Distributed Sensor Networks, Article ID 90636, 7 pages http://dx.doi.org/10.1155/014/90636 Research Article A Gathering Method Based on a Mobile Sink for Minimizing the Loss in Wireless Sensor Networks Junho Park, 1 Kyoungsoo Bok, Dongook Seong, Sanglae Kim, 3 and Jaesoo Yoo 1 Agency for Defense Development, Bugyuseong daero 488, Yuseong, Daejeon 305-600, Republic of Korea School of Information and Communication Engineering, Chungbuk National University, 5 Naesudong-ro, Heungdeok-gu, Cheongju 361-763, Republic of Korea 3 Software Center, Samsung Electronics Co., Ltd., Maetan 3-dong, Yeongtong-gu, Suwon 443-74, Republic of Korea Correspondence should be addressed to Jaesoo Yoo; yjs@chungbuk.ac.kr Received 6 December 013; Accepted 7 February 014; Published 17 April 014 Academic Editor: Jason J. Jung Copyright 014 Junho Park 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. The sensor nodes near to the fixed sink node suffer from the quickly exhausted energy. For this, many methods have been researched to distribute the energy consumption into all wireless sensor nodes using a mobile sink. Since the mobile sink changes its location in the network continuously, it has limited time to communicate with the sensor nodes and needs the time to move to each sensor node. Therefore, before the mobile sink approaches the sensor node, the node can collect huge data by event occurrence. It causes thememoryoverflowofthesensornodeandthenthedataloss.weproposeanoveldatagatheringmethodbasedonamobilesink considering the data loss in wireless sensor networks. The proposed scheme actively sends the stored data to the mobile sink by considering the amount of collected data in the cluster header and the mobile patterns of the sink node. By doing so, it minimizes the data loss of each sensor node. To show the superiority, we compare it with the existing scheme. Our experimental results show that our proposed scheme minimizes the data loss and has similar network lifetime over the existing scheme based on a mobile sink. 1. Introduction With the remarkable development of microelectronic device and wireless communication technologies, sensor nodes used for the sensor network have enabled miniaturization, low cost, and low power. The sensor networks provide information obtained by detection events on interesting targets by taking advantage of sensor nodes. These sensor networks can be utilized in various applications such as the environmental monitoring, building safety monitoring, military applications, home networking, and a variety of commercial and public areas [1, ]. The sensor nodes used for these sensor networks generally use small built-in batteries. The lifetime of the sensor network is determined by the duration time of the batteries of sensor nodes. Therefore, it is important to maintain the long network lifetime by minimizing the energy consumption of limited batteries in the sensor nodes. Basically, the sensor network has one or more sink node to deliver to a user by collecting sensing values by the sensor nodes. According to the data gathering method to collect the data in the sink nodes, the sensing values by the sensor nodes are delivered efficiently to a query node without any data-loss. Therefore, the data gathering method by the sink nodes is one of essential technologies. Various data gathering methods based on static sink nodes have been proposed. Figure 1 shows the sensor network based on static sink. It is classified into two categories such as cluster-based methods and tree-based methods. The cluster-based methods firstly select the header nodes based on certain conditions and then construct clusters that consist of nodes near to them. Because the cluster-based methods collect the sensing values from the member nodes and process the data in header nodes before sending the data to the sink node, they have the advantage thatenergycanbeusedequally. LEACH [3] and HEED [4] are the representative clusterbased methods. The tree-based methods create tree-structure based routing paths that a base station which ultimately collects data is a root node. The sensing values from each
Distributed Sensor Networks Static sink Anchor point Mobile sink Figure 1: Sensor network based on a static sink. Figure 3: collection of mobile sink based on anchor points. Mobile sink Moving path Figure : Sensor network based on a mobile sink. nodearetransmittedtothebasestationalongwiththe tree-structure based routing path. TAG [5] and Cougar [6] are representative tree-based methods. However, these data gathering methods based on a static sink have several weaknesses in order to apply them to practical applications of the sensor networks [7]. First, in a sensor network environment based on a static sink, the network lifetime is shortened by unbalanced energy consumption because of the frequent data transfer in nodes near to the sink node. Second, if the routing distance is longer, data-losses occur frequently. The sensor network deploys a lot of sensor nodes in a wide area and collects interesting environmental information. Due to the limitation of a communication range, sensor nodes use multihop routing to transmit the collected data to a sink node. In this multihop routing, if sensor nodes which intend to transfer data are far from the sink node, more nodes participate in the routing. According to the increase of the sensor nodes, it increases the probability of communication failure and error. To overcome this problem, the data gathering schemes using mobile sinks have been proposed [8 10]. As shown in Figure, a mobile sink collects data from each sensor node while it moves the whole network. Therefore, the mobile sink solves a problem of the concentrated energy consumption around the sink node and reduces the data-loss due to the multi-hop routing. However, the mobile sink has a limited communication time to gather the sensing values from the sensor nodes. To overcome the problem, the existing schemes collect the data uniformly from all the sensor nodes in the sensor network [11]. However, the size of an internal memory in a sensor node is still limited. Therefore, if a sensor node can collect huge data by event occurrence, it causes the memory overflow and the dataloss. Therefore, a data gathering scheme should be studied tominimizethedata-losswhileutilizingthemobilityofthe mobile sink. In this paper, we propose a novel data gathering method considering the data-loss in a wireless sensor network based on a mobile sink. The proposed method maximizes the balanced energy consumption in entire sensor nodes due to themobilityofthemobilesink.inaddition,thedata-lossin the routing path and in the hot-spot area is minimized. To this end, the cluster header nodes send the stored data to the mobile sink by considering the amount of collected data in each cluster header and the mobile patterns of the sink node. Toshowthesuperiorityofourproposedscheme,wecompare it with the existing method based on a mobile sink. The remainder of this paper is organized as follows. Section overviews the existing data collection method using the mobile sink and analyzes their problems. In Section 3, we present our data gathering method based on a mobile sink for minimizing the data-loss in wireless sensor networks. Section 4 performs the simulated experiments and compares the existing method with the proposed method. Finally, we present concluding remarks in Section 5.. Related Works In the existing data collection scheme using the mobile sink, while the mobile sink moves along the anchor points as shown in Figure 3, it collects the sensing data. The sensor network consists of various clusters with sensor nodes. The sensed data in each cluster are collected in the cluster header. The cluster header as an anchor point of a mobile sink is waiting for data collection by the mobile sink. Therefore, before the mobile sink approaches the cluster header, the cluster header can collect huge data. As a result, the memory overflow of the cluster header and the data loss can occur. The existing scheme considering the data-loss was proposed [1]. Reference [1] is composed of clusters as shown in Figure 4. After a cluster header of each cluster is first chosen, the mobile sink moves to the most efficient location
Distributed Sensor Networks 3 CH ID D max 1 18 8 3 1 4 0 Optimal position d tobs 0 1 14 Average D =.5 D 4 1 4 8 1 0 18 0 Figure 4: Optimal positioning of the mobile sink. for data transmission by the cluster header and collects data. Therefore, it reduces energy consumption and minimizes data loss caused by the memory overflow of the cluster header: D = d max d toch. (1) The cluster header informs the own location (d toch )and the communication range (d max ) to each cluster member. Formula (1) is used to obtain the optimal position of mobile sink. Each member node calculates the D of each cluster header using formula (1). And then each sensor node calculates the average of the D. The position that the average of the D is the lowest is energy-efficient and minimizes the data loss. The process is repeated during regular rounds determined according to the energy consumption of all nodes. The existing scheme uses the mobility of a mobile sink but causes data loss due to the routing path, the problem of static sink. As the number of nodes increases, the benefits of the mobile sink are reduced since the moving path becomes longer. Therefore, we propose a novel data gathering method considering the data-loss in a wireless sensor network based on a mobile sink. Our scheme reduces data-loss using the mobilityofthemobilesinkandmultihoproutingpath.italso minimizes data-loss that occurs due to limited cache memory of sensor nodes in the data hot spot. 3. The Proposed Gathering Method In the proposed scheme, the base station establishes the moving path between the mobile sink and each cluster header. And then, it builds the moving table for determining thelocationofthemobilesink.bygeneratingtheoptimal collection paths for the first data collection, it is possible to minimize data-loss by the lack of internal memory in the cluster headers at the typical data collection. In order to build an optimal collection path, the proposed scheme establishes the shortest collection path that collects the stored data in the cluster headers by the mobile sink by using a convex hull-based algorithm [7]. This algorithm builds the initial collection path based on the maximum radius of the circle. Next, it generates the optimal data collection path by adding the cluster headers in the circle to the initial path. Figure 5 shows the shortest moving path to collect the data. By using this shortest moving path of the mobile sink, it can minimize the data-loss in typical data collection. Figure 5: Shortest moving path of the mobile sink. Table 1: Structure of the moving path table. Head 1 Head Destination Time Cluster header (H1, H) Table:Themovingpathtableofthemobilesink. Distance (D) Moving time (T) C1, C 3 6.4 C, C3 30 6.0 C3, C4 77 15.4 C4, C5 1 4. C5, C6 54 10.8 C6, C7 70 14 C7, C1 34 6.8 As shown in Table 1, the moving path table of the mobile sink is built based on the locations of the cluster headers and shortest moving path of the mobile sink.table shows the moving path table of the mobile sink. The moving time between cluster headers (H1, H) is calculated using formula (). The base station generates a moving path table considering the location information of cluster headers and themovingspeedofthemobilesink.andthenthebase station distributes the generated table to each cluster header. By this piece information, the cluster headers can recognize thelocationofthemobilesinkandthedatacollectiontime: (H1 x H x ) +(H1 y H y ) T=. () Mobile Sink speed In this paper, the proposed scheme is based on the following sensor network environments. Assumption 1. On the moving path of the mobile sink, one or more sensor node exists within the communication range of themobilesink,andtheyarepossibletosendandreceivethe data. Assumption. The communication error does not occur when it sends and receives data between the mobile sink and sensor nodes.
4 Distributed Sensor Networks New data New data Event 1 Empty full loss New data full (a) collection in cluster header (b) -loss due to the lack of cache memory Figure 6: collection in a cluster header and data-loss. Partial transmission Empty Overflow preliminary threshold position of a mobile sink to the overflow cluster header. It is calculated as a GT (total time to traverse all cluster headers by themobilesink)andact(timetocollectdatabythemobile sink in the current round). DCR and RT are operated when the overflow event occurs. They determine the partial or full data transmission to the mobile sink by predicting whether the overflow occurs using formula (5): Full data transmission DCR ( Collecting Rate) = Collected (D (T ) D(T 1 )) Time (T T 1 ), (3) Figure 7: management of the proposed scheme. RT (Remaining Time) = GT (General Time) CT (Current Time), (4) In general situations, the data gathering scheme by the mobile sink collects the data based on the constructed moving path. When the cluster header is located in the communication range of the mobile sink during its movement, the mobile sink collects the stored data in the cluster header. However, when a query event occurs as shown in Figure 6, the cluster header receives the large data by shortening the transmission cycle of the detected node. Accordingly, due to the lack of cache memory in the cluster header, the data-loss can occur. To solve this problem, the proposed scheme uses the data transmission policy as follows. At the deployment of the sensor network, the cluster header sets the preliminary threshold as shown in Figure 7. Whenaqueryeventoccurs, the cluster member nodes transmit the sensing values to the clusterheader.whentheamountofthestoreddatareachesan overflow preliminary threshold in the cluster header, the data overflow event occurs. According to the data overflow event of the cluster header, the stored data in the cluster header are transmitted to the mobile sink. Formulas (3) (5) are computational models to determine thewaytotransmitthestoreddatainthecaseofdataoverflow events.thedcrofformula(3)representsthedatacollected intheheadernodeforaperiodoftime.thertofformula(4) also represents the remaining time to reach from the current DCR RT > Reserve Threshold. (5) If formula (5) is satisfied, it means that the data collection rate from the cluster member nodes is fast or the mobile sink is far from the cluster header. Therefore, the probability of the overflow occurrence is high until the mobile sink accesses the cluster header. In this case, the cluster header transmits theentirestoreddatatothemovingpathofthemobilesink. Figure 8 shows how the entire data are transmitted to the mobile sink. When the cluster header transmits the entire stored data, its memory is empty. Therefore, since the mobile sink is not necessary to visit the cluster header that already transmits the entire stored data, the proposed scheme updates the moving path of the mobile sink. If formula (5) is satisfied when the data overflow event occurs, the cluster header transmits only partial data as shown in Figure 9. By doing so, it minimizes the energy consumption and the data loss that occurs in the case of the transmission of the full data. The mobile sink does not traverse the updated path in the partial data transmission unlike the full data transmission as shown in Table 3. Therefore, if an additional overflow does not occur in the cluster header until the mobile sink accesses it again, the partial data transmission is more efficient.
Distributed Sensor Networks 5 1 Overflow event Packet transmission Moving the location of mobile sink Figure 8: transmission when an overflow occurs. 1 Packet transmission Figure 9: transfer failure according to the shortest moving path. Table 3: The moving path table of the mobile sink after the change. Cluster header (H1,H) Distance (D) Moving time (T) C1, C 3 6.4 C, C3 30 6.0 C3, C4 77 15.4 C4, C5 1 4. C5, C7 (new moving path) 85 17 C7, C1 34 6.8 4. Performance Evaluation Toshowthesuperiorityofourproposedmethod,wecompare it with an anchor-based mobile sink moving data collection method in various environments. We have developed a simulator based on JAVA to evaluate our proposed scheme and the existing schemes. The performance evaluation was carried out through the simulation parameters in Table 4.The transmitted data size to a cluster header from sensor nodes is {Size of Sensing value} {Number of Sensor Nodes}.We assume that 10,000 sensors are deployed uniformly in a 1,000 1,000 (m) network field. The energy consumption for sending a message is determined by a constant function S (C t +C a D ),wheres is the message size, C t is the transmission cost, C a is the amplification cost, and D is the distance of message transmission. We set C t =50nJ/b and C a = 100 pj/b/m in the simulation. The energy consumption for receiving a message is determined by a cost function (S C r ), where S is the message size and C r is the transmission cost. We set C r =50nJ/b in the simulation. The preliminary threshold of data overflow in cluster headers in the proposed scheme was set at the 450 Kbytes. Figure 10 shows the amount of the data-loss according to the number of sensor nodes. In the case of the existing scheme, as the number of sensor nodes increases, the number of the transmitted data increases. Therefore, the data-loss increases in the existing scheme. In contrast, when the stored
6 Distributed Sensor Networks Table 4: Simulation parameters. Parameters Values Size of sensor network fields (m m) 100 100 Moving speed of mobile sink (m/s) 500 Size of memory of cluster header (KBytes) (0, 0) Cycle of sensing data transmission 0 5 Initial energy of sensor nodes (J) 0000 Size of data-loss (Kbyte) 70000 60000 50000 40000 30000 0000 10000 0 50 100 00 400 800 Number of sensor nodes (EA) Proposed Existing Figure 10: The amount of the data-loss according to the number of sensor nodes. data is piled up than a preliminary threshold in the cluster header, our proposed scheme actively sends the stored data to the mobile sink by considering the amount of collected data in the cluster header and the moving patterns of the sink node. By doing so, it minimizes the lost data. Figure 11 shows the network lifetime according to the number of sensor nodes. The existing scheme does not cause additional communication costs since it does not consider the data loss even though the overflow occurs. In the case of the data-overflow, the proposed scheme needs additional communication costs to send the stored data. However, the proposed scheme actively sends the stored data to the mobile sink by considering the amount of the collected data in the cluster header and the moving patterns of the sink node. By doing so, it minimizes additional communication costs. The proposed scheme has similar network lifetime over the existing scheme based on a mobile sink. 5. Conclusion In this paper, we have proposed a novel data gathering method based on a mobile sink considering the data-loss in wireless sensor networks. Our proposed scheme actively sends the stored data to the mobile sink by considering the amountofcollecteddataintheclusterheaderandthemobile patterns of the sink node. By doing so, it minimizes the loss data of each sensor node. It has been shown through various experiments that the proposed scheme reduced the data loss by sending data toward the moving path of a mobile sink. In spite of the minimization of the data-loss, the network Network lifetime (rounds) 60000 50000 40000 30000 0000 10000 0 50 100 00 400 800 Number of sensor nodes (EA) Proposed Existing Figure 11: The network lifetime according to the number of sensor nodes. lifetime of the proposed scheme is similar to the existing scheme based on a mobile sink. In the future work, we plan to extend our work to apply the proposed scheme to the dynamic clustering environment that considers the changes of the cluster headers. Conflict of Interests The authors declare that there is no conflict of interests regarding the publication of this paper. Acknowledgments This research was supported by the MSIP (Ministry of Science, ICT and Future Planning), Republic of Korea, under the ITRC (Information Technology Research Center) Support Program (NIPA-014-H0301-14-10) and the C- ITRC (Convergence Information Technology Research Center) Support Program (NIPA-014-H0401-14-1007) supervised by the NIPA (National IT Industry Promotion Agency) and by the National Research Foundation of Korea (NRF) Grant funded by the Korea government (MSIP) (no. 013R1AAA01015710). References [1] J. J. Jung, Semantic preprocessing for mining sensor streams from heterogeneous environments, Expert Systems with Applications, vol. 38, no. 5, pp. 6107 6111, 011. [] M. Wu and C. Kim, A cost matrix agent for shortest path routinginadhocnetworks, JournalofNetworkandComputer Applications,vol.33,no.6,pp.646 65,010. [3] M. J. Handy, M. Haase, and D. Timmermann, Lower energy adaptive clustering hierarchy with deterministic cluster-head selection, in Proceedings of the 4th International Workshop on Mobile and Wireless Communications Network, pp.368 37, 00. [4] O. Younis and S. Fahmy, HEED: a hybrid, energy-efficient, distributed clustering approach for ad hoc sensor networks,
Distributed Sensor Networks 7 IEEE Transactions on Mobile Computing,vol.3,no.4,pp.366 379, 004. [5] S. Madden, M. J. Franklin, J. M. Hellerstein, and W. Hong, TAG: a tiny aggregation service for ad-hoc sensor networks, in Proceedings of the Symposium on Operating Systems Design and Implementation (OSDI 0),pp.131 146,00. [6] J. Gehrke and Y. Yao, The cougar approach to in-network query processing in sensor networks, in Proceedings of the ACM SIGMOD International Conference on Management of,vol. 31, no. 3, pp. 9 18, 00. [7]J.M.Kahn,R.H.Katz,andK.S.J.Pister, Nextcentury challenges: mobile networking for Smart Dust, in Proceedings of the Annual ACM/IEEE International Conference on Mobile Computing and Networking, pp. 71 78, 1999. [8] J. Luo, J. Panchard, M. Piorkowski, J. Piorkowski, and M. Grossglauser, Mobiroute routing towards a mobile sink for improving lifetime in sensor networks, in Proceedings of the International Conference on Distributed Computing in Sensor Systems,pp.480 497,006. [9] W. Zhao, M. Ammar, and E. Zegura, Controlling the mobility of multiple data transport ferries in a delay-tolerant network, in Proceedings of the 4th Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM 05), vol.,pp.1407 1418,Miami,Fla,USA,005. [10] J. Luo and J. Hubaux, Joint mobility and routing for lifetime elongation in wireless sensor networks, in Proceedings of the Annual Joint Conference of the IEEE Computer and Communications Societies,vol.3,pp.1735 1746,005. [11] D. Seong, J. Lee, M. Yeo, and J. Yoo, An energy efficient data-centric probing priority determination method for mobile sinksinwirelesssensornetworks, KIISE: Computing Practices and Letters,vol.16,no.5,pp.561 565,010. [1] F.Wu,C.Huang,andY.Tseng, gatheringbymobilemules in a spatially separated wireless sensor network, in Proceedings of the International Conference on Mobile Management: Systems, Services and Middleware, pp. 93 98, 009.
Rotating Machinery Engineering The Scientific World Journal Distributed Sensor Networks Sensors Control Science and Engineering Advances in Civil Engineering Submit your manuscripts at Electrical and Computer Engineering Robotics VLSI Design Advances in OptoElectronics Navigation and Observation Chemical Engineering Active and Passive Electronic Components Antennas and Propagation Aerospace Engineering Modelling & Simulation in Engineering Shock and Vibration Advances in Acoustics and Vibration