energy-efficient sensing coverage and

0 downloads 0 Views 530KB Size Report
In this paper, we present our work on energy efficient sensing ... Key words Broadcasting protocol, sensing coverage, sensor networks, target detection.
Jrl Syst Sci & Complexity (2007) 20: 225–234

ENERGY-EFFICIENT SENSING COVERAGE AND COMMUNICATION FOR WIRELESS SENSOR NETWORKS∗ Xi CHEN · Qianchuan ZHAO · Xiaohong GUAN

Received: 2 February 2007 c 2007 Springer Science + Business Media, LLC Abstract Wireless sensor networks have a wide range of applications. Sensing coverage and communication coverage are two fundamental quality of service. In this paper, we present our work on energy efficient sensing coverage and communication. We design several schemes for sensing coverage subject to different requirements and constraints respectively. We also propose a broadcasting communication protocol with high energy efficiency and low latency for large scale sensor networks based on the Small World network theory. Simulation and experiment results show that our schemes and protocol have good performance. Key words Broadcasting protocol, sensing coverage, sensor networks, target detection.

1 Introduction Recent advances in microelectronic technology have enabled the development of sensor networks with low-cost, low-power, multi-functional sensor nodes that are small in size and can communicate with each other via wireless communication[1] . It is considered as one of the most important technologies in the 21st century[2] . Many programs and initiatives have been launched to investigate the key technology of sensor networks with applications. For example, the project “Smart Dust” at University of California, Berkeley aims to develop small, light and cheap sensors and to deploy the sensors on the ground, in the air, under water, in human bodies and vehicles, and inside buildings[3] . Sensor networks that consist of a large number of densely deployed sensors have typical applications including military target detection, security watching, environment monitoring, traffic surveillance, etc.[1,4] . When the area of interest is inaccessible or located in a hostile region, sensors may be dropped airborne, resulting in random placement[4] . For example, an aircraft can sprinkle a number of cheap wireless magnetic sensors along a road; once they hit the ground, these sensors automatically form a network and begin to scan the electric-magnetic environment Xi CHEN · Qianchuan ZHAO Center for Intelligent and Networked Systems, Department of Automation, Tsinghua University, Beijing 100084, China. Email: [email protected]; [email protected]. Xiaohong GUAN Center for Intelligent and Networked Systems, Department of Automation, Tsinghua University, Beijing 100084, China; SKLMS Lab and MOE KLINNS Lab, Xi’an Jiaotong University, Xi’an 710049, China. Email: [email protected]. ∗ The research is supported in part by National Natural Science Foundation (No. 60574087, 60574064) and the “111 International Collaboration Project” of China.

226

XI CHEN · QIANCHUAN ZHAO · XIAOHONG GUAN

signals. When a vehicle rolls by, its type is detectable from its magnetic signature, and its speed and moving direction can be obtained by fusing the measurements from multiple sensors. One of the most salient characteristics of the wireless sensor network (WSN) is that the sensor node is only equipped with limited power (< 0.5 Ah, 1.2v)[1] . For example, an off-theshelf sensor model, MICA2, has a full-duty lifetime of about one day with a battery capacity of 0.5Ah. Replenishment of power may be impossible. However, if appropriate approaches for power conservation and management are applied, the lifetime of a sensor network without changing battery can be greatly extended[1] . Therefore, fulfilling sensing tasks with low power consumption is one of the most important topics in sensor network research. Two fundamental measures of QoS of WSN are sensing coverage and communication coverage (network connectivity). As the sensor devices are supposed to be very cheap, an effective solution to prolong lifetime of a sensor network is to deploy more numbers of sensors than necessary but only let a part of them be active at any time. In this way some sensors are scheduled OFF (a power saving mode of sensors, with communication and sensing modules powering down) at a time, while the remaining sensors should provide required sensing coverage. Therefore, how to schedule the sensors to achieve longer system lifetime and meet the requirements on sensing coverage and network connectivity is extremely important and has attracted the interests of many researchers. Different methods and results have been reported as surveyed later. Our work on sensing coverage mainly includes five aspects. Firstly, we propose a novel location-free scheme for coverage preserving. A distributed, localized and location-free node scheduling algorithm, Stand Guard Algorithm, is proposed for meeting the requirements on coverage and connectivity in a unified scheme. Under appropriate conditions, this algorithm can guarantee network connectivity and any degree of sensing coverage. Secondly, we design a location-based percentage coverage configuration protocol. This protocol can guarantee the designed percentage coverage and save significant energy in comparison with the existing schemes. Thirdly, we develop a switching scheme without global time synchronization. Based on the scheme, a grid based coverage preserving protocol is proposed for the sensor nodes with different sensing ranges. Fourthly, We investigate the configuration problem of the information coverage in WSN. Since the concept of sensing area is not suitable for studying information coverage, we introduce occupation area for sensor and develop a simple algorithm for sensor to check whether it should be ON or OFF according to its occupation area. Then we design an information coverage configuration protocol. Fifthly, we explore the design and configuration of a sensor network for target detection. Subject to a budget limit for sensors, the network is required to secure a certain detection probability and to have a reasonable life time. We propose a novel model for randomly distributed network to solve this problem. Our work on network connectivity focuses on energy efficient broadcasting. We propose a simple randomized broadcasting protocol for large scale sensor networks. It is scalable and energy efficient. The key idea is to introduce a small portion of large radius nodes playing the similar role of “short-cuts” in small world networks. The rest of the paper is organized as follows. Section 2 presents literature reviews and related work. In Section 3, our work on sensing coverage is reported and a novel broadcasting communication protocol is discussed in Section 4. Concluding remarks are then given.

2 Related Work The existing algorithms for energy-efficient sensing coverage can be classified into two categories: location-dependent and location-free. In location-dependent coverage algorithms, each

COVERAGE AND COMMUNICATION

227

sensor needs to know the location information of its own and the neighbors. While in locationfree algorithms, node location information is not needed. There are several existing location-dependent algorithms for energy-efficient coverage in the literature. Tian and Georgana propose an algorithm which divides the lifetime of WSN into s rounds[5] . At the beginning of each round, every sensor checks whether its neighbors can help it to monitor its sensing area, if so, it is off duty eligible. The algorithms proposed by Carbunar et al.[6] , Carle et al.[7] and Zhang and Hou[8] also use the working-in-round scheme as in Tian and Georganas[5]. But they adopt different off duty eligibility rules. Yan et al.[9] propose a reference time based node scheduling algorithm, which can deal with heterogeneous sensors with different sensing ranges. Wang et al.[10] , and Zhang and Hou[8] consider the coverage as well as the connectivity of WSN simultaneously. They have proved that if the communication range is at least twice the sensing range, a complete coverage of a convex area implies network connectivity. The algorithms mentioned above are all location-dependent. Though many node localization methods have been proposed for WSN, the location error could be very large. For example, Langendoen and Reijers[11] study some existing localization methods and find that, for rangebased localization methods, “the average position error increases rapidly if noise is added to the range estimates”; when the range variance exceeds 10%, the average position error can be more than 50%! Due to the above drawbacks, the research on location-free algorithms for energy-efficient coverage is active. Ye et al. propose a location-free algorithm called PEAS, where PEAS stands for Probing Environment and Adaptive Sleeping[12,13] . In this algorithm, active sensors keep working until their energy are used up, and off sensors turn active once its sleeping time (following an exponential distribution) expires. Once an off sensor becomes active, it checks whether there is at least one active sensor within its probing range. If so, it is off again; otherwise, it stays active until running out of energy. In [14], Tian and Georganas propose three location-free algorithms, i.e., nearest-neighborbased, neighbor-number-based and probability-based schemes. In nearest-neighbor-based algorithm, if a sensor finds its nearest active neighbor is within a distance D, it will broadcast an off-message to request other sensors to turn off and then is off itself. The neighbor-numberbased algorithm has a little variable strategy such that if a sensor finds the number of its active neighbors is more than a threshold K, it will broadcast a turn-off message and then is off itself[14] . The probability-based algorithm is a random-off mechanism. A sensor will generate a random number x between [0, 1). If x is less than a threshold p, it is off; otherwise, it will remain active. Coverage quality can be measured in various ways. Since coverage is usually defined based on sensing radius, a point is covered if it is within the sensing radius of at least one sensor node. Such a definition is considered as physical coverage. The research on the coverage configuration mechanisms focuses on physical coverage problem. However, physical coverage hardly takes advantage of sensor data fusion that can help the WSN to improve its performance. Therefore, in [15], Wang et al. proposed the concept of information coverage via estimation theory. Different from physical coverage, information coverage focuses on the case when sensor nodes, in order to make more accurate estimation, fuse their measurements on a signal source at a particular position. The energy efficiency of communication protocols is another research focus. Many studies have been done including broadcasting and routed communications. Broadcasting plays very basic role in wireless communications. In fact, broadcasting is commonly used to gather neighborhood structure information before other more complicated point to point communication is established. For broadcasting, one of the most commonly used mechanisms, although the

XI CHEN · QIANCHUAN ZHAO · XIAOHONG GUAN

228

assumption of uniform radius at all nodes is simple and popular[16] , it is not clear whether these protocols are energy efficient. Although clustering-based protocol such as LEACH (Low-Energy Adaptive Clustering Hierarchy)[17] and its varieties exist and have demonstrated to be energy efficient, it is still interesting to seek simpler protocol without too much overhead to maintain the clusters. The study in [18] for small scale networks suggests that nodes with differentiated radius can perform much better than uniform radius. The challenge then becomes how to design scalable protocols which are energy efficient and simple to implement while avoiding complicated computation for global optimization. Recent network science discovery of “smallworld” feature on efficient organization of large scale complex networks[19] brings us inspiration to address this challenge in the context of sensor networks.

3 Energy-Efficient Sensing Coverage 3.1 Location-Free Scheduling Methods for Maintaining Sensing Coverage Due to the difficulties in localizing the sensor nodes, it is desirable to have location-free methods for maintaining sensing coverage since the sensor nodes can save energy for obtaining location information. It is shown how to maintain 1-dgree coverage without considering the connectivity in [12,13] and [14]. However, in many applications, such as target localization and tracking, more than 3-dgree coverage is usually needed. Moreover, fusing the data sent to the data sink requires the network be connected. We proposed a novel node scheduling algorithm in [20], StanGA, which can provide any degree of coverage as well as connectivity. Furthermore, we proved the conditions for StanGA to guarantee any degree of coverage and network connectivity. We also discussed how to choose system parameters to obtain a desired performance of the network. By comparing StanGA with other existing location-free algorithms, it can deal with any degree of coverage and connectivity simultaneously. Simulation results show that StanGA outperforms other algorithms, as shown in the Figure 1 below. For a desired coverage ratio, StanGA cause more sensors to turn off. With StanGA, the energy balance among the network is very good and therefore more energy-efficient. 1

0.04

0.99

0.035 0.03 STD of Coverage Ratio

Coverage Ratio

0.98 0.97 0.96 0.95 StanGA PEAS Nearest Neighbor Based Neighbor Number Based Probability Based

0.94 0.93 0.92

StanGA PEAS Nearest Neighbor Based Neighbor Number Based Probability Based

0

0.1

0.2 0.3 0.4 0.5 Off Sensor Percentage

0.025 0.02 0.015 0.01 0.005

0.6

0.7

0.8

0

0

0.1

0.2 0.3 0.4 0.5 Off Sensor Percentage

0.6

0.7

0.8

Figure 1 Location-free algorithm for maintaining sensing-coverage

3.2 Location-Dependent Percentage Coverage Most energy-efficient coverage problems investigated so far are for preserving complete coverage. Complete sensing coverage, which means sensor network can sense the whole area of interest without any vacancy, is energy consuming and demanding on sensors’ locations. Especially when sensors are deployed in a random way, complete coverage is harder to be obtained.

COVERAGE AND COMMUNICATION

229

Partial coverage, which does not require the network to fully cover the whole area of interest, aims to strike the balance between energy consumption and the performance of the network. With a weaker demand on coverage, sensors can save more energy and at last the network gains a longer life. Therefore, we proposed a method for preserving a percentage of sensing coverage other than preserving complete coverage[21] and a location-dependent Percentage Coverage Configuration Protocol (PCCP) is developed. In PCCP, the network lifetime is divided into a sequence of working stages. Each stage begins with a node scheduling phase followed by a sensing phase. ACTIVE-duty sensors enter the sensing phase and the OFF-duty sensors are off. Then each sensor stays in its state until the next round starts. PCCP can assure the proportion of the sensing area with a desired percentage coverage. Simulation results show that PCCP can not only guarantee the desired coverage percentage but also have more sensors turned off in comparison with other existing coverage preserving schemes. Therefore, PCCP can extend the lifetime of the sensor network. More details about PCCP can be found in our previous work[21] . 3.3 Coverage with Various Sensing Ranges Without Time Synchronization Most of the existing methods for sensing coverage require global time synchronization or uniform sensing ranges of all nodes. However, global time synchronization in a large scale sensor network is usually very costly and hard to achieve. Besides, the network may consist of sensor nodes with various capabilities and sensing ranges due to different energy levels or properties. Therefore, we proposed a grid based coverage preserving protocol (GBCPP), based on a novel scheme for state switching and a redundancy rule[22] when the sensors have various sensing ranges. The switching scheme allows every sensor to switch its state locally without global time synchronization. Only when there are a certain number of a sensor’s neighbors become active, it will check its off-duty eligibility according to its active neighbor’s locations and the grid based redundancy rule. With GBCPP, any new sensor can be added into or removed from the WSN at any time, and is thus robust to structural changes. For performance comparison, we choose two algorithms that can deal with heterogeneous WSN and need global time synchronization. One is proposed by Tian and Georganas in [5]. The core idea is: for Sensor A, if a sector of A’s sensing area is covered by one of A’s neighbors, record the central angle of the sector; if the union of the central angles reaches 360◦ , Sensor A is eligible for off-duty. We call it Sponsored Sector Based Algorithm. Yan et al. proposed another method in [9], where the schedule of each node is created by the reference time of each sensor. We call it Reference Time Based Algorithm. Simulation results show that GBCPP consumes much less energy, and can balance the energy consumption among sensor nodes as well as the two methods, as shown in Figure 2. 3.4 Information Coverage Configuration Based on the concept of “information coverage” in [15], we investigated the information coverage configuration problem. Firstly, we propose the concept of occupation area (OA) to divide the responsibility of different sensor nodes and an ON-duty eligibility check is performed. If a sensor node finds information gathered in its OA has satisfactory accuracy, that it is covered by other active nodes, it is off. The ON-duty eligibility check is actually a nonlinear programming problem but it is not practical to solve the problem with the limited computational resources of a sensor network. To address this issue, we develop a distance summation based method to fulfill the task of checking ON-duty eligibility. With OA and distance summation based algorithm, an information coverage configuration protocol (ICCP)[23] is proposed to configure the WSN. For performance comparison, we implement a physical coverage configuration protocol CCP

XI CHEN · QIANCHUAN ZHAO · XIAOHONG GUAN

230

40

0.45

Average Number of Working Nodes

STD of Energy Consumption in Percentage

0.4 0.35 0.3 0.25 Sponsored Sector Based Reference Time Based GBCPP (k=2) GBCPP (k=1)

0.2 0.15 0.1

35

30 Sponsored Sector Based Reference Time Based GBCPP (k=2) GBCPP (k=1)

25

20 0.05 0

100

200

300 400 Number of Deployed Sensors N

500

15 100

200

300 400 Number of Deployed Sensors N

500

Figure 2 Comparison of GBCPP with two other algorithms

used in [10] with the same parameters as a reference. Figure 3 shows that ICCP not only saves more energy but also balances the energy consumption more uniformly among the nodes in the WSN. 3.5 Node Scheduling Scheme and Target Detection We explore the design and configuration of a sensor network for target detection[24] . The problem considered is that, with a given budget for sensors, how to design a sensor network to monitor an area of interest for target detection with certain probability, e.g., 95%, and make the network work for a reasonable long time, e.g., ten days or longer. We present a novel model of randomly distributed sensor network (RDSN) to solve this problem. Suppose all sensors making up the RDSN are identical (in terms of sensing radius, energy as well as any other function and capability) and are independently and identically distributed in the area of interest following uniform distribution. To save energy, we design a state switching scheme (SSS). With SSS, each sensor independently switches its state, ON and OFF, with probabilities a and 1 − a respectively and keeps its state at least for UT before taking chance to change its state. SSS is a combination of randomized activation and duty-cycle. It has the advantages of both randomness (from parameter a) and regularity (from parameter UT ). SSS does not require sensors to coordinate their working schedule via communication so that its energy-efficiency performance is good. The randomness in state switching makes it hard to predict sensor’s state and enables the RDSN to operate robustly. SSS fundamentally controls the behavior of sensors and the RDSN at large. It lets the RDSN have a good scalability as the number of sensors increases. SSS is also easy to implement and its overheads are very small. With the above RDSN model and SSS scheme the relationships between detection probability and the parameters of sensors, network and target are obtained. In addition, we analyze the behavior of the RDSN in the course of the target moving through the area of interest. We formulate a two-level optimization problem. At the lower level, by choosing the value for (a, UT ), the behavior of each sensor is optimized subject to its energy constraint and then at the upper level, the lifetime of the network is examined. Equilibrium solution of the two-level optimization problem is obtained by numerical method. This solution delivers the optimal (a, UT ) and the maximal lifetime of the RDSN which not only satisfies the budget limit for sensors but also guarantees the predefined detection probability. Such analysis and results uncover important features and provide deep insights and useful information for the design of randomly distributed sensor networks for target tracking.

COVERAGE AND COMMUNICATION

231

4 Energy-Efficient Broadcasting Communication Protocol How to efficiently realize communication in large scale wireless sensor networks is challenging for at least three reasons. 1) In wireless sensor networks, energy (battery) of nodes is very limited. Computation, communication and sensing share the consumption of power. The life time of the entire network depends on balanced and efficient use of power. 2) The acquired data needs to be transmitted to the base station or data sink with small latency. However the communication radius is usually limited due to the capacity of nodes. Therefore, the data transmission has to be relayed through multi-hops before arriving at the base station. The latency for data transmission is proportional to the number of hops. 3) A central controller is not feasible for large networks due to the complexity. The organization of the network should be decentralized with scalability. Clearly, the requirements in 1) and 2) are contradictory. If we use large communication radius uniformly for all nodes, although there will be smaller number of hops to arrive at the base station, the energy consumption will be high since the energy consumption for each hop is E = Q(rq ), where q ∈ [2, 4]. If we use small communication radius uniformly for all nodes, the energy consumption at each hop may be small but the communication latency could be large due to many reply hops. Although theoretically a global optimization method can be applied to decide the best communication radius for each node, it is practically infeasible to do so for large sensor networks where nodes are deployed randomly in an area due to heavy computational requirements. We need to solve a multivariable nonlinear optimization problem with number of variables proportional to the number of nodes. To resolve this problem, we propose a practical broadcast protocol for large scale sensor networks deployed randomly in an area motivated by the recent progress in network science. Suppose N nodes are randomly deployed in an area of interest (AOA), a message of k-bit from a random source node is required to spread over the network. The short radius rs and the long radius rl randomly assigned to each retransmission node as the one-hop size with probability 1 − p and p, respectively. According to the distance-based relay scheme, the signal strength of received message is used to estimate the relative distance between neighboring nodes, and only those receivers far away from the transmitters are required to retransmit. During a fixed time slot, each recipient node may receive several copies of the same message. If all the relative distances measured by the copies’ signal strengths are above a threshold th, the node should retransmit; otherwise, no need to retransmit. Here we define a constant θ as the ratio of the threshold distance th to the transmission range r. In our two-radius broadcastings mechanism, the threshold distance th for small-radius and large-radius nodes is θrs and θrl , respectively. The protocol is illustrated in the Figure 4. There are a few parameters to choose in the protocol and we focus on the selection of the probability p of a node being chosen to broadcast with large radius, where its value is about 0.1, and small world phenomenon appears in Watts and Strogaze Small World model. Choice of other parameters can be found in [25]. We are inspired by the concept of “shortcut” from relational networks to random geometry networks. The large radius nodes play the role of shortcuts in our sensor network communication. The fraction p of such shortcuts needs not to be high to reduce the entire network’s average distance considerably.

Mean Energy Consumption in Percentage

0.22 0.2 0.18 CCP ICCP

0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 100

150

200

250

300

350

Total Number of Sensors

400

450

500

STD of Energy Consumption in Percentage

XI CHEN · QIANCHUAN ZHAO · XIAOHONG GUAN

232

0.22 0.2 0.18 CCP ICCP

0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 100

150

200

250

300

350

400

Total Number of Sensors

N

450

500

N

Figure 3 Performance Comparison of ICCP with CCP

Small Radius Large Radius

Figure 4 Illustration of the two-radius broadcasting protocol

The choice of p has significant impact on the system performance. We focus on the two most important performance indices, k(p) and L(p), where k(p) is the average energy consumption of each node and L(p) is the average number of relay hops between two node pair in the network. Under simplified assumption that there is no occurrence of repeated transmission and each node relay exactly once, we can predict the order of k(p) as a linear combination with the first order approximation k(p) ∝ (1 − p)rsq + prlq . Based on the law of large numbers, we can estimate the average number of hops between two nodes L(p) as 1 . L(p) ∝ (1 − p)rs + prl 

If we choose the short radius rs ∝ and the long radius

 rl ∝ N

rs N →∞ rl

then we find lim

logN N logN , N

which implies that with approximately only p ≈

1 10

additional energy

consumption, we achieve considerable reduction in time delay L(p). One thing worthwhile to

COVERAGE AND COMMUNICATION

233

emphasize is that the above analysis is only an approximation. An important observation not reflected in the above analysis but captured by our numerical test is that the protocol can even do better than the use of “medium” size radius (such as rm = (1 − p)rs + prl ) in both performance indices. More details can be found in [25]. In summary, we propose a simple randomized broadcasting protocol for large scale sensor networks. It is scalable and achieves balance between power consumption and transmission delay without solving global optimization problem. It has similar advantage to the clusterbased protocols in terms of efficiency yet needs no pre-communications, voting cluster head and synchronization overhead for head shifting. The method is a promising application of Small World effects from complex network theory to the design of large scale engineering networks.

5 Concluding Remarks Sensor networks have a wide spectrum of applications. Different applications may have different requirements on sensing coverage, communication and lifetime of sensor networks. How to schedule the sensors to achieve longer system lifetime and meet the requirements on sensing coverage and network connectivity is extremely important and are the key issues in sensor network design and operation. Systems and optimization theories are applicable to find the optimal tradeoff between quality of service and energy consumption. The Small World effects based on complex network theory are applicable to develop a scalable scheme to achieve balance between power consumption and transmission latency without solving global optimization problem. References [1] I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, A survey on sensor networks, IEEE Communications Magazine, 2002: 102–114. [2] Editors of Technical Review, Ten emerging technologies that will change the world, Technology Review, 2003, 106(1): 22–49. [3] J. M. Kahn, R. H. Katz, and K. S. J. Pister, Next century challenges: Mobile networking for “Smart Dust”, in Proceedings of ACM MobiCom 1999, Seattle, WA, 1999, 271–278. [4] C. Y. Chong and S. P. Kumar, Sensor networks: evolution, opportunities, and challenges, in Proceedings of the IEEE, 2003, 91(8): 1247–1256. [5] D. Tian and N. D. Georganas, A coverage-preserving node scheduling scheme for large wireless sensor network, in Proceeding of 1st ACM Workshop on Wireless Sensor Networks and Applications (WSNA 02), Atlanta, Geogia, 2002. [6] B. Carbunar, A. Grama, J. Vitek, and O. Carbunar, Coverage preserving redundancy elimination in sensor networks, in Proceeding of 1st IEEE International Conference on Sensor and Ad Hoc Communications and Networks (SECON 2004), Santa Clara, October 2004. [7] J. Carle, A. Gallais, and D. Simplot-Ryl, Preserving area coverage in wireless sensor networks by using surface coverage relay dominating sets, in Proceeding of 10th IEEE Symposium on Computers and Communications (ISCC 2005), Cartagena, Spain, 2005. [8] H. Zhang and J. C. Hou, Maintaining sensing coverage and connectivity in large sensor networks, Wireless Ad Hoc and Sensor Networks: An International Journal, 2005, 1(1–2): 89–123. [9] T. Yan, T. He, and J. A. Stankovic, Differentiated surveillance for sensor networks, in Proceedings of ACM International Conference on Embedded Networked Sensor Systems (SenSys 2003), Los Angeles, CA, USA, November 2003, 51–62. [10] X. Wang, G. Xing, Y. Zhang, C. Lu, R. Pless, and C. Gill, Integrated coverage and connectivity configuration in wireless sensor networks, in Proceedings of ACM International Conference on

234

[11] [12] [13]

[14] [15] [16]

[17]

[18] [19] [20]

[21]

[22]

[23]

[24] [25]

XI CHEN · QIANCHUAN ZHAO · XIAOHONG GUAN Embedded Networked Sensor Systems (SenSys 2003), Los Angeles, CA, USA, November 2003, 28– 39. K. Langendoen and N. Reijers, Distributed localization in wireless sensor networks: a quantitative comparison, Computer Networks, 2003, 43: 499–518. F. Ye, G. Zhong, S. Lu, and L. Zhang, Energy efficient robust sensing coverage in large sensor networks, UCLA Technical Report, 2002. URL: http://lion.cs.uiuc.edu/coverage-tech-report.ps. F. Ye, G. Zhong, S. Lu, and L. Zhang, PEAS: A robust energy conserving protocol for long-lived sensor networks, in Proceedings of International Conference on Network Protocols (ICNP) 2002, Paris, France, November 2002, 200–201. D. Tian and N. D. Georganas, Location and calculation-free node-scheduling schemes in large wireless sensor networks, Ad Hoc Networks, 2004, 2: 65–85. B. Wang, W. Wang, V. Srinivasan, and K. C. Chua, Information coverage for wireless sensor networks, IEEE Communication Letters, 2005, 9(11): 967–969. P. Gupta and P. R. Kumar, Critical power for asymptotic connectivity in wireless networks, in Stochastic Analysis, Control, Optimization and Applications: A Volume in Honor of W.H. Fleming (ed. by W.M. McEneany, G. Yin, and Q. Zhang), Birkhauser, Boston, MA, 1998, 547. W. R. Heinzelman, A. Chandrakasan, and H. Balakrishnan, Energy-efficient communication protocol for wireless sensor networks, in Proceedings of the 33rd Annual Hawaii International Conference on System Sciences, January 2000, 1–10. W. Krause, I. Glauche, R. Sollacher, et al., Impact of network structure on the capacity of wireless multihop ad hoc communication, Physica A, 2004, 338–633. D. Watts and S. Strogatz, Collective dynamics of ‘small-world’ networks, Nature, 1998, 393: 440– 442. H. Bai, X. Chen, B. Li, and D. Han, A location-free algorithm of energy-efficient connected coverage for high density wireless sensor networks, Journal of Discrete Event Dynamic Systems: Theory and Applications, to appear. H. Bai, X. Chen, Y. C. Ho, and X. Guan, Percentage coverage configuration in wireless sensor networks, in Proceeding of International Symposium on Parallel and Distributed Processing and Applications (ISPA’2005, Lecture Notes in Computer Science 3758), November 2005, 780–791. H. Bai, X. Chen, and X. Guan, Preserving coverage for wireless sensor networks of nodes with various sensing ranges, in Proceeding of 2006 IEEE International Conference on Networking, Sensing and Control (ICNSC’2006), Ft. Lauderdale, Florida, USA, April 23-25, 2006, 54–59. H. Bai, X. Chen, Y. C. Ho, and X. Guan, Information coverage configuration with energy preservation in large scale wireless sensor networks, in Proceeding of 2006 IEEE International Conference on Computer and Information Technology (CIT’2006), Seoul, Korea, September 20–22, 2006. X. Chen, H. Bai, L. Xia, and Y. C. Ho, Design of a randomly distributed sensor network for target detection, Automatica, to appear. F. Liu and Q. Zhao, An efficient organization mechanism for spatial networks, Physica A, 2006, 366: 608–618.