular in wireless sensor network clustering protocols. In first phase, algorithm chooses a node stochastically, the principal as explained in the coming: every sensor nodes generate a random number between 0 and 1, if the random number is lower than threshold, and then will be chosen as CH, otherwise, cannot be CH. And threshold calculation by (1): c 

IEICE 2012

DOI: 10.1587/elex.9.685 Received February 23, 2012 Accepted March 26, 2012 Published April 11, 2012

686

IEICE Electronics Express, Vol.9, No.7, 685–690

⎧ ⎨

p 1 − p(r mod (1/p)) T (n) = ⎩ 0

n∈G others

(1)

Where in this equation P = the desired percentage of CHs (e.g. P = 0.05) the current round, and G is the set of nodes that have not been CHs in the last 1/P rounds. In a real WSNs scenario, from the first round or in some of next rounds, sensor nodes distribution is heterogeneously and they are dense in part of given area. For example, in Fig. 1-a in which most of the nodes are grouped together close to some CHs for instance CH A and B. In the Fig. 1-a we assume, CH A and CH B are elected in CH selection phase and will broadcast to their neighbors and will end up having too many nodes in their cluster because in these cluster’s area there are too many sensor nodes. Due to LEACH does not have any limit on the number of nodes a CH can be accepted under it, these nodes will support too many members in their cluster and this would lead to a very fast depletion of energy in A and B and a part of the network would lose connectivity. Moreover, LEACH has no provision to account for distance from CHs to BS and movement of nodes or CHs during the network lifetime. If a CH stand far from BS it must consume more energy for sending data to BS. Therefore, the nodes that are near the BS can save more energy than other nodes. Fig. 1-b shows a network in this situation. Only the nodes near the BS are alive in end of LEACH rounds.

Fig. 1. Heterogeneous spread of nodes after LEACH rounds.

3

c 

IEICE 2012

The HCABS protocol

In HCABS the CHs are elected by a probability based on soft threshold. HCABS, for defining cluster members, determine a confidence value for any nodes that want to be a CH in each round.

DOI: 10.1587/elex.9.685 Received February 23, 2012 Accepted March 26, 2012 Published April 11, 2012

687

IEICE Electronics Express, Vol.9, No.7, 685–690

3.1 Cluster head election algorithm based on soft threshold In the HCABS protocol, CHs are also chosen according to the probability, but based on STCS algorithm. The LEACH algorithm mentions that, after a node has been selected as a CH once, the threshold in its CH selection will be changed to 0 and this node cannot be CH again, even if it still has enough energy. In our algorithm, the probability of a node to become a CH is also determined by a specific threshold. In this algorithm, when a node has acted as a CH once, threshold will be adjusted step-by-step instead of being changed to 0 directly. Therefore the threshold T (i) is set as:

T (i) |r =

⎧ ⎪ P ⎪ ⎪ ⎪ ⎪ ⎪ P2 ⎪ ⎪ ⎨ T (i) |r−1 − ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎩ T (i) |r−1 +

2

P2 2 N U M (C(i))

r=0 i ∈ Gr−1 (2) othererwise

Y |r = Σi=1 N xi |r denotes the number of CHs in the r th round, for each cluster, we can get [7]: T (Y |r ) = N

i=1

N i=1

1 ∗ T (i) |r =

1 ∗ T (i) |r−1 =

N

i=1

1 ∗ EY |r−1 ) = . . . = E(Y |0 ) = k

(3)

3.2 Cluster formation based on cluster member bounds In the formal description of HCABS, the proposed algorithm aims to rectify some of the loopholes left out by other algorithms. New algorithm adjusted on CH current battery power and a number of members currently under a CH are taken into account before a node decides which CH to attach. We suppose that all sensors have a processor, a memory and the hardware needed to execute sensing, information gathering and transferring. The main principle for an energy based approach is that received energy of a signal intensity decreases with propagation distance at a rate given by the equation [8, 9]: A x − xi (t) α x − xi −α , 1 ≤ i ≤ N

c 

IEICE 2012

DOI: 10.1587/elex.9.685 Received February 23, 2012 Accepted March 26, 2012 Published April 11, 2012

(4)

Where k x − xi  is the distance between nodes x and xi , A x − xi  is the attenuation for the distance the signal travels, and α(α ∈ [2, 5] ) is the path loss exponent. Our clustering model is based on confidence value associated with broadcasting from CHs. Confidence value of a CH is a function of some parameters: (1) distance between the CH and the node, (2) number of nodes already were a member of this CH, (3) the CH current battery power and (4) distance between the CH and the BS. Basically, our model checks first, if with current battery power of the CH, it would be able to support the current members at maximum data broadcasting rate. A node decides to join a CH if the head can still support the node with its rest power. Confidence value given by:

688

IEICE Electronics Express, Vol.9, No.7, 685–690

CV (i) =

⎧ ⎪ ⎨ 0 ⎪ ⎩

Bp Cm ∗ Dc ∗ Db

Bp < T (Bp) Bp > T (Bp)

(5)

Where in this equation BP is the battery power of given node, Cm is number of nodes already a member of given CH, Dc is distance between the CH and the node and Db is distance between the CH and the BS. T (Bp) is already the threshold of battery power to support Cm + 1 nodes of given CH.

4

Simulation results

HCABS, LEACH, LEACH-E and STSC are simulated with GCC and the simulation repeated for tow times with different simulation number of nodes to achieve the reliable results about proposed algorithm. We have considered a wireless sensor network with N = 100, 200 nodes dispersed with in a 300 m2 square field randomly. We assumed the BS to be fixed and located at the origin (0, 0) of the coordinate system. The network undergoes 1000 rounds in each run. The CHs perform data aggregation to reduce the redundancy before transmitting it to the BS. In order to compare with other algorithms, we also set the desired percentage of CHs P to 0.05. In simulation, we use the same radio model shown in [6] for the radio hardware energy dissipation. Fig. 2-a, 2-c presents the energy consumption of the clustering protocols. The results demonstrate that the energy consumption of HCABS is generally smaller than that of LEACH-E, STCS, and LEACH. Fig. 2-b, 2-d presents the number of alive nodes. This result has a significant relationship with the network lifetime. In the case of networks with 100 sensor nodes using HCABS, the time when the first node dies (FND) occurs after 209 rounds, whereas in the case of STCS, it occurs after 190 rounds. In the case of networks using LEACHE it occurs after 124 rounds and in LEACH, after 29 rounds. In the case of networks with 100 nodes using HCABS, the first event where half of the nodes run out of energy (HND) occurs after 637 rounds, whereas in the case Table I. Parameters used in simulations.

c 

IEICE 2012

DOI: 10.1587/elex.9.685 Received February 23, 2012 Accepted March 26, 2012 Published April 11, 2012

689

IEICE Electronics Express, Vol.9, No.7, 685–690

of STCS, it occurs after 467 rounds. In the case of networks using LEACH-E it occurs after 449 rounds and in LEACH, after 249 rounds. The results indicate that the most energy-saving protocol is HCABS, which performs better and keeps the sensor nodes alive for a longer number of rounds than all the other protocols.

Fig. 2. Simulation results.

5

Conclusion

In this paper, we adopt a way of adjusting the threshold and cluster formation to achieve our purpose of increasing the network lifetime. HCABS has better performance in CH election and forms adaptive power efficient and clustering hierarchy. The simulation results presented that HCABS significantly improves the lifespan and the energy consumption of the wireless sensor networks in comparison with existing clustering protocols. In order to further energy saving and extend the lifetime of the network, our future plans will involve how to optimize threshold more, based on other CH selection algorithms and adopt cluster forming by optimizer equations and considering more effective parameters.

c 

IEICE 2012

DOI: 10.1587/elex.9.685 Received February 23, 2012 Accepted March 26, 2012 Published April 11, 2012

690