About this Journal Submit a Manuscript Table of Contents
International Journal of Distributed Sensor Networks
Volume 2012 (2012), Article ID 397961, 10 pages
http://dx.doi.org/10.1155/2012/397961
Research Article

A Multiple-Dimensional Tree Routing Protocol for Multisink Wireless Sensor Networks Based on Ant Colony Optimization

School of Electronics and Information, Nantong University, Jiangsu, Nantong 226019, China

Received 13 January 2012; Revised 1 April 2012; Accepted 15 April 2012

Academic Editor: Shukui Zhang

Copyright © 2012 Hui Zhou et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Routing protocol is an important topic in the wireless sensor networks. For MultiSink wireless sensor networks, the routing protocol designs and implementations are more difficult due to the structure complexity. The paper deals with the problem of a multiple-dimensional tree routing protocol for multisink wireless sensor networks based on ant colony optimization. The proposed protocol is as follows: (1) listening mechanism is used to establish and maintain multidimensional tree routing topology; (2) taking into consideration hops, packet losses, retransmission, and delay account, a distributed ant colony algorithm is proposed. When nodes select routes in the data transmission, the algorithm is utilized to realize the real-time optimization by coordination between nodes. The simulation results show that the proposed protocol can realize the QoS optimization for multisink wireless sensor networks, and its performance is better than the routing protocol of minimum hop numbers.

1. Introduction

Multisink wireless sensor architecture networks have received more and more attention due to their advantages such as improving network throughput, balancing energy consumption, and prolonging network lifetime. Moreover, the reliability and robustness of networks are improved because multisink nodes increase the transmission routines of the sensor node information [1, 2].

Due to multiple sink nodes in multisink wireless sensor networks (multisink WSNs), the network topology is complex, which brings many difficulties to design and implement the network protocols. Currently, research on multisink WSNs is still insufficient, especially for the cooperation and quality of service (QoS) of multisink WSNs.

In this paper, the problem of a multiple dimensional tree routing protocol for multisink WSNs will be investigated based on listening and ant colony optimization (ACO), where multiple dimensional tree routing is defined in Section 3. Listening mechanism is first used to establish and maintain multidimensional tree routing topology in the proposed protocol. Then, a distributed ant colony algorithm is presented with the consideration of hops, packet losses, retransmission and delay.

The rest of this paper is organized as follows. (i) Section 2 states the related works about the multisink WSNs routing researches and the applications of the ant colony optimization in WSNs. (ii) Section 3 describes the multisink WSNs model. (iii) Section 4 introduces the routing establishment of the listening-based routes for the multisink WSNs and the routing selection based on the distributed ant colony optimization. (iv) Simulation experiments are performed and analyzed in Section 5. (v) Some concluding remarks are found in Section 6.

2. Related Works

2.1. Multisink Wireless Sensor Networks

Recently, research results of multisink WSNs routing have been reported in the literature [39]. In [3], Dubois-Ferrière et al. limited the sink node transmissions to deliver the query messages to the minimum number of the data collection nodes, using the Voronoi scoping algorithm. In [4], Ciciriello et al. proposed a scheme based on a periodic adaptation of the message routes, which could efficiently route data from multiple sources to multiple sinks. Min [5] proposed priority-based multisink routing protocol, which considered both the level of node energy and the routing energy, so that the energy consumption was balanced efficiently and the lifetime of the network was prolonged. In [6], Kawano and Miyazaki proposed a minimum multihop routing protocol (MMHR), aiming to minimize the communication hops between each sensor node and a sink node in wireless sensor networks with multiple sink nodes. In [7], the optimal multisink positioning and energy-efficient routing protocol showed that the method to choose the route can be attributed to the linear programming model in order to realize the best deployment of multiple sink nodes and optimize the throughput of the whole network. In [8], Kalantari and Mark took the partial differential equations of Maxwell to resolve the optimization problem of the multisink networks and proposed the partial differential equations protocol. In the opportunistic routing protocol proposed in [9], each node measures the received signal strength indication from sink nodes in order to calculate mobility gradient, information of both the best neighbor node and the best sink nodes. In [10], an efficient multiple sink transmission power control scheme is analyzed for a sink-centric cluster routing protocol in multiple sink wireless sensor networks. It is worth pointing out that some problems such as node coordination, balance of the communication load, and robustness of routing protocol have not been sufficiently investigated in the above works.

2.2. ACO-Based Routing Protocol for WSNs

ACO is a swarm intelligent algorithm which analogs the ant foraging and exchanges pheromones to optimize complex problems [1113]. Because of the inherent parallelism of ACO, it is appropriate to apply ACO to optimize wireless sensor networks [14]. ACO for single-sink WSNs has been investigated in the last decade [1518]. In [15], Zhang et al. proposed three new ant-routing algorithms to improve the performance of WSNs. In [16], the energy efficient routing algorithm based on ACO was designed to extend network lifetime by reducing communication overhead in path discovering. It was achieved by energy efficient paths, which were established by using fixed size ant agents and introducing energy and number of hops in pheromone update mechanism. Cai et al. [17] proposed ACO-based QoS routing, which was a reactive protocol that tries to cope with strict delay requirements, limited energy, and computational resources available at sensor nodes. Ant-based service-aware routing algorithm proposed in [18] was a QoS-aware routing protocol for multimedia sensor networks.

ACO for multiple sink WSNs has received attention recently [19, 20]. Kiri et al. [19] described a cluster-based data gathering scheme aimed to achieve reliability and scalability in WSNs. Since WSNs architecture with a single sink is not robust to energy depletion, the authors in [19] proposed a multisink WSNs in which the nodes can use an alternate sink in case of failure of the network. In [20], Paone et al. proposed a routing protocol for multisink WSNs with interesting properties: self organization, fault tolerance, and environmental adaptation, which was inspired by the well known behavior (in artificial life studies) of “slime mold.” However, some problems such as QoS and node coordination still need to be fully studied. This motivates the research of this paper.

3. Multisink Wireless Sensor Network Model

In this section, a multisink wireless sensor network model is provided under the following assumptions.

Assumption 1. Sink nodes and sensor nodes are deployed randomly and they cannot move.

Assumption 2. All the sink nodes have the same architecture, and so do the sensor nodes.

Assumption 3. Wireless channels are symmetrical, and the process of receiving and transmitting orientates all directions.

Assumption 4. Each node has its own ID address.

Definition 5 (one-dimensional tree routing). The wireless sensor network tree routing is said to be one-dimensional tree routing if in the WSNs with one sink node and sensor nodes; routing topology is tree-type structure where the sink node is its root and sensor nodes are elements.

Definition 6 (N-dimensional tree routing). The wireless sensor network tree routing is said to be N-dimensional tree routing if in the WSNs with sink nodes and sensor nodes; routing topology is tree-type structure which is composed of one-dimensional tree routings where sensor node belongs to .

The topology structure of the one-dimensional tree routing and the two-dimensional tree routing is shown in Figures 1(a) and 1(b) separately.

fig1
Figure 1: The wireless sensor network tree routing.

4. ACOMSR Protocol

In this section, a multiple dimensional tree routing protocol for multisink wireless sensor networks based on ant colony optimization (ACOMSR) will be proposed.

4.1. ACOMSR Description

The ACOMSR is mainly made up of two parts: (i) the establishment and maintenance of the multiple dimensional tree routing topology by means of listening, (ii) the route selection and the pheromone update based on the distributed ant colony optimization.

The protocol establishes multidimensional tree topology routing, and the dimensional number is the same of the sink number. Each sensor node establishes -dimensional routing tables, and every one-dimensional routing table takes sink ID, father node’s ID, link quality, load, hop numbers, and other information. Sensor nodes use ant colony optimization algorithm to select routes according to the information of the routing table before they send their data packet.

The corresponding variables are defined in Table 1, and the implementation of the proposed protocol is shown in Algorithm 1.

tab1
Table 1: Notation description.

alg1
Algorithm 1: Proposed main algorithm.

4.2. Establishment and Maintenance of ACOMSR

Listening-based minimum hop routing protocol adopts bottom-to-up approach to establish routes, forming a tree topology structure whose root node is the sink node. During the process of the routing establishment and maintenance, node broadcasts the routing request packets (RREQ). It is assumed that node has received the RREQ. If , where is hops of node to sinks , is hops of node to sink , node will send the routing reply packets (RREP) with broadcast. Then all nodes which are the neighbor nodes of node could receive RREP. These nodes will establish or update their own routing if they have more hops. Most nodes in the network establish or maintain routes only by listening RREP, thus the protocol has the advantage of lower overhead and faster routing. The packet frame is described by Figure 2.

397961.fig.002
Figure 2: Frame format of packet.

According to the listening mechanism, the process of routing establishment or maintenance can be shown in Algorithm 2.

alg2
Algorithm 2: Routing establishment and maintenance.

Remark 7. (i) Since there are sinks in multiple sink WSNs, -dimensional tree routing should be established. (ii) In order to avoid collision, nodes use CSMA at the MAC layer when sensor nodes send out packet. (iii) Transmission power control is used to improve the link quality [21]. (iv) Regular maintenance is also used in the routing protocol, and the maintenance period is set after the update of the routing messages.

4.3. The Routing Selection and the Pheromone Update Based on the Distributed Ant Colony Optimization

The basic idea of the ant colony optimization in multisink WSNs routing algorithm is to build mappings between routing protocol and ant colony optimization.

Path selection of multisink WSNs is treated as ant foraging, and then a distributed ant colony algorithm is designed. When sensor nodes send packets, the algorithm is used to choose a suitable route. The mapping between the multisink WSNs routing protocol and the ant colony search space is given in Table 2.

tab2
Table 2: Mapping relation between multisink WSNs routing and the ant colony optimization.

The variable descriptions of the distributed ant colony algorithm are shown in Table 3.

tab3
Table 3: Notation description of ACO.

Sensor nodes of multisink WSNs save the quality of links from them to all sink nodes, namely, pheromones when ants look for food sources. Sensor nodes choose routes according to the link quality, that is, ants select foraging path and food sources based on the pheromone concentration. When ants choose food sources at time t, the transition probability is

The pheromone can be presented by where is the initial pheromone.

Define the visibility as the reciprocal of the hop number from sensor node to sink node , which is given by

The initial pheromone from sensor node to sink node is given by where represents the initial pheromone when .

Define the pheromone update value associated with the link quality as follows: where is the weight of , and represent the influence size on for packet loss and retransmission, respectively.

In order to avoid the pheromone unlimited accumulation which caused the imbalance evaluation of the link quality, we define the pheromone volatile coefficient as follows where is the initial value of the pheromone volatile coefficient, is the basic number of the volatile coefficient, is the maximum value of the buffer in a sensor node, is a threshold of the average link load from sensor node to sink node , and denotes the average link load from sensor node to sink node , which is defined by

The implementation procedure of the distributed ant colony algorithm is displayed in Algorithm 3.

alg3
Algorithm 3: Distributed ant colony algorithm.

5. Performance Evaluation

In this section, a simulation example is illustrated to show the effectiveness of the proposed protocol. Moreover, results compared with the minimum multihop multisink routing protocol (MMHR) are also provided.

5.1. Simulation Environment

Our simulation environment consists of 200 sensor nodes and 4 sink nodes randomly deployed in a field of 7000 m * 7000 m. Table 4 shows the simulation parameters of the network. The parameters of distributed ant colony algorithm are shown in Table 5 which are selected by means of a lot of simulation.

tab4
Table 4: Network simulation parameters.
tab5
Table 5: Parameters of the distributed ACO.

We can analyze the performance of this proposed protocol by changing generate packet rate. Here, generate packet rate means the amount of packet the perception nodes produced in unit working time.

5.2. Simulation Results and Analysis

According to routing establishment and maintenance process given in Algorithm 2, the ACOMSR routing topology structure at the moment of 60 s simulation time shown by Figure 3 is obtained, where red represents the routing to sink 0, orange the routing to sink 1, blue the routing to sink 2, and green the routing to sink 3.

397961.fig.003
Figure 3: The topology structure of 4-sink network.

As displayed in Figure 4, when the packet generate rate (PGR) is 0.04, the average hop numbers differ from each other with the change of time in the 4-demensional routing. We can see from Figure 4 that the average hop number decreases step by step with the increase of time. This is the routing continuous optimization result from the minimum routing maintenance mechanism.

397961.fig.004
Figure 4: Average hop counts vary for 4-dimensional routing.

For the proposed protocol and MMHR [6], lost packet rates, packet retransmission rates, and average transmission delays at different PGR are shown by Figures 5, 6, and 7, respectively.

fig5
Figure 5: Lost packet rates at different PGR.
fig6
Figure 6: Packet retransmission rates at different PGR.
fig7
Figure 7: Average Buffer Usage at different PGR.

From Figure 5 it can be seen that ACOMSR outperforms MMHR in term of the packet loss rate. When the packet generate rate is 0.01, the difference between ACOMSR and MMHR is not obvious. When the packet generate rate is 0.03, 0.04, 0.05, or 0.07, the proposed protocol shows a better performance in reducing packet loss rate.

From Figure 6 it can be seen that ACOMSR outperforms MMHR in term of the packet retransmission rate. But the differences of the packet retransmission rate between the two protocols are not obvious when the packet generate rate is 0.01 and 0.07.

From Figure 7 it can be seen that ACOMSR outperforms MMHR in the average buffer usage of all sensor nodes. When the packet generate rate is 0.01, the difference between ACOMSR and MMHR is basically the same. When the packet generate rate is 0.03, 0.04, 0.05 or 0.07, the proposed protocol shows a better performance in reducing average buffer usage.

From Figure 8, it is easy to see that the average transmission delays obtained by using ACOMSR are less than those obtained by using MMHR.

397961.fig.008
Figure 8: Average transmission delays at the different PGR.

The simulation results show that the proposed protocol ACOMSR has better performances than MMHR and the QoS optimization of multisink WSNs is achieved. The main reasons are two folds: (i) for selecting routing, MMHR takes the hop as the only performance index of selecting routes, whereas both hop numbers and QoS (including load balance, packet loss, and retransmission) in our protocol are simultaneously considered. The cooperation on the link information feedback and the nodes is achieved because of the interaction of the node link status information in DATA-ACK form. (ii) The introduction of the distributed ant colony algorithm makes the proposed routing protocol intelligent, and thus, a reasonable balance of network load is possible. Especially when the network load is heavy, the proposed protocol shows much better performance. It is worth pointing out that the real-time nature of the network transmission has been improved although the packet transmission route of the proposed protocol (ACOMSR) is not the shortest.

However, because of ACO algorithm, the computational overhead of the proposed protocol is more than the MMHR. With the case of 4 Sinks, each time the computation overhead of ACOMSR in a node is much more than that of MMHR’s about 21 times multiplications. But sensor node can be perfectly qualified for this overhead. Taking the microprocessor LPC2131 of Philips Corp as an example, the microprocessor comes with a hardware multiplier so that it only costs 21 instruction cycles, which is about 2 ms to complete these multiplications. Therefore, the proposed protocol is feasible in actual application.

6. Conclusion

A multiple dimensional tree routing protocol for multisink WSNs based on listening and ant colony optimization has been proposed in this paper. The advantages of the proposed protocol can be summarized as follows. (i) In the process of the routing establishment and maintenance, the waste of resources is avoided and the reliability of routing is improved by utilizing the listening mechanism and the power control, respectively. (ii) The fault tolerance and robustness of routing are increased because multidimensional tree routes from each sensor node to all sink nodes are set up. (iii) The QoS optimization of multisink WSNs is achieved by using the proposed ACOMSR. Simulation experiments have been made to show that the performance of the proposed ACOMSR is better than the routing protocol of minimum hop numbers. As a future work, we are intended to study the cross-layer optimization and multiobjective optimization for the multisink wireless sensor networks based on the ant colony algorithm.

Acknowledgment

This work was supported by National Natural Science Foundation of China under Grants nos. 61174065, 61071086, and 60901041.

References

  1. M. Ayaz, I. Baig, A. Abdullah, and I. Faye, “A survey on routing techniques in underwater wireless sensor networks,” Journal of Network and Computer Applications, vol. 34, no. 6, pp. 1908–1927, 2011. View at Publisher · View at Google Scholar · View at Scopus
  2. Y. Yang, M. I. Fonoage, and M. Cardei, “Improving network lifetime with mobile wireless sensor networks,” Computer Communications, vol. 33, no. 4, pp. 409–419, 2010. View at Publisher · View at Google Scholar · View at Scopus
  3. H. Dubois-Ferrière, E. Deborah, and S. Thanos, “Efficient and practical query scoping in sensor networks,” in Proceedings of the IEEE International Conference on Mobile Ad-Hoc and Sensor Systems, pp. 564–566, Lauderdale, Fla, USA, October 2004. View at Scopus
  4. P. Ciciriello, M. LucaPicco, and G. Pietro, “Efficient routing from multiple sources to multiple sinks in wireless sensor networks,” in Proceedings of the 4th European Conference on Wireless Sensor Networks (EWSN '07), pp. 34–50, Delft, The Netherlands, January 2007.
  5. M. Min, “PBR: priority based routing in multi-sink sensor networks,” in Proceedings of the Conference on Wireless Sensor Networks, pp. 25–28, Las Vegas, Nev, USA, 2007.
  6. R. Kawano and T. Miyazaki, “Distributed data aggregation in multi-sink sensor networks using a graph coloring algorithm,” in Proceedings of the 22nd International Conference on Advanced Information Networking and Applications Workshops/Symposia (AINA '08), pp. 934–940, March 2008. View at Publisher · View at Google Scholar · View at Scopus
  7. H. Kim, Y. Seok, N. Choi, Y. Choi, and T. Kwon, “Optimal multi-sink positioning and energy-efficient routing in wireless sensor networks,” in Proceedings of the International Conference on Information Networking (ICOIN '05), pp. 264–274, Jeju Island, Korea, February 2005. View at Scopus
  8. M. Kalantari and S. Mark, “Design optimization of multi-sink sensor networks by analogy to electrostatic theory,” in Proceedings of the IEEE Wireless Communications and Networking Conference (WCNC '06), pp. 431–438, Las Vegas, Nev, USA, April 2006. View at Scopus
  9. A. Lukosius, Opportunistic Routing in Multi-Sink Mobile Ad Hoc Wireless Sensor Networks, University of Bremen, Bremen, Germany, 2007.
  10. L. Cao, C. Xu, W. Shao et al., “Distributed power allocation for sink-centric clusters in multiple sink wireless sensor networks,” Sensors, vol. 10, no. 3, pp. 2003–2026, 2010. View at Publisher · View at Google Scholar · View at Scopus
  11. M. Dorigo, V. Maniezzo, and A. Colorni, “Ant system: optimization by a colony of cooperating agents,” IEEE Transactions on Systems, Man, and Cybernetics B, vol. 26, no. 1, pp. 29–41, 1996. View at Scopus
  12. M. Dorigo and T. Stützle, Ant Colony Optimization, MIT Press, Cambridge, UK, 2004.
  13. N. Jiang, R. G. Zhou, and S. Q. Yang, “An improved ant colony broadcasting algorithm for wireless sensor networks,” International Journal of Distributed Sensor Networks, vol. 5, no. 1, pp. 45–45, 2009. View at Publisher · View at Google Scholar
  14. M. Saleem, G. A. Di Caro, and M. Farooq, “Swarm intelligence based routing protocol for wireless sensor networks: survey and future directions,” Information Sciences, vol. 181, no. 20, pp. 4597–4624, 2011. View at Publisher · View at Google Scholar · View at Scopus
  15. Y. Zhang, L. Kuhn, and M. Fromherz, “Improvements on ant routing for sensor networks,” in Proceedings of the 4th International Workshop on Ant Colony Optimization and Swarm Intelligence (ANTS '04), pp. 154–165, Brussels, Belgium, September 2004.
  16. T. Camilo, C. Carreto, J. S. Silva, and F. Boavida, “An energy-efficient ant-based routing algorithm for Wireless Sensor Networks,” in Proceedings of the 5th International Workshop—Ant Colony Optimization and Swarm Intelligence (ANTS '06), pp. 49–59, Brussels, Belgium, September 2006.
  17. W. Cai, X. Jin, Y. Zhang, K. Chen, and R. Wang, “ACO based QoS routing algorithm for wireless sensor networks,” in Proceedings of the 3rd International Conference on Ubiquitous Intelligence and Computing (UIC '06), pp. 419–428, Wuhan, China, September 2006.
  18. Y. Sun, H. D. Ma, and L. Liu, “Ant-colony optimization based service aware routing algorithm for multimedia sensor networks,” Tien Tzu Hsueh Pao/Acta Electronica Sinica, vol. 35, no. 4, pp. 705–711, 2007. View at Scopus
  19. Y. Kiri, M. Sugano, and M. Murata, “Self-organized data-gathering scheme for multi-sink sensor networks inspired by swarm intelligence,” in Proceedings of the 1st International Conference on Self-Adaptive and Self-Organizing Systems (SASO '07), pp. 161–170, Cambridge, Mass, USA, July 2007. View at Publisher · View at Google Scholar · View at Scopus
  20. M. Paone, L. Paladina, M. Scarpa, and A. Puliafito, “A multi-sink swarm-based routing protocol for wireless sensor networks,” in Proceedings of the IEEE Symposium on Computers and Communications 2009 (ISCC '09), pp. 28–33, Sousse, Tunisia, July 2009. View at Publisher · View at Google Scholar · View at Scopus
  21. S. L. Zhu, C. Xu, Q. Sun, and X. Huang, “A novel transmission power selection mechanism for wireless sensor networks,” in Proceedings of the International Conference on Electronics, Communications and Control (ICECC '11), pp. 661–665, Ningbo, China, September 2011.