International Journal of Application or Innovation in Engineering & Management (IJAIEM) Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com Volume 2, Issue 8, August 2013 ISSN 2319 - 4847 An Approch For Classification and Clustering The Moving Objects V.Satish Kumar1, S.Deepthi2 1 2 Pursuing M.tech In computer science engineering at vignan’s lara institute of technology & science Asst.Professior In computer science engineering at vignan’s lara institute of technology & science ABSTRACT In this computer era there is lot of research going on the real time applications useful in our day to day life. present research topic which is having most concentration is on data mining and networks. In this paper we are going to deal with sensor networks based real time object tracking ,data compression, energy efficiency. In data mining we are dealing with clustering ,aggregation, classification. In this we apply these networks, data mining concepts in real time object tracking in wild life for bird sanctuaries, forests for tracking birds, animals etc. i.e living organisms natural habitation . In this paper we mainly concentrate on efficient usage of sensors and grouping similar objects by clustering and then classifying them into classes by svm and naive Bayesian classifiers. Key words: Wireless sensors, classifiers, clustering. 1.INTRODUCTION: In this modern age real time, day to day activities oriented research has much importance. In this living organisms birds, animals we need much attention in this because we has so many animals, birds etc. In this chain of living organisms one organism depends on other either in food or help. if the chain is broken then it would be serious issue. In order to maintain stability in food chain we track moving objects. actually tracking is done by using sensors with wireless network support. If we see in animal planet channel they keep a tracking device to the respective animal or bird for tracking it. In this way we track objects. In this loss of information ,connectivity loss ,sensor problems have to be faced and eliminate these problems in networks. In this sensor devices are connected to wireless sensor networks .In this we have mobile node, cell, base station, receiver etc equipment normally in wireless networks .In this tracking of objects we need much effort in wireless sensor networks in dealing with loss of information we must get information without loss using routing protocols like can routing algorithm by using this energy efficiency can also be resulted. and then applying clustering algorithm in similarities matching and classifying them on object behavior, activity etc by using svm, naive Bayesian classifier. 2.BACK GROUND WORK : 2.1 Movement pattern mining : Movement patterns are nothing but movement positions and actions of objects i.e animals &birds.In this pattern are action events of objects. In this patterns are like their actions, sounds, behavior etc.In this objects like animals ,birds etc we apply movement pattern mining on these objects. mining is applied on different species of birds and their classes ,sounds ,features, actions etc. we perform mining by applying clustering by clubbing similar objects in group and then apply classification done by classifiers. 2.2 Clustering : Clustering is nothing but grouping similar object based on their values by applying methods of clustering examples like binning, smoothing, bin by means etc. clusturing helps in predicting similarities in all types in finding similar sounds ,actions, features, event etc. by this using clustering it is helpful in finding similar objects and predicting to which class based on its values. and then applying classifiers for classification into classes. Fig 1: Clustering the element Volume 2, Issue 8, August 2013 Page 5 International Journal of Application or Innovation in Engineering & Management (IJAIEM) Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com Volume 2, Issue 8, August 2013 ISSN 2319 - 4847 2.3 Data Comprssion : Data compression[16] is needed in networks for efficient transmissions because if data is compressed and sent in wireless sensor networks[1] it would be useful by lessing the overhead in less error control,,bandwidth efficiency can be increased, sensor overhead be reduced etc beneficiaries can be achieved.. 2.4 Net Work And Location Models : In this sensors networks are connected to a wireless transmitter and receiver it can be mobile network or satellite network. in this network is connected with base station ,cell, service provide etc. while changing from one location to another location the network provider station changes from one base station to another base station. In this we can apply gsm or satellite etc communication models for wireless network communication. in this tracking objects birds fly from one place other places when season changes for food. in this we must use wireless sensor networks for tracking the objects so we must use a very high profile wireless network .every time location changes then based on location we must use best resources for tracking moving objects efficiently. wireless sensor networks are very helpful in providing best tracking.we must also consider on the facts like energy efficiency,bandwidth,error control,flow control,transmission properties, 3. RELATED WORK : 3.1. Mining Of Group Movement Patterns: Data mining has importance in mining huge data bases and getting efficient results by applying data mining concepts. mining is useful in getting extact results from huge data bases and data ware houses.Grouping movement patterns is nothing but based on the similarities of the object we group them on the similarity values and calculations of their patterns like sound,action ,features etc.We collect all the data bases and data ware house of all object and their images,sounds ,actions,features and then we apply data mining clustering concepts for grouping movements of the objects and we get the results of mining efficiently for our respective queries. 3.2 The Group Movement Pattern Mining Algorithm : In this algorithm we take data base from object tracking sensor tracking data by wireless sensor networks. we mine the data by applying a clustering algorithm on meta data i.e sounds, images, actions, feelings, actions etc. we group the objects First we take the data sets like parrot we take its speaking sounds, its images, feelings ,actions etc. we group the parrots on their color, sounds, actions etc.This is how grouping movements pattern mining algorithm. To provide better discrimination accuracy, we propose a new similarity measure simp to compare the similarity of two objects. For each of their significant movement patterns, the new similarity measure considers not merely two probability distributions but also two weight factors, i.e., the significance of the pattern . The similarity score simp of oi and oj based on their respective Ti and Tj, is defined as follows: The similarity score simp includes the distance associated with a pattern s, defined as 3.2 The Cluster Assemblage Algorithm : In this cluster assemblage phase grouping using normal clustering gives less results in efficiency.but using cluster assemblage [14]we can group different objects of same group in different location we can group them using this cluster assemblage phase.By this we can group similar object if slightest difference also in location or in any other similarity or patterns.The ensembling problem involves finding the partition of O that contains the most information about the local grouping results. Let C denote the ensemble of the local grouping results, represented as C={G0,G1,….Gk } where K denotes the ensemble size, Our goal is to discover the assemblage result G’ that p contains the most information about C, ie where G denotes all possible ensembling results. However, enumerating every G G in order to find the optimal ensembling result G’ is impractical, especially in the resource-constrained environments. To overcome this diffi- culty, the CE algorithm trades off the grouping quality against the computation cost by adjusting the partition parameter D, i.e., a set of thresholds with values in the range [0,1]such that a finer-grained configuration of D achieves a better grouping quality but in a higher computation cost. Therefore, for a set of thresholds D, we rewrite our objectivefunction as Volume 2, Issue 8, August 2013 Page 6 International Journal of Application or Innovation in Engineering & Management (IJAIEM) Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com Volume 2, Issue 8, August 2013 ISSN 2319 - 4847 The algorithm includes three steps. First, we utilize Jaccard Similarity Coefficient as the measure of the similarity for each pair of objects. Second, for each , we construct a graph where two objects share an edge if their Jaccard Similarity Coefficient is above . Our algo- rithm partitions the objects to generate a partitioning result G . Third, we select the ensembling result G ’ . Because of space limitations, we only demonstrate the CE mining algorithm with an example in Appendix C, which can be found on the Computer Society Digital Library at http:// doi.ieeecomputersociety.org/10.1109/TKDE.2010.30. In the next section, we propose our compression algorithm that leverages the obtained group movement patterns. 3.3 A Compression Algorithm With Group Movement Patterns: Compression is lessing the data size but not losing data information.it is necessary for every network related transmission or storage in data base or data ware house. By transmission can be easy if network speed also is less and energy efficiency can also be achieved. In our concept data transmission over wireless sensor networks is reliable but some situations we can get troubles in transmission of less bandwidth and capacity and speed .so we compress the data then we can get efficiency even in such conditions we get good results. 3.4 Trajectory Clustering : In this trajectory is path of the object or group of objects of its location tracking. trajectory clustering [7]is needed in grouping similar location paths data from wireless networks & data warehouses. Trajectory helps not only in clustering but also this data it is useful for other uses also. 3.5 Similarity Masure : Similarity measures are needed in clustering and less the over head occurred in various situations & circumstances. Similarity measures is nothing but grouping similar objects based on its color, action, features etc. by this clustering could be easy in grouping similar objects .by using this similarities we can do many operation..similarities measures are done by some distance calculations . we define the distance a pattern s associated with two objects Oi and Oj as 3.6 Svm Classifier : Support Vector Machines (SVMs) are supervised learning methods used for classification and regression phenomena’s that begin from statistical learning theory . As a classification method, SVM is a global classification model that generates non-overlapping partitions and usually employs all attributes. The entity space is partitioned in a single pass, so that flat and linear partitions are generated. SVMs are based on maximum margin linear discriminants, and are similar to probabilistic approaches, but do not consider the dependencies among attributes . Let D be a classification dataset with n points in a d-dimensional space D = {(xi , yi )}, with i = 1, 2, ..., n and let it have only two class labels such that yi is either +1 or -1. A hyperplane h(x) gives a linear discriminant function in d dimensions and splits the original space into two half-spaces: , where w is a d-dimensional weight vector and b is a scalar bias. Points on the hyperplane have h(x) = 0, i.e. the hyperplane is defined by all points for which wTx = -b. Fig 2. SVM Classifier 3.7 Navive Bayesian Classifier : Naive Bayes classifiers can p an perform arbitrary number of independent variables whether continuous or categorical. Given a set of variables, X = {x1, x2, x3……., xd}, we want to construct the posterior probability for the event Cj among a set of possible outcomes C = {C1, C2, C3…., Cn}. In a more familiar language, X is the predictors and C is the set of categorical levels present in the dependent variable. Using Bayes' rule: Volume 2, Issue 8, August 2013 Page 7 International Journal of Application or Innovation in Engineering & Management (IJAIEM) Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com Volume 2, Issue 8, August 2013 ISSN 2319 - 4847 where is the posterior probability of class membership, i.e., the probability that X belongs to Cj. 3.8 Layer By Layer Network Structure And Location Model : In this network structure & location model in wireless sensor network helps in getting efficiency in storage ,sensor capacity, scalability, network life time. In this we apply clustering on nodes of networkby similar location ids and sensor ids in that location by combining layer by layer into a network of cluster. This a network structure with location model. 3.9 Distributed Mining Algorithm: In this distributed mining algorithm we apply on location, group, time etc. In this wireless sensor network are distributed in different location we then combine this distributed wireless sensor network data And apply distributed mining algorithm for getting results .this is distributed data from distributed wireless sensor networks are collected in data bases &data warehouse repository is performed by applying data mining for getting efficient results . 4. ENERGY EFFICIENT ENTITY TRACKING SENSOR NETWORK : In this sensor networks object tracking energy efficiency must be attained because we must track objects going in different direction and sensor network packet transmission requires more bandwidth ,capacity so in order to over come this We need a energy efficient entity tracking sensor network in this we must use probabilistic prediction for detecting object tracking directions by this we get energy efficiency. 4.1 Group Data Aggregation : Aggregation is clubbing similar items based on their score of measurement . aggregation is done on data items on their values similar on their avg values or for probabilistic values. In this way first we aggregate similar items and then apply group aggregation. In this group aggregation the items are cluster and then formed into groups. In this we combine one by applying primary key like id of the sensor, group id, and time details like time stamp. by this similar items are aggregated. 4.2 Network Data Aggregation : In wireless sensor networks data aggregation is necessary for saving energy in transmission. In this every transmission from one to other stores the data of transmission from one sensor to other sensor network by this energy is saved because network location data is transmitted from one to other sensor network.By this we reduce traffic, energy efficiency. 5 WORKING OF MOVEMENT PATTERN OBJECT TRACKING IN WIRELESS SENSOR NETWORKS: Fig 3 : Working of movement pattern object tracking in wireless sensor networks. Volume 2, Issue 8, August 2013 Page 8 International Journal of Application or Innovation in Engineering & Management (IJAIEM) Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com Volume 2, Issue 8, August 2013 ISSN 2319 - 4847 5.1 Execution of movement pattern object tracking in wireless sensor networks: Step1: in this we apply sensor on every object we want track using wsn network Step2: tracking of movement patterns of object in wsn is like this Fig 4 : Object tracking in wireless sensor networks. Step3:clustering the object based on the similarity measures and grouping them accordingly Step4:aggregating network and data of objects by aggregation operation Step5:classifying the details of the object based on clustering and aggregation and grouping details into their respective classes .by performing probabilistic formulas &calculations by Storing in data ware house for mining and then we perform mining for results. Step6:compression of data in wireless sensor networks& data bases. Step7:movement pattern object tracking is performed. Example: result we take tracking data and compare the details of parrots data with missing parrot or a parrot which changed its location from bird sanctuary to other location then we compare data of group of parrot. and know the details of parrot to which kind of parrot group it is and track it to which location it resides and presently it is at which location. this s how movement pattern object tracking is done efficiently with energy efficiency. 6. CONCLUSION: In Future work we are going to determine efficiency results for various classisfication algorithm and applying better sensor for object tracking in wireless sensor networks which would be having good performance results. References: [1.] S.S. Pradhan, J. Kusuma, and K. Ramchandran, “Distributed Compression in a Dense Microsensor Network,” IEEE Signal Processing Magazine, vol. 19, no. 2, pp. 51-60, Mar. 2002. [2.] A. Scaglione and S.D. Servetto, “On the Interdependence of Routing and Data Compression in MultiHop Sensor Networks,” Proc. Eighth Ann. Int’l Conf. Mobile Computing and Networking, pp. 140-147, 2002. [3.] R. Agrawal and R. Srikant, “Mining Sequential Patterns,” Proc.11th Int’l Conf. Data Eng., pp. 3-14, 1995. [4.] J.N. Al Karaki and A.E. Kamal, “Routing Techniques in Wireless Sensor Networks: A Survey,” IEEE Wireless Comm., vol. 11, no. 6, pp. 6-28, Dec. 2004. [5.] H. Ayad, O.A. Basir, and M. Kamel, “A Probabilistic Model Using Information Theoretic Measures for Cluster Ensembles,” Proc. Fifth Int’l Workshop Multiple Classifier Systems, pp. 144-153, June 2004. [6.] D. Bollegala, Y. Matsuo, and M. Ishizuka, “Measuring Semantic Similarity between Words Using Web Search Engines,” Proc. 16th Int’l World Wide Web Conf., pp. 757-766, 2007. [7.] L. Chen, M. Tamer Ö zsu, and V. Oria, “Robust and Fast Similarity Search for Moving Object Trajectories,” Proc. ACM SIGMOD, pp. 491-502, 2005. [8.] J.-G. Lee, J. Han, and K.-Y. Whang, “Trajectory Clustering: A Partition-and-Group Framework,” Proc. ACM SIGMOD, pp. 593-604, 2007. [9.] C.-Y. Lin, W.-C. Peng, and Y.-C. Tseng, “Efficient In-Network Moving Object Tracking in Wireless Sensor Networks,” IEEE Trans. Mobile Computing, vol. 5, no. 8, pp. 1044-1056, Aug. 2006. [10.] M. Morzy, “Mining Frequent Trajectories of Moving Objects for Location Prediction,” Proc. Fifth Int’l Conf. Machine Learning and Data Mining in Pattern Recognition, pp. 667-680, July 2007. [11.] M. Nanni and D. Pedreschi, “Time-Focused Clustering of Trajectories of Moving Objects,” J. Intelligent Information Systems, vol. 27, no. 3, pp. 267-289, 2006. [12.] S. Pandey, S. Dong, P. Agrawal, and K. Sivalingam, “A Hybrid Approach to Optimize Node Placements in Hierarchical Hetero- geneous Networks,” Proc. IEEE Conf. Wireless Comm. and Network- ing Conf., pp. 3918-3923, Mar. 2007. [13.] W.-C. Peng, Y.-Z. Ko, and W.-C. Lee, “On Mining Moving Patterns for Object Tracking Sensor Networks,” Proc. Seventh Int’l Conf. Mobile Data Management, p. 41, 2006. [14.] A. Strehl and J. Ghosh, “Cluster Ensembles—A Knowledge Reuse Framework for Combining Partitionings,” Proc. Conf. Artificial Intelligence, pp. 93-98, July 2002. Volume 2, Issue 8, August 2013 Page 9 International Journal of Application or Innovation in Engineering & Management (IJAIEM) Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com Volume 2, Issue 8, August 2013 ISSN 2319 - 4847 [15.] S.S. Pradhan, J. Kusuma, and K. Ramchandran, “Distributed Compression in a Dense Microsensor Network,” IEEE Signal Processing Magazine, vol. 19, no. 2, pp. 51-60, Mar. 2002. [16.] S. Baek, G. de Veciana, and X. Su, “Minimizing Energy Consumption in Large-Scale Sensor Networks through Distrib- uted Data Compression and Hierarchical Aggregation,” IEEE J. Selected Areas in Comm., vol. 22, no. 6, pp. 1130-1140, Aug. 2004. AUTHOR PROFILE: V.Satish Kumar, pursuing M.Tech in Computer Science Engineering at Vignan's LARA Institute Of Technology and Science, Vadlamudi, Guntur Dist., A.P., India. His research interests are Data Mining and Data Warehousing and networks. S.Deepthi, Asst.Prof, Department of CSE, Vignan's LARA Institute Of Technology & Science, Vadlamudi Guntur Dist., A.P., India. Her research interests are networking and Data Mining Data Warehousing. Volume 2, Issue 8, August 2013 Page 10