TheImplementation ofanAdaptive DataReduction Technique forWireless Sensor Networks

andthebasestation. Thissolution requires knowledge ofthe Abstract Wireless sensor networks (WSN)havegainedlocation ofthenodestodetermine theshortest distance significant attention duetotheir ability tomonitor physical between thesource andthesink. Analternative methodto phenomena within awiderange ofapplications. These networks are reduce powerconsumption istoswitch offthetransceiver generally deployed inremote areas andarebattery powered This during theperiods ofinactivity (1). Thisrequires that the meansthatthelifetime ofthenetwork depends ontheenergytransmissions arescheduled within timeslots andthuseach consumption, thusnecessitating careful hardware andsoftware design tosustain thelongperiod ofoperation without human nodemusthavethenecessary knowledge about itstimeslot, intervention. Onewayofreducing theenergy required isto theexpected amount ofdata toberelayed from its neighbors, minimize thenumber ofdatatransmissions. Aprediction-based data andtheaddress ofthenodetowhichitwillforward the reduction algorithm basedon theleast-mean-square (LMS)aggregated data. Thisnecessitates accurate synchronization algorithm wasimplemented onafield programmable gatearraybetween thesensor nodesandthebasestation, whichis (FPGA) toreduce thecommunication between thesensor nodes and controlled by thebasestation through synchronization thebase station. Measurement results showthat communication can packets. Otherthansaving nodepower, this protocol also bereduced byasmuchas90%ina temperature monitoring avoids collisions andthuseliminates retransmissions, which application ifanerror of0.5degree isacceptable. Thishasalargewouldotherwise drain unnecessary power. impact onthelifetime ofthewireless sensornetwork sincethe Monitoring aslowvarying parameter, suchastemperature, transceiver canbeswitched off during non-communication periods saving precious energy. meansthatsuccessive readings fromasensor nodewill be similar. Therefore, thecommunications overhead canbe Keywords-field programmable gate arrays, least-mean-square, reduced bytransmitting onlysparse datavalues fromeach prediction-based algorithm, wireless sensor networks. sensornodesincemostofthetimetheinformation is redundant. Theobservation dataisthenobtained either from