Abstract— The area of wireless sensor network (WSN) isshrouded with the issues of routing protocol that has been witnessed in thelast few decade in the research community. One of the direct influences ofrouting protocol is the performance of data aggregation in WSN. This paperpresents a novel routing protocol that performs a cost effective, reliable androbust routing mechanism in wireless sensor network with its highlycompatibility with in-network data aggregation mechanism. A simulation study isperformed Matlab, where along with proposed routing algorithm, two moreconventional algorithm is chosen to perform the performance comparativeanalysis. The performance parameters selected for simulation study shows thatthe proposed routing protocol is robust against the conventional routingprotocol with respect to efficiency and data delivery phenomenon.
Keywords-component; In-network Data Aggregation, Routing Protocol, WirelessSensor Network I. INTRODUCTION A wireless sensor network consists of a sensornodes that has the perceptibility characteristics for some of the physicalattributes e.g. heat, motion, pressure, smoke, moisture etc 1. A sensor nodecan be termed as very small devices which has very low computational capabilityalong with less availability of resources 2. Here resources mainly meanenergy, bandwidth, and memory.
The prime operation that are performed by thesensor node is called as data aggregation 3, which is a process where thesensor node collects the physical information from the surrounding area andforwards it to the base station. Normally, such forms of data aggregationoccurs using clustering. Each cluster consists of a specific number of sensornodes and cluster heads. A sensor node (which is also called as candidatemember node) is responsible for capturing the raw data from the surroundingarea to encapsulate certain events 4. This collected data are then forwardedto cluster head by removing the redundancies. However, it doesn’t happen soeasily. In wireless sensor network, there are various forms of routing protocolthat are responsible for making the communication happens 567. Theserouting protocols make the decision whether to forward the aggregated data tothe base station or to some other cluster head.
The prime reason behind this isto remove data redundancies in order to avoid overhead towards the basestation. These mechanisms indirectly conserve a significant amount of batterylifetime of a sensor node. The biggest challenge is there are very few andstandardized energy-efficient routing protocol in wireless sensor network.According to the theory, standard books, and standard research journals, thefrequently used standard routing protocol that has the supportability of energyefficiency are LEACH, PEGASIS, TEEN, and APTEEN 8. These routing protocolsare said to significantly conserve energy on the defined test-bed. Although allof them are has started it is based from efficient clustering, it significantsaves energy. Another interesting observation that we have made is that 98% ofthe research papers that deals with energy-efficient routing considers LEACH,whereas in reality there are many variants of LEACH as well as other energyaware routing techniques.
LEACH is the one of the standard hierarchical routingprotocol that was tested on Berkley nodes considering the 1st order radioenergy model 9 designed using core RF antenna circuitry system. This willmean that LEACH is tested in real-time, whereas the original versions of otherrouting protocols were basically tested from simulation viewpoint. Even therecent variants of LEACH were also enhanced from soft computational viewpointand were validated only empirically and not experimentally. This is the primereason, why 98% of the research manuscript chooses to compare their energyefficient techniques with LEACH. Due to such differences, many new algorithmshave been proposed for the routing problem in WSNs. These routing mechanismshave taken into consideration the inherent features of WSNs along with theapplication and architecture requirements.
The problem definition of theproposed system can be stated as-“It is quite a computationally challengingtask to design a framework that can ensure energy efficient and reliablerouting protocol that can accomplish superlative innetwork data aggregation inwireless sensor network.” The discussed problem is basically intended to bemitigated by the proposed system. This paper proposes an energy efficientscheme called as REEDA that is essentially meant for controlling unwantedenergy dissipation for large scale wireless sensor network. The study hasfocused on achieving energy efficient in-network data aggregation to showbetter performance in network lifetime.
Section II discusses about some of therelevant literatures pertaining to energy efficient routing followed bydiscussion on proposed system on Section III. Algorithm implementations arediscussed in Section IV while result outcomes were discussed in Section V.Finally, summary of the paper is made in Section VI.II. RELATED WORK This section presents the prior researchwork that has been carried out in the area of energy-efficient routing and dataaggregation in wireless sensor network. Our prior work has reviewed some of theexisting issues in wireless sensor network along with different approaches ofenergy conservations 10. Our recent work has introduced a technique called asMLO i.
e. Multi-Level Optimization that has significantly enhanced networklifetime for large scale wireless sensor network 11. Jian et al. 12 havedeveloped an optimization technique based on bio-inspired algorithms. Thetechnique was mainly meant for enhancing the lifetime of radar sensors.Enhanced version of ant colony optimization was used in the design of theproposed study.
The outcome of the proposed study was compared with LEACHalgorithm to find the technique better with respect to energy conservation.Similar direction of the study was also carried out by Mao et al. 13, wherethe authors have focused on developing energy efficient routing forheterogeneous sensor network. The protocol design was carried out in GloMoSimand is found to have slight improvement of existing routing techniques in thosetimes with respect to energy efficiency. Masazade et al. 14 have discussed atechnique that uses Cramer-Rao bound rule along with Monte Carlo mechanism forsolving localization problems. Interestingly, the entire focus of the study wasmainly to retain maximum energy of the nodes. Srivastava et al.
15 haveproposed a mechanism that can perform controlling of the transmitter node foreffective scheduling of the data packets. The design of the system model iscompletely based on the control information using a unique feedback scheme thatgives the response of energy required to perform data aggregation. The outcomeof the study was found to be energy efficient however, it was done withoutbenchmarking. Wang et al.
16 have presented a both decentralized as well ascentralized scheme for scheduling the routing channels of the sensors consideringthe mobility factor. The entire stress of the study was to achieve energyefficiency in the mobility condition. The outcome of the study was evaluatedusing number of alive nodes and energy mainly and it shows good minimization ofenergy. The study carried out by Wang et al. 17 has addresses the ambiguityfactor in the energy dissipation for long term utility in energy harvesting ofwireless sensor network.
The design was built using ZigBee module, MCU Atmegamicrocontroller, Battery Voltage monitor, ADC etc.The uniqueness of this study is the discussion of powerexpenditure profiling system that is found to quite helpful even in otherresearch work too. It assists to understand the relationship between energyconsumption and data transmission. Dziengel et al. 18 have presented anotherunique study where a distributing surveillance method is presented with focuson energy efficiency accomplishment. The technique is totally hardware basedapproach. The outcome of the study basically shows the detection rate andimpact of data transmission over energy. Liu et al.
19 have discussed atechnique for addressing energy depletion in wireless sensor network. Theschema presented by the author has mainly three essential components i.e. i)reward computation, ii) punishment computation, and iii) decision making asshown in Fig.2. Figure 2 Schema presented by Liu et al. 19 The study towardsenergy efficiency was also seen in the work of Rezai 20 and Jain et al. 21.
An analytical modelling is presented in these studies where Adhoc routingprotocols were seen to be used in wireless sensor network in order to ensureenergy efficiency. It is to be noted that Adhoc routing protocols were mainlyimplemented in mobile Adhoc network. Most recently, Jorio et al. 22 havepresented a study that focuses jointly on clustering, routing, and energyefficiency. The study has introduced a new hierarchical routing technique justlike LEACH in order to conserve energy while performing clustering. The studyhas established an empirical relationship between clustering and energyefficiency which are directly connected to each other. The study says ifefficient clustering is performed, energy is significantly conserved or else itdrains and drastically minimizes the network lifetime of wireless sensornetworks.
Hence, it can be seen that there are massive archives in theliteratures that deals with energy efficiency. Each technique has its ownadvantage factor as well as limiting factor. However, we find that less focusis laid towards innetwork data aggregation energy efficiency while performingrouting. Hence, the presented paper is a continuation of the research workwhere the emphasis is laid over attaining more energy conservation with somerobust benchmarking aspects in wireless sensor network.
III. PROPOSED SYSTEM The prime goal of the study is todesign a reliable and efficient routing protocol that can perform energyefficient data aggregation considering in-network approach in wireless sensornetwork. Therefore the proposed system is termed as REEDA i.e.
routing withEnergy Efficiency in Data Aggregation. In order to accomplish the discussedgoal, following objectives were ascertained, • To perform an in-depthinvestigation of the prior literatures that has attempted to solve the similarissues in order to understand the effectiveness of the techniques used. • Todesign a simulation test bed that can perform the simulation using thecluster-based approach considering indicative simulation parameters. • Todesign an election method of cluster head for performing in-network dataaggregation. • To apply shortest path algorithm for route generation. • Toimplement an algorithm for maximizing the aggregation point and to use lessernumber of control packets to build the routing tree. The system architecture ofREEDA is shown in Fig.
3. The design of the architecture uses a coworker nodethat is responsible for identifying any significant events as well as transmitsthe aggregated data to the controller node. The controller node is responsiblefor gathering the aggregated data and forward to the base station. The designof REEDA also consists of a relay node which represents the sensor nodetransmitting the data to the base station.
A Steiner tree is designed thatessentially performs mapping of the operational behavior of in-network dataaggregation along with some of the significant functionalities e.g. Designingof Graph model, clusters design, generation of stabilized routes, andrectification of unstabilized routes.
The next section discusses about thetechniques applied for designing the proposed REEDA architecture. IV.IMPLEMENTATION TECHNIQUES The implementation of the proposed study of REEDA iscarried out over Matlab.
In order to accomplish the design requirements of REEDA,following implementation steps were carried out: • Designing of Graph Model: Aspatial-based approach is applied for computing the distance from the basestation to all the sensor nodes. The base station forwards a control messagethat is essentially consisted of node identifier and distance. A graph usingtreebased approach is used that maps various hops to be formulated. The basestations than forwards a control message with value 1 and with the lowestenergy.
After receiving this message, the sensors reconfigure their value as 1.This phenomenon happens in first event. However, in second event, the basestation increases the value to be 2 for the control message, which is alsoupdated by the sensor nodes receiving it. A hello message consisting of maximumand minimum limit of each value is set and each sensor estimates itsEuclidean’s distance from the sink using received signal strength. Thereby aspecific graph model is formulated with hierarchical edges as well as verticesto support the in-network data aggregation.
• Cluster Design: After asignificant event is detected in the simulation area, all the correspondingcandidate as well as cluster head will actively participate in innetwork dataaggregation process. Exclusively for the generation of the initial event, theselection of the cluster head is done on the basis that candidate sensor nodeshould be in the nearest distance from the base station. However, in case ofsimilar distance, the selection of the cluster head will be carried out on the basisof the node bearing smallest identifier. In the ultimate cycle of the clusterhead selection process, the system will select only one candidate node to beeligible as cluster head. The cluster head will now act as controller nodewhile the other nodes that are not selected as cluster head will act ascoworker node. The controller node will aggregate all the significantinformation from the coworker node and will forward them to the base station.
The phenomenon significant reduces the work load of cluster and therebyconserves significant amount of energy required in data aggregation in WSN(Fig.4 and Fig.5). • Generation of Stabilized Routes: We define stabilizedroutes are a route that is established between the nodes with sufficientcut-off residual energy.
The cut-off energy values can be different fordifferent applications and thereby we don’t discrete discuss about the cut-offvalue. Our range is fixed around 0.3 Joule for cut-off energy value. The simplealgorithm is used for performing data transmission as exhibited in Fig.6. Thecluster head initiates formulating new communication channel for the purpose ofdisseminating events.
In this situation, the controller node transmits acontrol message for generating a communication channel to its immediate edgesin the Fig.10 shows that LEACH has sharp fall of residual energy owing to thecentrality of the base station position. Intanagonwiwat et al. 23 approachwas found better than LEACH as it supports directed diffusion avoiding muchdata redundancies and thereby conserves more power. But cumulatively, REEDA hasoutperforms both Intanagonwiwat et al. 23 and LEACH.
The prime issue ofIntanagonwiwat et al. 23 approach was that frequent dissemination of data forevery generation in the simulation area in order to update the other nodes.This action required additional energy while performing routing and thereforethis charecteristics restricts the energy efficient performance of theIntanagonwiwat et al. 23 approach.
The proposed REEDA uses identification ofthe unstabilized routes and then it performs rectification of unstabilizedroutes and substitutes it by exploring more robust route. Moreover REEDA has nodependency on the position of the base station making it more flexible tosupport better energy efficient routing. The outcome of the study washypothetically compared with EHE-LEACH too 24, which has supportability ofenergy efficiency for heterogeneous sensor network also.
The outcome shown inREEDA excels better in comparison to Intanagonwiwat et al. 23, LEACH 9,andEHE-LEACH 24 with respect to throughput, energy efficiency, and processingtime. VI. CONCLUSION This paper discusses about an energy efficient routingtechnique that is essentially focused on achieving better innetwork dataaggregation. The proposed system has introduced different type ofcharacteristic behavior of the sensor nodes in order to reduce the load ofcluster head during data aggregation in order to retain more amount of energy.The proposed system is also capable of identifying the unstabilized routes andreplaces the unstabilized routes with more energy efficient routes. Theemphasis of the proposed study was purely on energy-efficiency, however, it isequally important to understand the necessary impact of other parameters e.g.
bandwidth, QoS parameters e.g. propagation delay, etc on energy effectivenesstoo. Our future work will be to further consider these parameters forincreasing the scope of outcomes on energy effectiveness. We are also planningto implement a novel design of energy optimization considering the physicallevel of the sensor network, which is quite challenging to achieve till date.We will also investigate the possible applicability of energy efficiency onexisting security protocols frequently in used. As we strongly believe thesecurity protocols do have higher consumption of energy, hence, it is importantto ensure that our future work should address all these issues.