Dhananjay M. Sable , Latesh G. Malik / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.022-02822 | P a g eDesign & Implementation Of An Algorithm For ProximityEstimation In Sensor NetworkDhananjay M. Sable , Latesh G. Malik1IV Sem M.E.( Embedded System and Computing) Deptt. Of Computer Science and Engg G.H.Raisoni collegeof Engg. Nagpur16(India)2H.O.D Deptt. Of Computer Science and Engg G.H.Raisoni college of Engg. Nagpur16(India)AbstractSensor networks are often used toperform monitoring tasks, such as in animal orvehicle tracking and in surveillance of enemyforces in military applications. In this project wewill be introduce the concept of proximity queriesthat allow us to report interesting events that areobserved by nodes in the network that are withincertain distance of each other. An event istriggered when a user programmable predicate issatisﬁed on a sensor node. To study the problemof computing proximity queries in sensornetworks using existing communication protocolsand then propose an efficient Algorithm that canprocess multiple proximity queries, involvingseveral different event types. This project givessolution utilizes a distributed routing index,maintained by the nodes in the network that isdynamically updated as new observations areobtained by the nodes. This project presents anextensive experimental study to show the beneﬁtsof this techniques under different scenarios.Keywords:—position estimation, wireless sensorsnetwork, decentralized algorithm, position-basedrouting, location identiﬁcationI. INTRODUCTIONSensor networks are used in a variety ofmonitoring tasks. This survey will be study of newdecentralized algorithms for the detection of eventsthat are observed by nodes within certain distance ofeach other, i.e. events that are reported by sensornodes in spatial proximity. An event is triggeredwhen a user-programmable predicate is satisﬁed on asensor node. The deﬁnition allows different types ofevents, depending on the sensing capabilities of thenodes in the network and on the application. Forinstance, in an application where nodes are used forcollecting meteorological data, an event may bedeﬁned for when the temperature readings on asensor exceed a certain threshold. Another type ofevent, in the same application may be deﬁned whentemperature readings fall below another, lower value.When both events are detected, each by a differentnode and these two nodes are in proximity, this mayindicate an abrupt, abnormal shift in temperature inthe terrain. In a military Surveillance application,events may be used to detect the movement offriendly and enemy forces. Proximity alerts thenmay be used for the early warning of approachingenemy forces. In another application of wild animaltracking, we may want to raise an alert when apredator in spotted in an area occupied by a ﬂockthat observe, assuming that the presence of eachanimal can be detected by the use of RFIDtechnology. Sensor networks consist of wireless,battery-powered sensing devices, have introducenew challenges in data management and have spawnseveral recent proposals for embedded databasesystems, such as COUGAR and TinyDB. Most ofthe proposed techniques explore in networkprocessing to carefully synchronize the operation ofthe nodes and utilize the, Multi-hop communicationlinks to leverage the computation of expensivequeries, such as those involving aggregation.Continuous monitoring queries and distributed joinalgorithms have also been considered. Alternativemethods try to reduce the cost of data processing insensor networks through probabilistic techniques,data modelling or through the use of decentralizedalgorithm. In this project algorithms will fall in thelatter category. Application of existing methods forcomputing set-expressions in data streams in theevaluation of proximity queries is an open researchquestion due to the different settings and costconsiderations. Most of these fundamentaltechniques have been devised to support event-basedmonitoring applications. For example, in animaltracking, an event such as the presence of an animalcan be determined by matching the sensor readingsto stored patterns. This project proposes an eventdetection mechanism based on matching the contourmaps of in-network sensory data distributions. Inkernel-based techniques are used to detect abnormalbehaviour in sensor readings. Since mostapplications depend on a successful localization, i.e.to compute their positions in some fixed coordinatesystem, it is of great importance to design efficientlocalization algorithms. It can also be used inhospital environments to keep track of equipments,patients, doctors and nurses. For these advantagesprecise knowledge of node localization in ad hocsensor networks is an active field of research in
Dhananjay M. Sable , Latesh G. Malik / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.022-02823 | P a g ewireless networking. Recent advances in wirelesssensor network (WSN) technology have enableddeployment of sensor networks in numerousapplications, ranging from medical andenvironmental monitoring to detection and trackingof intruders. Localisation of sensor nodes of awireless sensor network is an essential prerequisitefor its successful operation. However, localisation inWSNs is often constrained by the low cost, lowenergy and conjuration requirements. Thus, it isunreasonable to expect that every sensor node can beequipped with a GPS receiver (a GPS receiver islarge, heavy and expensive, as opposed to a sensornode which is supposed to be small, light andcheap). While there are many algorithms for nodelocalisation reported in the literature this projectfocus on range-free, a cost-effective alternative tothe more expensive range based methods. In range-based systems, every sensor node requires hardwarethat can deduce the distance between the sensors.While this information improves localisationaccuracy. Range-free systems, on the other hand, useonly the connectivity or proximity information tolocalise themselves. The assumption is that (small)subsets of sensor nodes, called anchors, know theirexact position within some global coordinate system.The anchors broadcast their position and thisinformation is received by sensor nodes that arewithin their respective communication range.II. LITERATURE REVIEWMiloš Jevtic, Nikola Zogovic, (2009) donewireless sensor network (WSN) simulator surveys.They examine four WSN simulators: ns-2, Castalia(OMNeT++ based), TOSSIM, andCOOJA/MSPSim, and define a set of criteria toevaluate and compare the simulators. They provideshort descriptions of simulators and tabularcomparison based on the criteria. Since none of thesimulators under survey is a universal solution,rough guidelines on which simulator to use inparticular situation’s Sanjiv Rao and V Vallikumari,(2012) published, the recent advances in electronicsand wireless communication technologies haveenabled the development of large-scale wirelesssensor networks that consist of many low-powers,low-cost and small size sensor nodes. Sensornetworks hold the promise of facilitating large-scaleand real time data processing in complexenvironments. Some of the application areas arehealth, military, and home. In military, for example,the rapid deployment, self-organization, and faulttolerance characteristics of sensor networks makethem a very promising technique, for militarycommand, control, communications, computing, andthe targeting systems. Deployment of nodes inWireless Sensor Networks (WSNs) is a basic issueto be addressed as it can influence the performancemetrics of WSNs connectivity, resilience and storagerequirements. Many deployment schemes have beenproposed for wireless sensor networks. In this paperwe consider the implications of various deploymentschemes for the connectivity and resilience of theWSNs. Our approach to deal with the affectivetrade-offs between resilience, connectivity andstorage requirements for each deployment schemesin the literature, we survey four deployment modelsrandom, grid, group and grid-group to show whichdeployment scheme can be used to increase networkconnectivity, without increasing storagerequirements or sacrificing resilience with respect tosome factor. Their contribution is tobe implementedof WSNs using random, grid and grid-groupdeployment Knowledge. WSNs have been simulatedwith Network Simulator 2.34 for nodeconfiguration, sink node configuration, topologycreation, with sensing capabilities, temperature andenergy by using Mannasim.Khalid K. Almuzaini(2010) presented Node localization is an essentialcomponent of many wireless networks. It can beused to improve routing and enhance the security ofa network. Localization can be divided into rangefree and range-based algorithms. Range-basedalgorithms use measurements such as ToA, TDoA,RSS, and AoA to estimate the distance between twonodes. Range-free algorithms are based on proximitysensing between nodes. Although this is typicallyless accurate than range-based, it is cheaper andeasier to implement. A new range-free scheme isproposed. Every target forms two sets of anchors.The ﬁrst set contains one-hop anchors from thetarget. The second set contains two-hop and three-hop anchors away from the target. Each target usesthe intersections between these anchors to estimateits position. This new approach is simple, but ismore accurate than DRLS when the anchor ratio islow.Cuong Pham Duc A. Tran (2011) Designeddata-centric sensor networks, sensor data is notnecessarily forwarded to a central sink for storage;instead, the nodes themselves serve as distributed in-network storage, collectively storing sensor data andwaiting to answer user queries. A key problem indesigning such a network is how to map data andqueries to their corresponding rendezvous nodes sothat a query can ﬁnd its matching data quickly andefﬁciently. Existing techniques are mostly aimed toaddress a certain type of queries. The capability tosupport queries of any type is desirable, yet remainsa challenge. We propose search efﬁcient architecturefor data-centric sensor networks that cansimultaneously address range queries and top-k, thetwo most popular query types. Importantly, theproposed technique does not require locationinformation, hence applicable to a wide range ofsensor networks. Our theoretical ﬁndings arecomplemented by a simulation-basedevaluation.Razia Haider, Dr. Muhammad YounusJaved, Naveed S. Khattak (2008) they had invented
Dhananjay M. Sable , Latesh G. Malik / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.022-02824 | P a g erouting in sensor networks is a challenging issue dueto inherent constraints such as energy, memory andCPU processing capabilities. The energy efficiencyis one of the key concerns in sensor networks fortheir better performance, as sensor nodes are limitedin their battery power. In this research work,different routing algorithms have been studied andthe main focus was on geographic routing in sensornetworks. A location based protocol EAGR (EnergyAware Greedy Routing) has been presented forsensor networks to extend the lifetime of thenetwork and to get higher data delivery rate and tobalance the energy consumption of nodes. In EAGR,each node makes the local decision to choose itsnext hop. This algorithm works on forwarding rulebased on location and energy levels of nodes. Eachnode knows its own geographic location and itsenergy levels and the location and energy level of itsNeighbours. The transmitting node writes thegeographic position of destination into the packetheader and forwards it to the destination byestablishing the sub-destinations. The sub-destination nodes must be alive and geographicallynear to the destination node to route the packet bychoosing the shortest and reliable path. Simulationhas been made by using OMNET++. Simulationresults show that the proposed algorithm gives betterperformance in terms of higher data delivery rateand less number of dead nodes. It has been notedthat the ratio of successful packet delivery willincrease in EAGR as number of nodes increased inthe network. Consequently, the proposed algorithmcan scale to thousands of nodes in future sensornetworks and can effectively increase the lifetime ofthe sensor networks. Jaehun Lee, Wooing Chung,and Euntai Kim (2009) proposed an efficient range-free localization algorithm based on shortest pathinformation. The proposed method is a kind ofrange-free localization algorithm which only usesthe proximity information between sensor nodes. Inthe proposed method, an efficient distance vectorbased on shortest path information is proposed. Theproposed method is applied to various kinds ofnetwork topologies. The simulation resultsdemonstrate that the proposed method showsexcellent and robust location estimationresults.Miloš Jevtic, Nikola Zogovic, (2009) theyexamined four WSN simulators: ns-2, Castalia(OMNeT++ based), TOSSIM, andCOOJA/MSPSim, and define a set of criteria toevaluate and compare the simulators. They providedshort descriptions of simulators and tabularcomparison based on the criteria. Since none of thesimulators under survey is a universal solution,rough guidelines on which simulator to use inparticular situation .Amitangshu Pal (2010)Publishesh that recent advances in radio andembedded systems have enabled the proliferation ofwireless sensor networks. Wireless sensor networksare tremendously being used in differentenvironments to perform various monitoring taskssuch as search, rescue, disaster relief, target trackingand a number of tasks in smart environments. Inmany such tasks, node localization is inherently oneof the system parameters. Node localization isrequired to report the origin of events, assist groupquerying of sensors, routing and to answer questionson the network coverage. So, one of thefundamental challenges in wireless sensor network isnode localization. This paper reviews differentapproaches of node localization discovery inwireless sensor networks. The overview of theschemes proposed by different scholars for theimprovement of localization in wireless sensornetworks is also presented. Future researchdirections and challenges for improving nodelocalization in wireless sensornetworks.Vaidyanathan Ramadurai, Mihail L.Sichitiu (2009) considered a probabilistic approachto the problem of localization in wireless sensornetworks and propose a distributed algorithm thathelps unknown nodes to determine conﬁdentposition estimates. The proposed algorithm is RFbased, robust to range measurement inaccuracies andcan be tailored to varying environmental conditions.The proposed position estimation algorithmconsiders the errors and inaccuracies usually foundin RF signal strength measurements. They alsoevaluate and validate the algorithm with anexperimental testbed. The test bed results indicatethat the actual position of nodes are well bounded bythe position estimates obtained despite ranginginaccuracies.Jin Zhu and Symeon Papavassiliou, (2006)published that due to hardware, energy, cost andother physical constraints, sensor-based networkspresent various design, implementation anddeployment challenges. In this letter an analyticalmodel to estimate and evaluate the node and networklifetime in a randomly deployed multi-hop sensornetwork is presented. Based on this, we provide aprocedure for the creation of an energy efficientsensor network organization, that attempts to extendthe lifetime of the communication critical nodes, andas a result the overall network’s operationlifetime.Umair Sadiq, Mohan Kumar (2011)published a series of opportunistic contacts in spaceand time among devices carried by mobile users, canbe utilized to forward messages from one user toanother in the absence of an end-to end connectedpath. Existing routing metrics exhibit efﬁcientperformance in either homogeneous (users havesimilar mobility characteristics) or speciﬁcheterogeneous (users exhibit varied mobilitycharacteristics) scenarios. However, in practice,behaviour of users changes at different locations andtimes, making it hard to generalize any one routingalgorithm. The adaptive forwarding scheme,
Dhananjay M. Sable , Latesh G. Malik / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.022-02825 | P a g eProxiMol, proposed leverages two simple facts -some users have better likelihood of messagedelivery due to higher mobility, while others do dueto their location’s proximity to destination. Keycontributions of ProxiMol include: i) a model toinfer user’s location over time from its diffusion (ameasure of mobility); ii) an analytical result toestimate distance between users; and iii) anempirical method to estimate diffusion of a user.These are used to compute the likelihood of deliverytaking into account both the mobility of a user andher proximity to destination. In addition to thisrobust forwarding scheme, a novel concept ofdisconnected distance that captures partial paths innetworks with moderate levels of connectivity isintroduced. ProxiMol improves delivery ratios (10-20%) and reduces delays by up to 50%, whencompared against previously proposed algorithms, inuser environments that range from relativelyhomogeneous to highly heterogeneoussettings.Jaehun Lee, Wooyong Chung, Euntai Kimand In Wha Hong , (2010) proposed a novellocalization algorithm in Wireless Sensor Network.The proposed method is a kind of range-freelocalization algorithm which only uses the proximityinformation between sensor nodes. In the proposedmethod, a robust weighted algorithm is presented tocalculate the average hop distances between sensornodes and anchor nodes. The proposed method isapplied to an isotropic network and three anisotropicetworks. The simulation results demonstrate that theproposed method shows excellent and robustlocation estimation results Satoshi Kurihara , (2010)described a method for accurately estimating thetopology of sensor networks from time-series datacollected from infrared proximity sensors. Ourmethod is a hybrid combining two differentmethodologies: ant colony optimization (ACO),which is an evolutionary computation algorithm; andan adjacency score, which is a novel statisticalmeasure based on heuristic knowledge. We showthat, using actual data gathered from a real-worldenvironment, our method can estimate a sensornetwork topology whose accuracy is approximately95% in our environment. This is an acceptable resultfor real-world sensor network applications.Qiang Le(2009) presented the maximum likelihoodlocalization (ML) algorithm for multi-targetlocalization using proximity-based sensor networks.Proximity sensors simply report a single binaryvalue indicating whether or not a target is near. TheML approach requires a hill climbing algorithm toﬁnd the peak, and its ability to ﬁnd the global peak isdetermined by the initial estimates for the targetlocations. This paper investigates three methods toinitialize the ML algorithm:1) centroid of k-means clustering, 2) centroid ofclique clustering, and 3) peak in the 1-targetlikelihood surface.To provide a performance bound for theinitialization methods, the paper also considers theground truth target positions as initial estimates.Simulations compare the ability of these methods toresolve and localize two targets. The simulationsdemonstrate that the clique clustering techniqueoutperforms k-means clustering and is nearly aseffective as the 1-target likelihood peak methods at afraction of the computational cost.III. CURRENT STUDY This project will be introducing the conceptof proximity queries in sensor networks. Itcaptures a large number of interestingqueries that may be used in a variety ofmonitoring applications in sensor networks. This project will propose the use of adistributed routing index for capturing thespatial distribution of interesting types ofevents in the sensor network. This routingindex requires minimal resources at eachnode and is being updated dynamically,when the nodes collaborate to provideanswers to proximity queries. This project will provide a detailedexperimental evaluation where study theeffect of various parameters in the accuracyof the algorithms. This will resultsdemonstrate that techniques are very robustand can accurately process a variety ofproximity queries, while substantiallyreducing the number of messagesexchanged in the network.IV. PLATFORM USEMannasim is a Wireless Sensor Networkssimulation environment comprised of two solutions: the Mannasim Framework; the Script Generator Tool.The Mannasim Framework is a module for WSNsimulation based on the Network Simulator (NS-2).Mannasim extends NS-2 introducing new modulesfor design, development and analysis of differentWSN applications. The Script Generator Tool (SGT)is a front-end for TCL simulation scripts easycreation. SGT comes blunded with MannasimFramework and its written in pure Java making itplataform independent.V. MANNASIM OBJECTIVESMannasim goal is to develop a detailedsimulation framework, which can accurately modeldifferent sensor nodes and applications whileproviding a versatile testbed for algorithms andprotocols.Numerous challenges make the study of realdeployed sensor networks very difficult andfinancially infeasible. At the current stage of thetechnology, a practical way to study WSNs is
Dhananjay M. Sable , Latesh G. Malik / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.022-02826 | P a g ethrough simulations that can provide a meaningfulperspective of the behavior and performance ofvarious algorithmsConfiguring Access PointsTo create an access point (AP) the first thing it to setnode-config structure. This is a ns-2 structure usedto define a bunch of mobile node parameters. InMannasim context the only difference from thetradicional settings is the inclusion of -sensorNodeON command. With this command whenever aninstance of a mobile node is instantiated a wirelesssensor node is automatically created. The box bellowshows a sample configuration (note the use of val()vector).$ns_ node-config -sensorNode ON-adhocRouting $val(rp)-adhocRouting $val(rp)-llType $val(ll)-macType $val(mac)-ifqType $val(ifq)-ifqLen $val(ifqlen)-antType $val(ant)-propType $val(prop)-energyModel $val(en)On demand networks require that the AP orother outsider send requests messages to the WSN.These messages specify the situation when thenetwork should provide data to the requester.To create a request message to be send by the AP thefollowing parameters should be set: When requested data should be sent back toAP (for example, send data whentemperature exceeds 25 celcius degrees). request type - two values are possible:REAL (0) when sensor node drops all data all datafrom its buffer, gather new one, process and deliverit to the disseminating module. BUFFER (1) whensensor node process data from its buffer and give theresults to the disseminating module.The sample codeprovided bellow defines the request time, scheduleswhen data structure that encapulates request data iscreated and when request message will be sent.VI. CONFIGURING CLUSTER HEADSThe Cluster Head (CH) configuration issimilar to what was made for the Access Point. TheCH however, can execute data sensing andprocessing so its configuration is a bit more complex(as the sensor nodes executes more different tasksmore configuration adjustments should be made).Tocreate a sensor node just put in the tradicional node-config structure the -sensorNode ON command. Justlike it was made in the AP configuration.ContinuingCH configuration, the user should set a node field torepresent the sensor node just defined, create atansport protocol (again an UDP transport protocolis used in the example presented in the box bellow),a sensor node application (as expected aClusterHeadApp) and an instance of a processingmodule.VII. CONFIGURING COMMON NODEThe Common Node (CN) configurationpresents the greater number of parameter to be set.Since a CN has an application, a processing moduleand a data generation module used to simulatesensing tasks.Sensor node creation is equal to AccessPoint node creation. Just put in the tradicional node-config structure the -sensorNode ON command.Next step is to create common nodefunctionality modules including a transport protocol,an sensor application, a processing and a datagenerator. The data generator module is thedifference between a Cluster Head and a CommonNode configuration. In the example box presentedbellow an UDP transport protocol, an AggregateProcessing processing module, aTemperatureDataGenerator data generator moduleand a CommonNodeApplication application modulewere used.
Dhananjay M. Sable , Latesh G. Malik / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.022-02827 | P a g eVIII. RESULTSNodes in the proximity of 010Nodes in the proximity of 1928293248Nodes in the proximity of 27183849Nodes in the proximity of 32942Nodes in the proximity of 4Nodes in the proximity of 5243840Nodes in the proximity of 61723Nodes in the proximity of 721446Nodes in the proximity of 82326Nodes in the proximity of 9122Nodes in the proximity of 1003642Nodes in the proximity of 1118Nodes in the proximity of 121523Nodes in the proximity of 13182943Nodes in the proximity of 14736Nodes in the proximity of 15122145Nodes in the proximity of 1649Nodes in the proximity of 17640Nodes in the proximity of 182111343Nodes in the proximity of 193446Nodes in the proximity of 2040Nodes in the proximity of 2115444648Nodes in the proximity of 22927Nodes in the proximity of 236812Nodes in the proximity of 24533Nodes in the proximity of 2534Nodes in the proximity of 26833Nodes in the proximity of 2722Nodes in the proximity of 281Nodes in the proximity of 29131334
Dhananjay M. Sable , Latesh G. Malik / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.022-02828 | P a g eNodes in the proximity of 3045Nodes in the proximity of 31Nodes in the proximity of 321Nodes in the proximity of 332426Nodes in the proximity of 34192529Nodes in the proximity of 35Nodes in the proximity of 361014Nodes in the proximity of 37Nodes in the proximity of 3825Nodes in the proximity of 39Nodes in the proximity of 4051720Nodes in the proximity of 41Nodes in the proximity of 42310Nodes in the proximity of 431318Nodes in the proximity of 4421Nodes in the proximity of 451530Nodes in the proximity of 4671921Nodes in the proximity of 47Nodes in the proximity of 48121Nodes in the proximity of 49216Proximity TCP ACK Event estimatedIX. CONCLUSION & FUTURE WORKThis survey will introduce the concept ofproximity queries in snsor networks. The definitioncaptures a large number of interesting queries thatmay be used in a variety of monitoring applicationsin sensor network.This survey will be propose theuse of a distributed routing index for capturing thespatial distribution of interesting types of events iathe sensor network. This routing index requiresminimal resources at each node and is being updateddynamically,REFERENCES Sanjiv Rao and V Vallikumari, AssociateProfessor, Dept of IT, Sri Sai AdityaInstitute of Science And Technology,Surampalem, AP India. Professor, Dept ofCS&SE, College of Engineering, AndhraUniversity. International Journal OfAdvanced Smart Sensor Network Systems (IJASSN ), Vol 2, No.2, April 2012 Khalid K. Almuzaini Student Member andT.AaronGulliverSenior MemberDepartment of Electrical and ComputerEngineering University of VictoriaVictoria, BC, Canada, Cuong Pham Duc A. Tran Department ofComputer Science University ofMassachusetts, Boston Razia Haider, Dr.Muhammad Younus Javed, Naveed S.Khattak Department of Computer Science,Military College of Signal, NationalUniversity of Science & Technology,Rawalpinid, Pakistan. Jaehun Lee, Wooyong Chung, and EuntaiKim School of Electrical and ElectronicEngineering, Yonsei University, Seoul,Korea. Vaidyanathan Ramadurai, Mihail L.Sichitiu Department of Electrical andComputer Engineering North Carolina StateUniversity Umair Sadiq, Mohan Kumar Department ofComputer Science and EngineeringUniversity of Texas Satoshi Kurihara , Toshiharu Sugawara,Department of Computer Science andEngineering, Waseda University, Japan