Modeling In-Network Processing and Aggregation in Sensor NetworksSensor NetworksIn-Network ProcessingExisting ApproachesExisting Approaches … cont.DEEPADS – A Novel ApproachC-DEEPADSSimulationSensor Node ModelPerformance EvaluationPerformance Evaluation … cont.DiscussionQuestions ??? Comments !!!Modeling In-Network Processing and Modeling In-Network Processing and Aggregation in Sensor NetworksAggregation in Sensor NetworksAjay MahimkarEE 382C Embedded Software SystemsProf. B. L. EvansMay 5, 2004Sensor NetworksSensor NetworksMonitor physical environment from remote locationsChallenges–Battery is the most pressing–Deployment of sensors in thousandsNo manual intervention–Design protocols that extend network lifetimeNetwork lifetime is the time at which first node diesIn-Network ProcessingIn-Network ProcessingWhy data aggregation???–Individual sensor readings are of limited use–Delivering large amount of data from all nodes to a central point consumes lot of energyConserves limited energy and bandwidthIncreases system lifetimeExisting ApproachesExisting ApproachesDirected Diffusion [Intanagonwiwat, 2003]LEACH (Low Energy Adaptive Clustering Hierarchy) [Heinzelman, 2000]–Cluster-Head responsible for data aggregationExisting Approaches … cont.Existing Approaches … cont.PEDAP (Power Efficient Data gathering and Aggregation Protocol) [Tan, 2003]–MST (Minimum Spanning Tree) based routing using energy as the metric–DisadvantagesLocally optimizes energyIncreases end-to-endlatencyDEEPADS – A Novel ApproachDEEPADS – A Novel ApproachDistributed Energy-Efficient Protocol for Aggregation of Data in Sensor Networks (DEEPADS)–Novel approach that globally maximizes the energy and increases system lifetimeSA BGECFDHPEDAPDEEPADS34752632 1C-DEEPADSC-DEEPADSUses Clustering Approach–Two Tier MethodologySensors organize themselves into clusters, each cluster represented by a cluster-headGlobal energy metric similar to DEEDAPCluster-head aggregates data and transmits to the base stationReduces end-to-end latencySimulationSimulationUsing Ptolemy-II, VisualSense and Java–Discrete Event Model–Network Simulation Setup–Environment100 m x 100 m area–Sensors location Uniformly distributed x and y random variablesBattery Energy at Bootstrap 2.0 JEnergy Consumed during TX or RX 50 nJ/bitThreshold Power 6.3 nWTransceiver Maximum Range 50 mMessage Length 500 BytesWavelength 0.325 mHeight of TX & RX antenna 1.5 mGain of TX & RX antenna 0 dBSimulation ParametersSensor Node ModelSensor Node ModelPerformance EvaluationPerformance EvaluationPerformance Evaluation … cont.Performance Evaluation … cont.DiscussionDiscussionResults–DEEPADS & C-DEEPADS perform much better than existing approachesIncrease in the system lifetimeReduction in the total energy consumptionFuture Work–Repeat experiments taking into consideration the sleep mode in sensorsQuestions ???Questions ???Comments !!!Comments
View Full Document