A Survey of Sensor Network ApplicationsNing XuComputer Science DepartmentUniversity of Southern [email protected] the past few years, many wireless sensor networks had been deployed, these applica-tions serve to explore the requirements, constraints and guidelines for general sensor networkarchitecture design. In this paper,we present a snapshot of the recent deployed sensor networkapplications and identify the research challenges associated with such applications.1 IntroductionThe recent advance in Micro-electro-mechanical system(MEMS) and wireless communicationtechnology makes it a pragmatic vision[13, 12] to deploy a large-scale, low power ,inexpensivesensor network. Such an approach promises advantage over the traditional sensing methods inmany ways:large- scale, densely deployment not only extends the spatial coverage and achieveshigher resolution, but also increases the fault-tolerance and robustness of the system, the ad-hocnature and ”deploy’em and leave’em” vision make it even more attractive in military applicationsand other risk-associated applications, such as habitat monitoring and environmental observation.[21, 9, 30, 29, 6, 7, 25]During the past few years, lots of efforts have been directed to make this vision a reality. Re-search prototype sensor nodes(UCB motes[15, 16],uAMPS[1],PC104[4],GNOMES[11] etc.) aredesigned and manufactured, energy effecient MAC[27], topology control protocols [32, 31, 18]and routing schemes[17, 8, 19, 14, 26] are implemented and evaluated, various enabling tech-nologies such as time synchronizations[10], localization and tracking[28] are being studied andinvented. In this paper, we intend to take a snapshot of the recent deployed sensor networks, andidentify the research challenges these applications brought forward.Although sensor network research is initially driven by military applications such as battle-field surveillance and enemy tracking,we will survey only civil applications in this paper. Underthis civil catagory, the existing applications can be classfied into habitat monitoring, environmentobservation and forecast system, health and other commerical applications.The remainder of the paper is organized as follows: section 2 surveys habitat monitoringapplications, section 3 surveys EOFS applications, section 4 discusses health applications, section5 presents other commercial applications, section 6 summaries the field and identifies researchchallenges.12 Habitat monitoring applicationsCerpa et al.[9] describe habitat monitoring as a driver application for wireless sensor network:habitat sensing for biocomplexity mapping. In this first cut on habitat monitoring sensor networkapplication, they propose a tiered architecture for such applications and a frisbee model that opti-mizes energy effeciency when monitoring moving phenomenon.2.1 Great Duck Island(GDI) systemIn August 2002, researchers from UCB/Intel Research Laboratory deployed a mote-based tieredsensor network on Great Duck Island, Maine, to monitor the behavior of storm petrel[21].2.1.1 UCB Mica moteUC Berkeley Mica mote deployed in this application use an Atmel Atmega 103 microcontrollerrunning at 4MHz, 916MHz radio from RF monolithics to provide bidirectional communication at40kbps, and a pair of AA batteries to provide energy.The Mica Weather Board, stacked to the pro-cessor board via the 51 pin extension connector, includes temperature, photoresistor, barometer,humidity and thermopile sensors. Some new designs to preserve energy on this version include anADC and an I2C 8x8 power switch on the sensor board,the bypassing of the DC booster etc. Toprotect from the variable weather condition on GDI,the Mica mote is packaged in acrylic enclo-sure, which will not obstruct the sensing functionality and radio communication of the motes.2.1.2 System Architecture32 motes were placed at area of interest(e.g., inside a burrows). Those motes, grouped intosensor patches, transmit sensor reading to a gateway(CerfCube),which is responsible for forward-ing the data from the sensor patch to a remote basestation through a local transmit network. The2basestation then provides data logging and replicates the data every 15 minutes to a Postgressdatabase in Berkeley over satellite link.Users can interact with the sensor network in 2 ways. Remote users can access the replicadatabase server in Berkeley, a small PDA-size device can be used to perform local interactionssuch as adjusting the sampling rates, power management parameters etc.2.1.3 Other works on habitat monitoring applicationsWang et al.[29] discuss methods for habitat monitoring, such as target classification by maxi-mum cross-correlation between measured acoustic signal and reference signal, localization usingTDOA-based beamforming, and data reduction using zero-crossing rate technique. A prototypetestbed consisting of iPAQs is built to evaluate the performance of those target classification andlocalization methods.Energy effeciency shall be one of the design goals at every level:hardware, local process-ing(compressing, filtering etc.),MAC and topology control, data aggregation,data-centric routingand storage. Wang et al.[30] proposed preprocessing in habitat monitoring applications. Theyargue that the tiered network in GDI[21] is solely used for communication, they then present a2-tier network for the purpose of collaborative signal and information processing. The proposednetwork architecture consists of micronodes and macronodes,the micronodes perform local filter-ing and data reduction as 2 types of preprocessing that significantly reduce the amount of datatransmitted to macronodes. A preliminary experiment shows that data reduction and event filter-ing using cross-zero rate are effective, especially in the high data volume scenario such as acousticsampling.2.2 PODS-A Remote Ecological Micro-Sensor NetworkPODS[6] is a research project in University of Hawaii that built wireless network of environmentalsensor to investigate why endangered species of plants will grow in one area but not in neighboringareas. They deployed camouflaged sensors node, called Pods, in Hawaii Volcanos National Park.The Pods, consist of a computer, radio transceiver and environmental sensors sometimes includinga high resolution digital camera, relay sensor data via wireless link back to the Internet. Bluetoothand 802.11b are chosen as MAC, data are deliveried in IP packets. Energy efficiency is identified asone of the design