Coverage Issues in Wireless Sensor Networks

Transcription

1 ModernComputerApplicationsTechnologies Course Coverage Issues in Wireless Sensor Networks Presenter:XiaofeiXing GuangzhouUniversity

2 Outline q Wirelsss Sensor Networks q Coverage Issues q Our Works ILP-Based Polytype Target Coverage Scheme in Heterogeneous Wireless Sensor Network 2

3 Faster, Smaller, Numerous q Moore s Law Stuff (transistors, etc) doubling every 1-2 years. q Bell s Law New computing class every 10 years. Streaming Data to/from the Physical World log(peoplepercomputer) year 3

4 What is Wireless Sensor Networks (WSN) q Wikipedia Definition: A WSN is a wireless network consisting of spatially distributed devices using sensors to cooperatively monitor physical or environmental conditions at different locations. q Types Temperature Sound Vibration Pressure Motion 4

5 Hardware of Sensor q Sensor: 5

6 Hardware of Sensor q The Berkeley TinyOS group 6

7 Hardware of Sensor q Sensor node: components 7

8 Characteristics A special wireless ad hoc network q Large number of nodes are deployed randomly and densely Scalability & Self-configure q Battery powered Energy efficiency q Topology and density charge Data centric q In-network data processing (Data aggregation) Message-level latency 8

9 Key Technology q Key technologies of WSN: Topology control Network protocol Network security Positioning technology Data aggregation q Energy is scarce For this reason, algorithms and protocols need to address the following issues: Lifetime maximization Robustness and fault tolerance Self-configuration 9

10 Applications q Area monitoring Air pollution monitoring Forest fires detection Greenhouse monitoring Landslide detection q Industrial monitoring Machine health monitoring q Water/wastewater monitoring Agriculture q Structural monitoring q Others: nearly anything you can imagine 10

11 Applications q SensorMap Project-Microsoft 11

12 Applications q VigilNet: An Integrated Sensor Network System for Energy- Efficient Surveillance - Univ. of Virginia 12

13 Applications q Perimeter Security System -Shanghai Pudong International Airport, Expo

14 Applications q Structural Health Monitoring of High-Rise Slender Structures 14

15 Outline q Wirelsss Sensor Networks q Coverage Issues q Our Works ILP-Based Polytype Target Coverage Scheme in Heterogeneous Wireless Sensor Network 15

16 Coverage Issues q Two fundamental tasks of Wireless Sensor Network: Sensing - sense changes in the temperature, pressure, acoustic signal, or electronic signal. Communications - transmit the sensed signal (or its digest) to the sink node. Coverage Connectivity q Definitions of Coverage and Connectivity Problems(CCP): Given a monitoring area A, a set of demand points D, a set of sensor nodes S and a sink node m, the CCP of WSN consists of assuring that at least n sensor nodes from S are covering each demand point j D in the monitoring area A, and there is a path between these nodes and the sink node m. 16

17 Sensor deployment q In terms of the deployment way of nodes, the subjects about coverage in WSN mainly include 3 classes, namely, area coverage, target coverage and barrier coverage. Area Coverage Target Coverage Barrier Coverage It addresses the problem how the whole sensor field is covered. It mainly deals with how to cover a set of discrete targets (some space points) with known locations. It focuses on finding a penetration path across the sensor field with some desired property. 17

18 WSN Formulation q Deployed densely and randomly Dense means exits redundant nodes Density control Random means topology is indefinite Topology control q Self-Configuration & Self-Organization Scalability Energy 18

19 Sensor s Operation q On-Duty (working) nodes Forming a sensor network Am I redundant? Off-duty? Energy Consideration Role-change? Off-duty? q Off-Duty (sleeping) nodes When to wakeup? On-duty? q Duty cycle policy Scheduling vs. Adaptive Duty period 19

20 Coverage, Connectivity q Is every point covered by 1 or K sensors 1-covered, K-covered q Is the sensor network connected q Others K-connected R

21 Coverage & Connectivity q Not independent, not identical (1) If region is continuous & Rt > 2Rs (2) Region is covered sensors are connected X. Wang (Sensys 03) H. Zhang & J. Hou (2004) Rs Rt 21

22 Real Products Product HMC1002 Magnetometer sensor Reflective type photoelectric sensor Thrubeam type photoelectric sensor Pyroelectric infrared sensor (RE814S) Acoustic sensor on Berkeley Motes Sensing Range 5m 1m 10m 30m TypicalApplication Detecting disturbance from automobiles Detecting targets of virtually any material Detecting targets of virtually any material Detecting moving objects ~1m Detecting acoustic sound sources Product MPR300 MPR400CB MPR410CB MPR420CB MPR500CA MPR510CA MPR520CA Transmission Range 30m 150m 300m 300m 150m 300m 300m 22

23 Simple Coverage Problem q Given an area and sensor deployment q Question: Is the entire area covered? R

24 Geometric Problems Art Gallery Problem: how to use a minimum set of guards in a polygon such that every point of the polygon is watched by at least one guard. Circle Covering Problem: Given a unit disk, find the smallest radius required for equal disks to completely cover the unit disk. Example: 24

25 K-Coverage Problem q Given: region, sensor deployment, integer k q Question: Is the entire region k-covered? C.-F. Huang & Y.-C Tseng R

26 Is the perimeter k-covered? 26

27 Is a belt region k-barrier covered? q Construct a graph G(V, E) V: sensor nodes, plus two dummy nodes L, R E: edge (u,v) if their sensing disks overlap q Region is k-barrier covered iff L and R are k- connected in G. L R 27

28 Density Control q Given: an area of interest and sensors deployed q Problem: turn on/off sensors to maximize the sensor network s life time. 28

29 Density Control (cont d) q Nodes are on-duty or off-duty by Scheduling or Probing Resulting monitoring area still covered q Sensing range Determined (disc) Irregular in shape, or even follow a probabilistic model 29

30 Approaches for Density Control q Adaptive PEAS (ICNP, ICDCS) CCP (SenSys) q Scheduling SET K-COVER (ICC) Co-Grid (IPSN) OGDC (International Workshop on Theoretical and Algorithmic Aspects of Sensor, Ad hoc Wireless and Peer-to- Peer Networks) 30

31 Surveillance (a)thevoronoidiagram andthemaximalbreachpath (b)thedelaunaytriangulationandthemaximalsupportpath S. Meguerdichian, F. Koushanfar, M. Potkonjak, and M. Srivastava. Worst and Best-Case Coverage in Sensor Networks. IEEE TRANSACTIONS ON MOBILE COMPUTING, 4(1):84-92,

32 Exposure The exposure for an object in the sensor field during the interval [ t, t ] along a path p ( t) t 1 ( ) t2 dp t E ( p ( t), t, t ) = I ( F, p ( t) ) dt 1 2 dt minimal exposure path the worst coverage of a sensor network S. Meguerdichian, F. Koushanfar, G. Qu, M. Potkonjak, Exposure In Wireless Ad Hoc Sensor Networks. Procs. of MobiCom '01, July

33 Problem Tree for Coverage and Connectivity Problems Barrier Density control Coverage Target Area # of sensors are needed? Adaptive Surveillance & exposure Scheduling PEAS OGDC Deployment Network # of sensors? K-coverage Per-node Xue&Kumar Network K-connectivity Topology control Probabilistic Homo Penrose Connectivity Topology information Deterministic Algorithmic Homo Scheduling Various connected subgraphs Adaptive Per-node (Max Rt) K-connected ASCENT LEACH 33

34 Outline q Wirelsss Sensor Networks q Coverage Issues q Our Works ILP-Based Polytype Target Coverage Scheme in Heterogeneous Wireless Sensor Network 34

35 Motivations q Heterogeneous WSN consists of sensor nodes with different characteristics. q Previous work There have been lots of efforts for target coverage problem in WSNs. Most of the previous studies concentrated on homogeneous wireless sensor networks. The sensor only has a single sensing unit. q In practice, a sensor may have multiple sensing units. MICA2 mote is equipped with several sensing units, which can sense the attributes of temperature, humidity, light, sound, vibration, etc. 35

36 Problem Definition q What is Polytype Target Coverage (PTC) problem? Given M targets and N sensors with multiple sensing units randomly deployed in the target s vicinity, turning on the sensor s sensing units to cover the targets in its sensing range, the PTC problem is to find a family of cover sets SC 1, SC 2, SC k, such that: (1) each target can be covered by each cover set, (2) k is maximized, and (3) each sensor in all sets consumes at most energy E. 36

37 Network model q HWSN consists of some sensors with multiple sensing units. It consists of: Base station; Clusterhead sensor nodes; Member sensor nodes. q Cluster-based structure [1] Managing and saving power by the mode scheduling of sensor; Efficient data aggregation; Many researchers have been addressing their achievements in cluster-based WSNs. q Uniform distribution. [1]Ameer Ahmed Abbasi, Mohamed Younis, A survey on clustering algorithms for wireless sensor networks. Computer communications archive, 30(14-15): , October

38 PTC problem with ILP constraint q Linear programming (LP) It is a mathematical method for determining a way to achieve the best outcome (such as maximum profit or lowest cost) in a given mathematical model for some list of requirements represented as linear equations. Standard form: Maximize: c T x Subject to: Ax b. where x represents the vector of variables (to be determined), c and b are vectors of (known) coefficients and A is a (known) matrix of coefficients. q Integer Linear Programming (ILP) If the unknown variables are all required to be integers, then the problem is called an ILP problem. 38

39 PTC problem with ILP constraint q PTC problem with ILP constraint can be formulated as below: N i= 1 ( ϕ δ θ ) SC γ, j, t, k. j t t t i i i, k k j Maximize: K k = 1 SC k K T k = 1 t = 1 t ( θ e ) E, i. t i, k SCk {0,1}, k. 39

40 PTC problem with ILP constraint q Notation description: N : the number of sensors; M : the number of targets; u t : the sensing units of sensor; t δ : denote whether the sensor S is equipped with i j i t the sensing unit u ; ϕ : denote whether the sensor S can cover t j k the target T ; t by the sensing unit u ; j γ : denote the target T needs to be covered SC : bool variable; SC =1 if is a cover sets; θ : bool variable; θ = 1 if the sensor S turns on t t i, k i, k i j k t t the sensing unit u in cover set SC ; u SC. i i k k 40

41 Energy-efficient Target Coverage Scheme q In order to solve polytype target coverage problem, an Energy-efficient Target Coverage (ETC) scheme is proposed, which is an optimal scheduling way in a cluster. q The executive procedure of ETC scheme: Step 1: In each cluster, member sensors send mcover message to its clusterhead, which contains the node s information, such as node's ID, residual energy and sensing capability, etc. Step 2: The clusterhead sorts the received message into a list according to the node's (1) residual energy, (2) sensing capability after a waiting time. Step 3: Then, the clusterhead traverses the list and sends a mnotice message to the member node when this node satisfies the requirement. Step 4: The node turns on its corresponding sensing units to cover the targets according to the received mnotice message. 41

42 Energy-efficient Target Coverage Scheme q Figure (a) is the network topology and (b) is the corresponding bipartite graph. Sensing unit type A. B. C. Cover mapping Target type Sensor Target 42

43 Energy-efficient Target Coverage Scheme Step 1: {1,(A,T1)} {1,(C,T5)} T 4 {2,(A,T1)} {2,(B,T4)} {2,(C,T6)} mcover message: {ID,(sensing unit, target),e i } A- B- C- T 5 T1 T 6 {4,(A,T1)} {4,(A,T2)} {4,(B,T3)} {4,(C,T5)} {4,(C,T7)} T 7 T 3 T2 {3,(A,T2)} {3,(B,T3)} {3,(C,T6)} 43

44 Energy-efficient Target Coverage Scheme Step 2: {1,(A,T1)} {1,(C,T5)} T 4 {2,(A,T1)} {2,(B,T4)} {2,(C,T6)} Sensing capability: {ID,(sensing unit, target),e i } A- B- C- T 5 T1 T 6 {4,(A,T1)} {4,(A,T2)} {4,(B,T3)} {4,(C,T5)} {4,(C,T7)} T 7 T 3 T2 {3,(A,T2)} {3,(B,T3)} {3,(C,T6)} Sort the message list. 44

45 Energy-efficient Target Coverage Scheme Step 3: T 4 A- B- C- T 5 T1 T 6 T 7 T 3 T2 {2,(B,T4)} {3,(A,T2)} {3,(B,T3)} {1,(A,T1)} {3,(C,T6)} {1,(C,T5)} {4,(C,T7)} 45

46 Energy-efficient Target Coverage Scheme q Five cover sets are formed: 46

47 Energy-efficient Target Coverage Scheme q We propose an algorithm, ETCA, to show this coverage scheme. Some notations and message types used in ETCA algorithm Notation t w Description the duration that clusterhead receives messages from its member nodes. It is less than the initial time in a round. E i the residual energy of node S i ; E thr β i CL the energy threshold of node. A node cannot be scheduled to detect targets when its residual energy is less than this energy threshold; the sensing capability of S i. it is the union of sensing capability of all sensing units equipped on S i can cover the targets in its sensing ranges; a cover list formed by clusterhead received mcover message from its member nodes; mcover a message that contains E i and of S i, etc; mnotice a message that can schedule the sensing units of a node to cover the corresponding targets. 47

48 Energy-efficient Target Coverage Scheme q Algorithm pesudo-code: 48

49 Performance Evaluation q We use the optimization toolbox MATLAB to evaluate the efficiency of ETCA through conducting some simulations and measuring network lifetime with different number of sensors, targets and sensing attributes. q The performance of ETCA is compared with ILP solution and Energy First (EF) [2] scheme, which is a greedy approach to make decisions for a sensor to enable its sensing units only considering its remaining energy. q Simulation parameters: [2]Vu, C.T., Gao, S.: Distributed Energy-Efficient Approach for K-Coverage in Wireless Sensor Network. In: Military Communications Conference, pp.1-7,

50 Performance Evaluation q Network lifetime VS number of sensors q Network lifetime VS number of targets 50

51 Performance Evaluation q Network lifetime VS number of sensing attributes q Execution Time VS number of Sensors 51

52 Conclusions q Proposed a provably optimal polynomial time algorithm for coverage in sensor networks. q Contributions: combine existing optimizing techniques and constructs such as the linear programming, with theoretical algorithmic techniques. Linear programming: get an optimal coverage in each cluster. q Heuristics for sensor deployment and asymptotic coverage behavior of random wireless sensor networks. 52

53 Useful Links q Paper Retrieve Library: IEL(IEEE/IET Electronic Library): ieeexplore.ieee.org ACM: portal.acm.org SpringerLink: Wiley: Elsevier Science: www. sciencedirect.com Google Scholar: scholar.google.com q WSN Projects: Research groups and projects: TinyOS: China Academic of Science: q Simulation Tools: NS2: isi.edu/nsnam/ns OMNeT++: 53

54 Questions? Contact Info. Xiaofei Xing 54