A survey on routing protocols for wireless sensor networksIntroductionSystem architecture and design issuesNetwork dynamicsNode deploymentEnergy considerationsData delivery modelsNode capabilitiesData aggregation/fusionRelated workData-centric protocolsFlooding and gossipingSensor protocols for information via negotiationDirected DiffusionEnergy-aware routingRumor routingGradient-based routingCADRCOUGARACQUIREHierarchical protocolsLEACHPEGASIS and Hierarchical-PEGASISTEEN and APTEENEnergy-aware routing for cluster-based sensor networksSelf-organizing protocolLocation-based protocolsMECN and SMECNGAFGEARNetwork flow and QoS-aware protocolsMaximum lifetime energy routingMaximum lifetime data gatheringMinimum cost forwardingSAREnergy-aware QoS routing protocolSPEEDConclusion and open issuesReferencesA survey on routing protocols for wireless sensor networksKemal Akkaya*, Mohamed YounisDepartment of Computer Science and Electrical Engineering, University of Maryland, Baltimore County, Baltimore, MD 21250, USAReceived 4 February 2003; received in revised form 20 July 2003; accepted 1 September 2003Available online 26 November 2003AbstractRecent advances in wireless sensor networks have led to many new protocols specifically designed for sensor net-works where energy awareness is an essential consideration. Most of the attention, however, has been given to therouting protocols since they might differ depending on the application and network architecture. This paper surveysrecent routing protocols for sensor networks and presents a classification for the various approaches pursued. The threemain categories explored in this paper are data-centric, hierarchical and location-based. Each routing protocol is de-scribed and discussed under the appropriate category. Moreover, protocols using contemporary methodologies such asnetwork flow and quality of service modeling are also discussed. The paper concludes with open research issues. 2003 Elsevier B.V. All rights reserved.Keywords: Sensor networks; Energy-aware routing; Routing protocols; Classification of protocols1. IntroductionRecent advances in micro-electro-mechanicalsystems and low power and highly integrated dig-ital electronics have led to the development ofmicro-sensors [1–5]. Such sensors are generallyequipped with data processing and communicationcapabilities. The sensing circuitry measures ambi-ent conditions related to the environment sur-rounding the sensor and transforms them into anelectric signal. Processing such a signal revealssome properties about objects located and/orevents happening in the vicinity of the sensor. Thesensor sends such collected data, usually via radiotransmitter, to a command center (sink) either di-rectly or through a data concentration center (agateway). The decrease in the size and cost ofsensors, resulting from such technological ad-vances, has fueled interest in the possible use oflarge set of disposable unattended sensors. Suchinterest has motivated intensive research in the pastfew years addressing the potential of collaborationamong sensors in data gathering and processingand the coordination and management of thesensing activity and data flow to the sink. A naturalarchitecture for such collaborative distributedsensors is a network with wireless links that can beformed among the sensors in an ad hoc manner.Networking unattended sensor nodes are ex-pected to have significant impact on the efficiencyof many military and civil applications such ascombat field surveillance, security and disastermanagement. These systems process data gathered*Corresponding author.E-mail addresses: [email protected] (K. Akkaya),[email protected] (M. Younis).1570-8705/$ - see front matter  2003 Elsevier B.V. All rights reserved.doi:10.1016/j.adhoc.2003.09.010Ad Hoc Networks 3 (2005) 325–349www.elsevier.com/locate/adhocfrom multiple sensors to monitor events in an areaof interest. For example, in a disaster managementsetup, a large number of sensors can be dropped bya helicopter. Networking these sensors can assistrescue operations by locating survivors, identifyingrisky areas and making the rescue crew more awareof the overall situation. Such application of sensornetworks not only can increase the efficiency ofrescue operations but also ensure the safety of therescue crew. On the military side, applications ofsensor networks are numerous. For example, theuse of networked set of sensors can limit the needfor personnel invo lvement in the usually dangerousreconnaissance missions. In addition, sensor net-works can enable a more civic use of landmines bymaking them remotely controllable and target-specific in order to prevent harming civilians andanimals. Security applications of sensor networksinclude intrusion detection and criminal hunting.However, sensor nodes are constrained in en-ergy supply and bandwidth. Such constraintscombined with a typical deployment of largenumber of sensor nodes have posed many chal-lenges to the design and management of sensornetworks. These challenges necessitate energy-awareness at all layers of networking protocolstack. The issues related to physical and link layersare generally common for all kind of sensorapplications, therefore the research on these areashas been focused on system-level power awarenesssuch as dynamic voltage scaling, radio communi-cation hardware, low duty cycle issues, systempartitioning, energy-aware MAC protocols [6–10].At the network layer, the main aim is to find waysfor energy-efficient route setup and reliable relay-ing of data from the sensor nodes to the sink sothat the lifetime of the network is maximized.Routing in sensor networks is very challengingdue to several characteristics that distinguish themfrom contemporary communication and wirelessad hoc networks. First of all, it is not possible tobuild a global addressing scheme for the deploy-ment of sheer number of sensor nodes. Therefore,classical IP-based protocols cannot be applied tosensor networks. Second, in contrary to typicalcommunication networks almost all applicationsof sensor networks require the flow of sensed datafrom multiple regions (sourc es) to a particularsink. Third, generated data traffic has significantredundancy in it since multiple sensors may gen-erate same data within the vicinity of a phenom-enon. Such redundancy needs to be exploited bythe routing protocols to improve energy andbandwidth utilization. Fourth, sensor nodes aretightly constrained in