Energy-Efficient Routing Approach for Wireless Sensor Networks

Energy-Efficient Routing Approach for Wireless Sensor Networks Daniel Fuentes A. Fernandez-Montes Juan A. Ortega Luis Gonzalez-Abril ETS Ing. Informatica ETS Ing. Informatica ETS Ing. Informatica EU Estudios Empresariales Univ. de Sevilla Univ. de Sevilla Univ. de Sevilla Univ. de Sevilla [email protected] [email protected] [email protected] [email protected] Abstract Wireless Sensor Networks (WSNs) are popular nowadays for multiple applications which include environmental monitoring, seismic detection and surveillance. The nodes in WSNs deplete of their very limited energy over time. Current approach proposes a method to save energy in a WSN by reducing the number of sensor transmissions. During a training period, the central server of the network analyzes arrived messages routes to determine routing rules for each sensor. After the training time, the server reconfigures the network adding the computed rules to the nodes to improve the message routing. Through the numerical results, we show that our approach can save up to 25.5% of the energy consumed by the nodes. 1 Introduction Since sensors typically carry their own power in Wireless Sensor Networks applications, it is crucial to reduce the energy consumption so lifetime of the sensor network can be maximized. Recently, several studies have considered energy efficient sensor networks. Data aggregation removes redundancy to improve bandwidth usage and energy-efficiency of sensor nodes as shown in [1-2]. A hardware model for wireless sensor nodes and the design of physical layer aware protocols, algorithms, and applications that minimize energy consumption of the system are described in [3]. Others methods save energy by reducing the number of sensor transmissions. In [4], the sensor transmissions are sorted according to the magnitude of their measurements, and the sensors with small magnitude measurements, less than a threshold, do not transmit. Other works propose sleep modes by adjusting sensing range to maximize the total number of rounds [56]. There are multiple application environments of energy-efficient wireless sensor networks and different routing techniques for them [7-8]. In the rest of the paper, we provide an overview of the objectives in Section 2. Section 3 describes our energy-efficient routing proposal. Implementation and test results are related in Section 4 and finally, concluding remarks are made in Section 5. 2 Objectives The main goal of this routing protocol approach for WSN is saving batteries while sending messages through the network. This way we can establish the best route a message must done to arrive its destination. Secondary goals are not-overloading the network by minimizing configuration messages needed to setup sensors routing constraints in an automated online way, and on the other hand, to free the administrator from manually setup the network. Finally, the method covers situations which imply one or more nodes. 3 Routing proposal The architecture of our sensor network is illustrated in Figure 1. We consider next elements: •A set of nmotes or sensors, S1, S2, ..., Sn. Devices that sense on aspects of the physical world like the humidity or temperature levels. •The home server, Hthat stores aspects of the mote network state and acts as a proxy to communicate between mote network and client applications. •A set of client applications, Athat interfaces with the host server to communicate with nodes, to manage existing applications or deploy new ones. Server H S 2 S 3 S 4 S 1 m 1 m 1 ' Client Application 1 Client Application 2 Client Application 3 Figure 1: Representation of typical WSN. A message is sent from S1node to Hserver. We assume that sensors interact each other through broadcast messages. A message from a sensor can be captured by all sensors that are into its coverage. The server can communicate with sensors within its scope, but client applications only communicate with the server. The size of a message is limited but it can be configured and modified to include or exclude information. The main goal of this routing proposal is the optimization of the number of messages in order to save batteries in the nodes by a previous training period. This way the H server computes rules for each sensor to determine when is necessary to forward a received message. When the training-time begins, all the sensors start sending messages to the server. In every message, the description of the route (nodes list, where a message has been routed) is attached to the message. When a message arrives to the Hserver, the message contains the whole route from the origin sensor to the server H. With this information, the Hserver can establish some rules to improve the communication between the nodes. For example, let’s consider two routes R1and R1 0of two copies m1and m1 0of the same message respectively: R1= [S1, S2, S3, S4, H] R1 0= [S1, S2, S4, H] With these sequences, the server can detect that it is not necessary that the S3sensor forwards the message m1. Hence, the rule "Don’t forward a message if the previous nodes sequence is [S1, S2]"will be established for the S3sensor when the server reconfigure the network. In this manner, the server H continues receiving messages and computing new rules until a time threshold is exceeded. In this case, the nodes stop sending messages to the server and the reconfiguration of the nodes starts. In this stage, the server establishes the computed rules to the nodes and the network is rebooted with the new configuration. Another simple example is shown in Figure 1. The S1node sends a message to the Hserver. This node sends two copies, m1and m1 0to the nodes that there are in its S3and S2respectively. When m1and m1 0arrive to the Hserver they contain the next routes R1and R1 0: R1= [S1, S3, H] R1 0= [S1, S2, S4, H] The Hserver checks that R1route is shorter than R1 0. It is no necessary that the S2node forward the m1 0message. Hence, the rule that Hserver will establish in node S2will be: "Don’t forward a message if the previous nodes sequence is [S1]" Obviously, during the training time, the decrease of the energy consumption is null. However, after this period, the key limitations of WSNs, the storage, power and processing, are treated. The server reconfigures the nodes with the calculated rules and the communication in the network. An efficient energy consumption is obtained through the reduction of the sent messages. With this optimization, in 22 the last example the message m1 0will not arrive to server. Hence, duplicated messages in the network are avoided and consequently, the information load in buffers, the time processing and the battery consumption in sensors are reduced. Furthermore, by deleting data redundancy and minimizing the number of data transmissions, there is less data load in the network, the possibilities of an overload are lower and the message management is simplified. Usually, communications in a WSN are only in one direction, from sensors to the host. However, sometimes the host needs to send a message to a specific sensor, e.g. the configuration of parameters. In this case, the server can use the previous computed rules to established a predefined way to send the message in the fastest way avoiding duplicated messages and overloading the network. Periodically, the Hserver sends a message to each node asking for its state. Hence, if a node stops working due to an attack, the lack of battery or another external factor it will not send any reply. In this case, the Hserver will start the whole training process again using the rest of nodes and computeing new rules. 4 Implementation, tests and results For test purposes we have implemented the routing proposal in Sentilla Work, the IDE supplied with Sentilla Development Kit. The real implementation is an extension of one of the protocols supplied by Sentilla, although the shown algorithm it was written in Java. //Called when the Dispatcher receives //a message for this protocol. public void stackReceive(Receiver rcvr){ //Get the TestMessage RoutedMsg msg=(RoutedMsg)rcvr.getData(); //Check hasn’t been received already //and route is allowed if((!msg.equals(prevMessage)) && !msg.currentRoute.equals(restrictedRoute)){ //Adds current sensorId to the route msg.currentRoute.addElement(id); //Not-relevant operations TestReceiver meta=new TestReceiver(); meta.addMetadata(rcvr); meta.data = msg.data; //This will forward the received data ds.dispatch(meta); prevMessage = msg; } } Algorithm 1: Forwarding routine. Results are shown in the table below. Tests have been done at the WSN installed at Computer and Systems Languages department of the University of Sevilla. It consists on 9 Zigbee sensors deployed in 9 offices. The retrieval of data was done during 360 hours (15 days). We would like to emphasize that 17.9% of messages were avoided from being sent through different routes and 25.5% of hops were avoided. Base Efficient #sensors 9 9 timeComputed 360h. 360h. #msgsPerHour 30 30 #msgs 10800 10800 trainingTime 0h. 1h. #duplicatedmsgs 1944 5 avgStepsPerMsg 2,3 2,02 #msgsReceived 12744 10805 #msgsAvoided 0 1938 %avoidedmsgs 017,9 #Steps 29311 21816 #StepsAvoided 0 7495 %avoidedSteps 025,5 Table 1: Base-Protocol and Efficient-Protocol comparison 5 Conclusion In this paper, we have presented a energy-efficient method for wireless sensor networks. We have got to save up energy by reducing the forwarding of messages. First, the central server determines the rules through the information of the messages routes. Second, the network is reconfigured avoiding duplicated messages. The paper describes practical experiences with implementations over Sentilla sensors base in Zigbee technology. 23 References [1] H. Çam, S. Özdemir, P. Nair, D. Muthuavinashiappan and H. O. Sanli. Energyefficient secure pattern based data aggregation for wireless sensor networks. In Computer Communications 29, pp. 446 - 455, 2006. [2] S. Chatterjea, T. Nieberg, M. Nirvana and P. Havinga. A distributed and self-organizing scheduling algorithm for energy efficient data aggregation in wireless sensor networks. In ACM Transactions on Sensor Networks (TOSN) 4(420), 2008. [3] E. Shih, S. Cho, N. Ickes, Rex Min., A. Sinha, A. Wang and A. Chandrakasan Physical Layer Driven Protocol and Algorithm Design for EnergyEfficient Wireless Sensor Networks. In Proceedings of the 7th annual International Conference on Mobile Computing and Networking, pp. 272 - 287, 2001. [4] X. Chen, R. Blum and B. M. Sadler. A New Scheme for Energy-efficient Estimation in a Sensor Network. In Proceedings of the Conference on Information Sciences and Systems (CISS), 2009. [5] M. Lu, J. Wu, M. Cardei and M. Li. EnergyEfficient Connected Coverage of Discrete Targets in Wireless Sensor Networks. In Proceedings of the International Conference on Computer Networks and Mobile Computing (ICCNMC), 2005. [6] G. Mathur, P. Desnoyers, P. Chukiu, D. Ganesan and P. Shenoy. Ultra-low power data storage for sensor networks. In Proceedings of the 5th international conference on Information processing in sensor networks, pp. 374 - 381, 2005. [7] N. A. Pantazis, S. A. Nikolidakis, and D. D. Vergados. Energy-Efficient Routing Protocols in Wireless Sensor Networks for Health Communication Systems. In Proceedings of the 2nd International Conference on PErvsive Technologies Related to Assistive Environments (34), 2009. [8] K. Zeng, K. Ren, W. Lou, and P. J. Moran. Energy Aware Efficient Geographic Routing in Lossy Wireless Sensor Networks with Environmental Energy Supply. In Proceedings of QShine06), 2006. 24