IEEE Communications Magazine • August 2002102A Survey on Sensor Networks0163-6804/02/$17.00 © 2002 IEEEABSTRACTRecent advancement in wireless communica-tions and electronics has enabled the develop-ment of low-cost sensor networks. The sensornetworks can be used for various applicationareas (e.g., health, military, home). For differentapplication areas, there are different technicalissues that researchers are currently resolving.The current state of the art of sensor networks iscaptured in this article, where solutions are dis-cussed under their related protocol stack layersections. This article also points out the openresearch issues and intends to spark new inter-ests and developments in this field.INTRODUCTIONRecent advances in wireless communications andelectronics have enabled the development of low-cost, low-power, multifunctional sensor nodes thatare small in size and communicate untethered inshort distances. These tiny sensor nodes, whichconsist of sensing, data processing, and communi-cating components, leverage the idea of sensornetworks. Sensor networks represent a significantimprovement over traditional sensors.A sensor network is composed of a largenumber of sensor nodes that are denselydeployed either inside the phenomenon or veryclose to it. The position of sensor nodes neednot be engineered or predetermined. This allowsrandom deployment in inaccessible terrains ordisaster relief operations. On the other hand,this also means that sensor network protocolsand algorithms must possess self-organizingcapabilities. Another unique feature of sensornetworks is the cooperative effort of sensornodes. Sensor nodes are fitted with an onboardprocessor. Instead of sending the raw data to thenodes responsible for the fusion, they use theirprocessing abilities to locally carry out simplecomputations and transmit only the required andpartially processed data.The above described features ensure a widerange of applications for sensor networks. Someof the application areas are health, military, andhome. In military, for example, the rapid deploy-ment, self-organization, and fault tolerance char-acteristics of sensor networks make them a verypromising sensing technique for military com-mand, control, communications, computing, intel-ligence, surveillance, reconnaissance, and targetingsystems. In health, sensor nodes can also bedeployed to monitor patients and assist disabledpatients. Some other commercial applicationsinclude managing inventory, monitoring productquality, and monitoring disaster areas.Realization of these and other sensor net-work applications require wireless ad hoc net-working techniques. Although many protocolsand algorithms have been proposed for tradi-tional wireless ad hoc networks, they are notwell suited to the unique features and applica-tion requirements of sensor networks. To illus-trate this point, the differences between sensornetworks and ad hoc networks are:• The number of sensor nodes in a sensor net-work can be several orders of magnitudehigher than the nodes in an ad hoc network.• Sensor nodes are densely deployed.• Sensor nodes are prone to failures.• The topology of a sensor network changesvery frequently.• Sensor nodes mainly use a broadcast com-munication paradigm, whereas most ad hocnetworks are based on point-to-point com-munications.• Sensor nodes are limited in power, compu-tational capacities, and memory.• Sensor nodes may not have global identifica-tion(ID) because of the large amount ofoverhead and large number of sensors.Many researchers are currently engaged in devel-oping schemes that fulfill these requirements.In this article we present a survey of protocolsand algorithms proposed thus far for sensor net-works. Our aim is to provide a better understand-ing of the current research issues in this emergingfield. We also attempt an investigation into per-taining design constraints and outline the use ofcertain tools to meet the design objectives.The remainder of the article is organized asfollows. We discuss the communication architec-ture of the sensor networks as well as the factorsthat influence sensor network design. We pro-vide a detailed investigation of current proposalsin the physical, data link, network, transport, andapplication layers, respectively. We then con-clude our article.Ian F. Akyildiz, Weilian Su, Yogesh Sankarasubramaniam, and Erdal CayirciGeorgia Institute of TechnologyACCEPTED FROM OPEN CALLIEEE Communications Magazine • August 2002103SENSOR NETWORKSCOMMUNICATION ARCHITECTUREThe sensor nodes are usually scattered in a sensorfieldas shown in Fig. 1. Each of these scatteredsensor nodes has the capabilities to collect dataand route data back to the sink. Data are routedback to the sink by a multihop infrastructurelessarchitecture through the sink as shown in Fig. 1.The sink may communicate with the task managernodevia Internet or satellite. The design of thesensor network as described by Fig. 1 is influ-enced by many factors, including fault tolerance,scalability, production costs, operating environment,sensor network topology, hardware constraints,transmission media, and power consumption.DESIGN FACTORSThe design factors are addressed by manyresearchers as surveyed in this article. However,none of these studies has a fully integrated viewof all the factors driving the design of sensornetworks and sensor nodes. These factors areimportant because they serve as a guideline todesign a protocol or an algorithm for sensor net-works. In addition, these influencing factors canbe used to compare different schemes.Fault Tolerance — Some sensor nodes may failor be blocked due to lack of power, or havephysical damage or environmental interference.The failure of sensor nodes should not affect theoverall task of the sensor network. This is thereliability or fault tolerance issue. Fault toler-ance is the ability to sustain sensor networkfunctionalities without any interruption due tosensor node failures [1, 2]. The reliability Rk(t)or fault tolerance of a sensor node is modeled in[2] using the Poisson distribution to capture theprobability of not having a failure within thetime interval (0,t):Rk(t) = e–λkt, (1)where λkis the failure rate of sensor node k andt is the time period.Scalability — The number of sensor nodesdeployed in studying a phenomenon may be onthe order of hundreds or thousands. Dependingon the application, the number may reach anextreme value of millions. New schemes must beable to work
View Full Document
Unlocking...