Document not found! Please try again

Energy Efficient Deployment of Wireless Sensor ... - IEEE Xplore

8 downloads 0 Views 517KB Size Report
Amrita Robotic Research Center. Amrita School of Engineering. Amrita Vishwa Vidyapeetham. Bengaluru Campus, India [email protected]. Anu George.
2015 Intl. Conference on Computing and Network Communications (CoCoNet'15), Dec. 16-19, 2015, Trivandrum, India

Energy Efficient Deployment of Wireless Sensor Network By Multiple Mobile Robots Rajesh M Amrita Robotic Research Center Amrita School of Engineering Amrita Vishwa Vidyapeetham Bengaluru Campus, India [email protected]

requires complex calculations. So an independent placement of sensor nodes to create a network by robots is performed. This is done with the support of communication over nodes of WSN which increase the coverage area, accurate localization of robots even when they are out of anchor nodes and faster completion of the process. The success of the application depends on the optimized deployment of sensor nodes. For deployment in an unknown territory using autonomous multiple robots requires an efficient exploration strategy which helps to build a map of the area. Frontier based exploration strategy is used with localization procedures to explore the unknown territory. Location information is obtained by the robot using trilateration method in which the distance of the robot from the anchor nodes is computed by using the RSS. The sensor nodes once placed in turn act as anchor node and helps the robot to map a very large area. The use of Zigbee protocol helps to reduce the energy consumed for localization and thus helps to achieve an energy efficient deployment of sensor node to create a WSN. Optimum positioning of sensor nodes is achieved by communication between already placed sensor nodes and the mobile robots. This significantly reduces the overlapping of the regions of sensors and redundant exploration of a region. Thus the territory is explored and sensors are optimally placed by reduced number of robots and in lesser time.

Abstract— Disaster management is one of the most critical applications that can be performed by a Wireless Sensor Network (WSN).The optimized deployment of sensor nodes is required for the successful relay of information. This paper proposes the deployment of sensor nodes by multiple autonomous mobile robots in an unexplored large disaster prone territory. The use of multiple robots provides important advantages over human-assisted placement like safety, accurate positioning and flexibility. For accurate location of an event, localization of the sensor nodes is very important which is achieved by using Received Signal Strength (RSS) from anchor nodes and the sensor nodes which act as anchor node after being placed. Placement of node by robots helps in achieving the location information of all the nodes that makes up the network. Communication and coordination between the multiple robots over the sensor nodes is used to achieve accurate localization, faster exploration and network creation. In this study, energy efficient utilization of the sensor node is achieved when it acts as an anchor node as it only responds when it receives the node discover command which a property of Zigbee protocol from the robot placing the sensor nodes. Hardware simulation of the proposed scheme is carried out by using Firebird V robots and Zigbee protocol is used for communication and coordination between the robots.

Keywords—Robotic Exploration, Wireless Sensor Network(WSN), Sensor Node Deployment Strategies, Localization, Firebird V

I.

II.

PREVIOUS WORK

A Wireless Sensor Network (WSN) is a network that consists of a large number of devices called sensor nodes each with sensing, thinking and wireless communication capacity. The network system can communicate and use data gathered mutually by each sensor node distributed spatially to achieve the purpose of the application. An ad-hoc network can be constructed only when each sensor node is connected to one or more nodes via wireless communication and this is possible only when a large number of nodes are deployed in the network. T. Suzuki et al. in [1] explains that when compared to a wired and fixed network, the process of enhancing a WSN in which nodes are deployed densely is much faster. This is a major advantage of WSNs as compared to other networks. The information collected by the sensor nodes are processed to

INTRODUCTION

When a disaster occurs, one of the most important concerns is a good search and rescue operation with a high level of meticulousness, timeliness and safety for dealing with both the casualties and the rescuers. They are inherently dangerous tasks to be carried out by humans. So robots can be used to help carry out the tasks of rescue. Robots perform this task by the deployment of sensor nodes for monitoring the environment and when the disaster strikes the already deployed sensor nodes can be used monitor the scenario and help robots or humans to carry out the rescue immediately without any delay. Manual sensor deployment represents a rather difficult task to achieve since the optimal deployment of sensor nodes

978-1-4673-7309-8/15/$31.00 ©2015 IEEE

Sudarshan T.S.B. Amrita Robotic Research Center Amrita School of Engineering Amrita Vishwa Vidyapeetham Bengaluru Campus, India [email protected]

Anu George Amrita Robotic Research Center Amrita School of Engineering Amrita Vishwa Vidyapeetham Bengaluru Campus, India [email protected]

72

2015 Intl. Conference on Computing and Network Communications (CoCoNet'15), Dec. 16-19, 2015, Trivandrum, India

strategies and they showed that the performance of the proposed method was better than both greedy method and generic tsp method. For minimum traversal, the robot traverses through the regions where the overlap of sensing range of placed sensor nodes is maximum. Gurkan Tuna et al. in [6] explains that it is advantageous and safe to use intelligent mobile robots to construct a WSN for disaster relief applications. This system has the benefit of autonomous deployment, cumulative intelligence of multiple robots and flexibility. During the deployment of sensor nodes, mobile robots perform localization and mapping of the area at the same time and communicate over the nodes. The extent to which these envisaged gains are realized depends on the information relaying and coordination capabilities of the system. When the area to be covered is large, communication between the robots is facilitated by the sensor nodes that constitute the WSN. Thus this helps the robots to overcome their dependence on the existing network infrastructures. Exploration of the unknown region is a primary requirement for mapping an unknown region. There are different exploration strategies, such as frontier-based exploration, market-driven exploration, and role based exploration that be employed depending on the governing factors for the situation. On comparing the three exploration strategies, exploration based on frontiers (i.e. the dividing line between explored and unexplored region) provides a proficient way to explore a real world scenario such as regions where walls and obstacles may be in arbitrary orientations. In the above approach proposed by B. Yamauchi in [8], each robot maintain their own their own global maps and decisions about where to explore is made without depending on data given by other robots. Whenever a robot arrives at a new boundary, it shares the newly gained information to all others within its communication range and also adds the information it receives from other robots into its own global map. Continuous localization using evidence grids which are Cartesian grids specified as cells can be incorporated with frontier-based exploration. Even when they are exploring unknown territory using this approach, the robots are able to have an accurate position estimate since localization is also carried out simultaneously. This method has the advantage of being both cooperative and distributed processing. Redundant exploration is avoided as the information obtained by any robot is communicated to all other robots that are within its communication range. To construct a WSN in an unknown territory such as a battlefield, deployment of sensor nodes is carried out during the exploration of the deployment zone. During exploration, the communication between the autonomous mobile robots is maintained over the sensor nodes. This allows the robots to remain connected to relay information even when they are not in the communication range of each other. For the success of the application, the optimum deployment of sensor nodes such that the entire area of application is covered with no sensing holes plays a critical role. To find out if the entire area is covered, each robot provides a map with the location of the deployed nodes.

provide various services. Because of the lower cost of implementation of WSN technology, it is applicable within many fields including disaster relief application. WSN helps to add intelligence into ambient environments. The autonomous structuring of WSNs is one of the most successful methods that have been formerly discussed in conventional studies on deployment. McMickell et al. in [3] proposes different methods of deployment which mainly consist of scattering many low-cost sensor nodes randomly. But in this method, there is a risk that the sensor nodes may not essentially be placed in the relevant locations. Thus it results in holes in the network and no events will be detected from these areas. The presence of holes results in bad network connectivity and inability to communicate the relevant information to the sink in the network. Another main shortcoming of random scattering of nodes is that it can result in redundant coverage of an area and hence two or more nodes will be reporting the same event. This wastage of energy in redundant communication will in turn reduce the lifetime of the WSN. Although it may be possible to conquer the problem of holes by scattering a huge number of sensor nodes, the costs of increased unwanted nodes covering the same area and redundant communication traffic associated with the deployment of a large number of sensor nodes may prove challenging. When the WSN is composed of mobile sensor nodes and a node is not able to communicate to any other node, then it can be directed to move to an appropriate location and thus resulting in better connectivity. However, mobility of sensor nodes increases its cost and to be deployed in large numbers becomes a very expensive job. From an application perspective, if the sensor nodes need not move to gather information, then the use of mobile nodes to cover holes is unwanted expenditure. In general, when a large number of sensor nodes are needed to gather data over a wide area then cost of the system becomes an important concern. Therefore, it is very important to lessen the expenditure on the sensor nodes and also the energy consumed by each to increase the lifetime of the network. The energy cost problem which is created by redundant deployment to cover holes can be solved by empowering autonomous mobile robots with localization capabilities to build WSNs. Mobile robots into which intelligence is embedded are being used for a variety of tasks like reconnaissance, rescue operations in disaster zones and monitoring of hazardous areas such nuclear radiation leakage sites to name a few. Arnoud Visser et al. in [2] explains that human operators can control the robots remotely but a minor misinterpretation can result in huge unrecoverable dangers. Also in the case of very large remote environments, the delay caused in relaying the control information can be significant. As a result, there is much encouragement to pass on some of the load to the robots, and partial or full independence of the robots is an advantageous capability in many scenarios. Y. Wang, C-H. Wu in [4] proposes deployment of nodes and data collection system for unmanned explorations or monitoring by a mobile robot. Based on the length of the path travelled by the robot, the authors evaluated three deployment

73

2015 Intl. Conference on Computing and Network Communications (CoCoNet'15), Dec. 16-19, 2015, Trivandrum, India

that comes within its communication radius that is fixed. • Line of Sight communication model: Only if there are no obstacle between two robots can communicate with each other. In practical scenarios, it will pose an important limitation. • Propagation model: In this model, depending on the thickness of obstacles in the environment communication range is computed. This is a realistic model of communication. For wireless communication between robots, sensor nodes and between robots and sensor nodes a communication protocol is to be used. Zigbee is a wireless communication technology which is designed to operate in 2.4 GHz ISM band used for Personal Area Networks (PAN). This is built with IEEE 802.15.4 standard as the foundation i.e. the application layer and network layer is defined by Zigbee but the physical layer and MAC layer is defined by IEEE 802.15.4. When compared with Bluetooth, Wi-Fi and GSM, the major advantages of this protocol are its low power consumption and high network security. Since in WSN, power consumption is one of the most important criteria, the use of Zigbee as communication protocol increases the lifetime of the network. The data rate is 250 kbps which is sufficient for WSN where only intermittent transmission of data is required. The commands incorporated in application layer helps to implement energy efficient localization and optimum exploration strategies. Zigbee supports a very large number of devices about 64,000 in a single network. There are three kinds of nodes in a Zigbee network – Zigbee Coordinator (ZC), Zigbee Router (ZR), Zigbee End Device (ZED). • Coordinator (ZC): Its function is to set up the network by allowing associations from reduced-function nodes that are capable of only relaying data and maintain routing tables. • Routers (ZR): They can talk to the coordinator, to other routers and to Zigbee end devices. • End devices (ZED): They can communicate with routers and the coordinator of the network of which it is a part of, but not to each other. A Zigbee network can have different topologies such as peer-to-peer, tree, star and mesh. Mesh topology is the most ideal as it facilitates to have more than a single path from source to destination and guarantees the delivery of data which is very important for applications such as surveillance. We propose to deploy sensor nodes using multiple Firebird V robot which has a LPC2148 microcontroller and Xbee-Pro module which is used for wireless communication using Zigbee protocol. We deploy sensor nodes equipped with Zigbee wireless module so that it can respond to the queries of the robot. Anchor nodes respond to the queries of the robot instead of continuously sending out beacons with information regarding its location. This helps to conserve energy of the network and also extend the battery life of the sensor nodes which act as anchor nodes after deployment.

Oguejiofor O.S et al. in [7] states that for effective use of relayed data, information of the position of sensor nodes is one of the most important requirements. The methods for localization can be widely divided into two groups. • Range methods in which the coordinates of the node is found out by computing the distances between two nodes. • Range-free methods in which distance is not used as a criterion to find the position. In range-based localization methods, the features of the communication signal that are frequently used are Angle of Arrival (AOA), Time of Arrival (TOA), and Time Difference of Arrival (TDOA). These methods necessitate the need for expensive hardware making it too expensive for WSNs. The simplest method to obtain location information of a blind node i.e. the node with no location data is to compute the distances to three nodes with known positions (called anchors or beacons) using the Received Signal Strength Indicator (RSSI) value and then use trilateration. In more advanced methods such as DV-Hop (Distance Vector), DV-Distance and Euclidian propagation methods it requires less severe assumptions. The main disadvantage of these methods is that they cost much more communications and in turn more energy consumption. There is no single criterion of a single best algorithm for localization in sensor networks. The algorithm is decided based on the environment in which WSN is constructed and the specifications of the motes used. A distributed beaconbased localization algorithm known as trilateration is used. Since RSSI measurements don’t need any extra hardware or line of sight, it is used to find out the range between nodes. Multipath and shadowing are two major phenomena that affect the reliability of RSSI measurements. Both constructive and destructive interferences are caused when signals of different magnitude from the transmitter arrive out of phase at the receiver. However, by averaging the received power over multiple frequencies spread spectrum radios have effectively mitigated these interferences. To smoothen out the fluctuation in RSSI values Erin-Ee-Lin Lau et al. in [12] proposes a two phase method. In the first phase a calibration is done to find out the path loss exponent for each of the anchor nodes and then smoothening algorithm is used to obtain a stable RSSI value by using at least 8 of the best RSSI value. This necessitates a high anchor-node ratio. After localization is done, a sensor node is deployed and the information is logged into their local map and whenever other robots are in its communication range (the deployed sensor nodes also act as connecting nodes in the path between robots if possible), they exchange their local maps. Three different communication models proposed in [6] proposes are Static Circle, Line of Sight and Propagation communication model. • Static Circle communication model: In this model, without considering the obstacles in the environment, any robot A can communicate with any other robot B

74

2015 Intl. Conference on Computing and Network Communications (CoCoNet'15), Dec. 16-19, 2015, Trivandrum, India

III.

PROPOSED APPROACH

A. Proposed Flowchart

To produce a map of an unexplored area, mobile robots coordinate between themselves and perform exploration based on trilateration and frontiers.WSN deployment is performed simultaneously while exploring the unknown area. The proposed system is an intelligent way of placing sensor nodes in which their approximate locations are known by the robots and in case of a disaster, rescue operations can be carried out accurately and efficiently. In this communication range of robots is not limited as sensor nodes are act as relays in between the robots. Thus an efficient exploration and deployment with minimum overlap and fast completion is achieved by the robots. In other words, in the proposed system placement of sensor nodes is not done randomly and so the locations of nodes are known while deploying and after. Thus, as soon as disaster information is received, rescue teams can be sent to the place near the known position of the alerting node and the information can also be verified by a robot if it is a disaster zone like a battlefield.

Start

Initialize the area to be covered as A and range of sensor nodes as a. Decide on the number of sensor nodes n=A/a

Perform localization procedure to find the initial location of all the robots employed to perform the mission of sensor deployment.

Is there interference in present location?

Yes

No Place the sensor node at that location and send the coordinate information, ID of sensor node to all other robots

The proposed system is a pre-deployment strategy in which the sensor network is deployed before a disaster occurs such as in a nuclear factory where the risk of disaster is very high. When compared to post-deployment strategies, it will save time and the complexity of the robots build to travel a disaster zone post-disaster will be much higher than in predisaster.

Check if X