Скачать презентацию Next Century Challenges Scalable Coordination in Sensor Networks Скачать презентацию Next Century Challenges Scalable Coordination in Sensor Networks

71cc3ccaedf753fab48d1cb67a52ece5.ppt

  • Количество слайдов: 80

Next Century Challenges: Scalable Coordination in Sensor Networks Deborah Estrin, Ramesh Govindan, John Heidemann Next Century Challenges: Scalable Coordination in Sensor Networks Deborah Estrin, Ramesh Govindan, John Heidemann and Satish Kumar In Proceedings of the Fifth Annual International Conference on Mobile Computing and Networks (Mobi. COM '99), August 1999, Seattle, Washington. Directed Diffusion: A Scalable and Robust Communication Paradigm for Sensor Networks Chalermek Intanagonwiwat, Ramesh Govindan and Deborah Estrin In Proceedings of the Sixth Annual International Conference on Mobile Computing and Networking (Mobi. COM '00), August 2000, Boston, Massachusetts. Presenter: Malik Tubaishat Department of Computer Science University of Missouri - Rolla

Topics Information gathering and processing, coordinate amongst themselves to achieve a larger sensing task Topics Information gathering and processing, coordinate amongst themselves to achieve a larger sensing task Necessary for sensor network coordination Communication model for describing localized algorithms

Sensor Networks that are formed when a set of small un-tethered sensor devices are Sensor Networks that are formed when a set of small un-tethered sensor devices are deployed in an ad-hoc fashion cooperate on sensing a physical phenomenon. 3

Sensor Network Scenario: n Sensors are used to analyze the motion of a tornado Sensor Network Scenario: n Sensors are used to analyze the motion of a tornado n Sensors are deployed in a forest for fire detection n Sensors are attached to taxi cabs in a large metropolitan area to study the traffic conditions and plan routes effectively 4

Sensor Network Characteristics n Sensor n Observer n Phenomenon 5 Sensor Network Characteristics n Sensor n Observer n Phenomenon 5

Sensor The device that implements the physical sensing of environmental phenomena and reporting of Sensor The device that implements the physical sensing of environmental phenomena and reporting of measurements (through wireless communication). Typically, it consists of five components: 1. sensing hardware, 2. memory, battery, 4. embedded processor, 5. trans-receiver. 3. 6

Observer n The end user interested in obtaining information disseminated by the sensor network Observer n The end user interested in obtaining information disseminated by the sensor network about the phenomenon. n The observer may indicate interests (or queries) to the network and receive responses to these queries. 7

Phenomenon The entity of interest to the observer that is being sensed and optionally Phenomenon The entity of interest to the observer that is being sensed and optionally analyzed/filtered by the sensor network.

Network Dynamics Models n A sensor network forms a path between the phenomenon and Network Dynamics Models n A sensor network forms a path between the phenomenon and the observer n The goal of the sensor network protocol is to create and maintain this path (or multiple paths) under dynamic conditions while meeting the application requirements of low energy, low latency, high accuracy, and fault tolerance. 9

Approaches to Construct and Maintain the path between the observer and the phenomenon: n Approaches to Construct and Maintain the path between the observer and the phenomenon: n Static Sensor Networks n Mobile Sensor Networks 10

Static Sensor Networks v In static sensor networks, there is no motion among communicating Static Sensor Networks v In static sensor networks, there is no motion among communicating sensors, the observer and the phenomenon v An elected node relays a summary of the local observations to the observer Example: A group of sensors spread for temperature sensing Ø Such algorithms extend the lifetime of the sensor network because they trade-off local computation for communication. 11

Mobile Sensor Networks n In dynamic sensor networks, either the sensors themselves, the observer, Mobile Sensor Networks n In dynamic sensor networks, either the sensors themselves, the observer, or the phenomenon are mobile n Whenever any of the sensors associated with the current path from the observer to the phenomenon moves, the path may fail n Either the observer or the concerned sensor must take the initiative to rebuild a new path. 12

Building Paths n During initial setup, the observer can build multiple paths between itself Building Paths n During initial setup, the observer can build multiple paths between itself and the phenomenon and cache them, choosing the one that is the most beneficial at that time as the current path n If the path fails, another of the cached paths can be used n If all the cached paths are invalid then the observer must rebuild new paths. 13

Motion of the Components n Mobile Observer n Mobile Sensors n Mobile Phenomena 14 Motion of the Components n Mobile Observer n Mobile Sensors n Mobile Phenomena 14

Mobile Observer For example, a plane might fly over a field periodically to collect Mobile Observer For example, a plane might fly over a field periodically to collect information from a sensor network. Thus the observer, in the plane, is moving relative to the sensors and phenomena on the ground. 15

Mobile Sensors n For example, consider traffic monitoring implemented by attaching sensors to taxis Mobile Sensors n For example, consider traffic monitoring implemented by attaching sensors to taxis n As the taxis move, the attached sensors continuously communicate with each other about their own observations of the traffic conditions 16

Mobile Sensors (Cont. ) n The overhead of maintaining a globally unique sensor ID Mobile Sensors (Cont. ) n The overhead of maintaining a globally unique sensor ID in a hierarchical fashion like an IP address is expensive and not needed n Instead, these sensors should communicate only with their neighbors n In such networks, repairing a path can be used so that the information about the phenomenon is always available to the observer regardless of the mobility of the individual sensors. 17

Mobile Phenomena n A typical example of this paradigm is sensors deployed for animal Mobile Phenomena n A typical example of this paradigm is sensors deployed for animal detection n In this case the infrastructure level communication should be event-driven n Depending on the density of the phenomena, it will be inefficient if all the sensor nodes are active all the time n Only the sensors in the vicinity of the mobile phenomenon need to be active n The number of active sensors in the vicinity of the phenomenon can be determined by application specific goals such as accuracy, latency, and energy efficiency. 18

Concern n The primary concern here is the ability of the sensor network to Concern n The primary concern here is the ability of the sensor network to report the desired level of accuracy under latency constraints within an acceptable deployment cost n The accuracy is a function of the sensing technology of the sensors and their distance from the phenomenon. 19

Distributed Sensor Networks Next Century Challenges: Scalable Coordination in Sensor Networks Distributed Sensor Networks Next Century Challenges: Scalable Coordination in Sensor Networks

Topics Information gathering and processing, coordinate amongst themselves to achieve a larger sensing task Topics Information gathering and processing, coordinate amongst themselves to achieve a larger sensing task Necessary for sensor network coordination Communication model for describing localized algorithms

Scenario Several thousand sensors are rapidly deployed (e. g. , thrown from an aircraft) Scenario Several thousand sensors are rapidly deployed (e. g. , thrown from an aircraft) in remote terrain 1. The sensors communicate to establish a communication network 2. Divide the task of mapping and monitoring the terrain amongst themselves in an energy efficient manner 3. Adapt their overall sensing accuracy to the remaining total resources 4. Re-organize upon sensor failure. 22

Applications: Environmental Analysis 23 Applications: Environmental Analysis 23

Applications: Contaminant Flow Monitoring 24 Applications: Contaminant Flow Monitoring 24

Applications: Traffic Control n Sensor attached to every vehicle. n Capable of detecting their Applications: Traffic Control n Sensor attached to every vehicle. n Capable of detecting their location, vehicle sizes, speeds and densities; road conditions… n Alternate routes, estimate trip times… 25

Current Networks Ø § § Internet Each PC on the Internet has a user Current Networks Ø § § Internet Each PC on the Internet has a user who can resolve or at least report all manner of minor errors and problems Automated Factories May contain hundreds of largely unsupervised computers Deployed with very careful planning and react to very few external events 26

Differences with Current Networks Sensor Network Dynamic Changes § Sensors may be inaccessible § Differences with Current Networks Sensor Network Dynamic Changes § Sensors may be inaccessible § extremely difficult to pay special attention to any individual node § Ratio of communicating nodes to users is much greater § They operate and must respond to very dynamic environments § Deployed in a very ad hoc manner (possibly thrown down at random) § Must automatically adapt to changes in environment and requirements 27

Localized Algorithms Localized algorithms – where sensors only interact with other sensors in a Localized Algorithms Localized algorithms – where sensors only interact with other sensors in a restricted vicinity, but nevertheless collectively achieve a desired global objective. Objective: Electing sensors that form the longest-baseline for triangulation for locating external objects. 28

Data Centric unlike traditional networks, a sensor node may not need an identity (e. Data Centric unlike traditional networks, a sensor node may not need an identity (e. g. , an address) n What is the temperature at sensor #27? ( ) n Where are nodes whose temperatures recently exceeded 30 degrees? ( ) n Applications focus on the data generated by sensors n This approach decouples data from the sensor that produced it n This allows for more robust application design: even if sensor #27 dies, the data it generates can be cached in other (possibly neighboring) sensors for later retrieval. 29

Localized Clustering Algorithm Clustering: efficient coordination cluster Cluster-head 30 Localized Clustering Algorithm Clustering: efficient coordination cluster Cluster-head 30

Overview: n A link level procedure is run on each sensor that adjusts the Overview: n A link level procedure is run on each sensor that adjusts the transmission power and thus the communication range to a minimum value that maintains full network connectivity n The clustering algorithm then elects cluster-head sensors such that each sensor in the multi-hop network is associated with a cluster-head sensor as its parent n Cluster-heads could summarize the object located in their clusters to provide a less detailed view to distant nodes. 31

Assumption: Associate sensors at a particular level with a radius q The radius specifies Assumption: Associate sensors at a particular level with a radius q The radius specifies the number of physical hops that a sensor’s advertisement will travel q Sensors at a higher level are associated with larger radii that those at lower levels q All sensors start off at the lowest level of 0 q Each sensor then sends out periodic advertisements to sensors within radius hops q Advertisement = { hierarchical level, parent ID, remaining energy } q 32

Localized Clustering Algorithm Advertisement L 0 L 0 Wait time L 0 In order Localized Clustering Algorithm Advertisement L 0 L 0 Wait time L 0 In order to allow advertisements from ID, remaining energy hierarchical level, parent various sensors to reach other 33

Localized Clustering Algorithm Promotion Timer • Start promotion timer if no parent. • Promotion Localized Clustering Algorithm Promotion Timer • Start promotion timer if no parent. • Promotion timer: inversely proportional (remaining energy, number of other sensors from whom level 0 advertisements were received) • If promotion timer expires GO 34

Localized Clustering Algorithm Promotion Timer • Start promotion timer if no parent. • Promotion Localized Clustering Algorithm Promotion Timer • Start promotion timer if no parent. • Promotion timer: inversely proportional (remaining energy, number of other sensors from whom level 0 advertisements were received) • If promotion timer expires GO New advertisement L 0 L 0 L 1 Cluster-head 35

Next: n Once a level 0 sensor picks a closest potential parent, it cancels Next: n Once a level 0 sensor picks a closest potential parent, it cancels its promotion time if running and drops out of the election process n After promotion, the level 1 sensors start a wait time proportional to their new larger radius n At the end of the wait period, the level 1 sensor may demote it self if it does not have any child sensor or if its energy level is less than a certain threshold function of its children’s energy Any change in network conditions, or in sensor energy level results in re-clustering with bounded delay 36

Application of Clustering Algorithm n Aim: To pinpoint in an energy-efficient manner, the exact Application of Clustering Algorithm n Aim: To pinpoint in an energy-efficient manner, the exact location of objects. n Accuracy: widest possible measurement baseline. n Energy efficiency: fewest number of sensors participating in the triangulation. 37

Triangulation • Determine position in space. • Can specify approx direction of object relative Triangulation • Determine position in space. • Can specify approx direction of object relative to its own 38 location.

Algorithm of Locating Objects There exists a simple rule whereby each clusterhead sensor locally Algorithm of Locating Objects There exists a simple rule whereby each clusterhead sensor locally determine (based on information from neighbor cluster-heads alone) whether it should participate in the triangulation computation: 1 - If all the neighboring cluster-heads of a clusterhead sensor lie on the same side of a line drawn between the sensor and the object, then that cluster-head sensor elects itself as a participant in the computation 2 - Once elected, these sensors report their 39

Base-line Estimation 40 Base-line Estimation 40

Advantages of Cluster-based Approach n Sensor algorithms only use local information. – generally lower Advantages of Cluster-based Approach n Sensor algorithms only use local information. – generally lower energy consumption in comparison to global communication. n Robust to link or node failures and network partitions – mechanisms for self-configuration can be simpler. 41

Advantages of Cluster-based Approach n Local communication and per-hop data filtering – avoid transmitting Advantages of Cluster-based Approach n Local communication and per-hop data filtering – avoid transmitting large amounts of data over long distances. – preserving node energy resources. n Node energy resources are better utilized – cluster-heads adapt to changing energy levels. 42

Disadvantage of Cluster-based Approach n Non-optimal under certain terrain conditions. 43 Disadvantage of Cluster-based Approach n Non-optimal under certain terrain conditions. 43

Several Sensors Electing Themselves Obstacle Allow a cluster-head to switch on some number of Several Sensors Electing Themselves Obstacle Allow a cluster-head to switch on some number of child sensors in its cluster to do 44 object location.

Adaptive Fidelity Algorithms Z Y A Some cluster-head sensors turn themselves off to conserve Adaptive Fidelity Algorithms Z Y A Some cluster-head sensors turn themselves off to conserve power. quality of the answer can be traded against battery lifetime, network bandwidth, or number 45 of active sensors.

Directed Diffusion: A Scalable and Robust Communication Paradigm for Sensor Networks Source: http: //www. Directed Diffusion: A Scalable and Robust Communication Paradigm for Sensor Networks Source: http: //www. isi. edu/scadds

Topics Information gathering and processing, coordinate amongst themselves to achieve a larger sensing task Topics Information gathering and processing, coordinate amongst themselves to achieve a larger sensing task Necessary for sensor network coordination Communication model for describing localized algorithms

Directed Diffusion n A communication paradigm n A new data dissemination paradigm 48 Directed Diffusion n A communication paradigm n A new data dissemination paradigm 48

Properties n Directed diffusion is data-centric in that all communication is for named data Properties n Directed diffusion is data-centric in that all communication is for named data (not nodes id). n All nodes are application-aware This enables diffusion to achieve energy savings by selecting empirically good paths and by caching and processing data in-network Interactions are localized Data can be aggregated or processed within the network Network empirically adopts to best distribution path. 49

Sensor Network Examples: n How many pedestrians do you observe in the geographical region Sensor Network Examples: n How many pedestrians do you observe in the geographical region X? n Tell me in what direction that vehicle in region Y is moving. 50

Examples Cont’d n Sensors within the specified region start collecting information n Once individual Examples Cont’d n Sensors within the specified region start collecting information n Once individual nodes detect object movements, they might collaborate with neighboring nodes to disambiguate object location or movement direction n One of these nodes might then report the result back to the human operator. 51

Basic Scenario Ø A data consumer express interest in data with certain attributes Ø Basic Scenario Ø A data consumer express interest in data with certain attributes Ø The interest will be propagated towards producers using only local interactions Ø The interest propagation will leave traces or gradients in the network along the way Ø The producers will send data back to the consumers using the established gradients Ø Reinforcement will be used to reduce the number of data delivery path for efficient distribution 52

Implementation Human operator’s query Diffused into region X and Y Descriptions: sensor’s location, intensity Implementation Human operator’s query Diffused into region X and Y Descriptions: sensor’s location, intensity of the signal, degree of confidence in its estimation Activate sensors to collect information Transformed into interest Receives interest Query: “Every I ms for the next T seconds, send me a location estimate of any four-legged animal in subregion R” 53

Directed diffusion n Data is named using attribute-value pairs n A sensing task is Directed diffusion n Data is named using attribute-value pairs n A sensing task is disseminated throughout the sensor network as an interest for named data n The dissemination sets up gradients within the network designed to “draw” events (i. e. , data matching the interest) n Events start flowing towards the originators of interests along multiple paths n The sensor network reinforce one, or small number of these paths 54

A simplified schematic for directed diffusion interest gradient reinforcement 55 A simplified schematic for directed diffusion interest gradient reinforcement 55

Issues n Naming n Interests and Gradients n Data Propagation n Reinforcement 56 Issues n Naming n Interests and Gradients n Data Propagation n Reinforcement 56

Naming In directed diffusion, task descriptions are named by, for example, a list of Naming In directed diffusion, task descriptions are named by, for example, a list of attribute-value that describe a task. 57

The Animal Tracking Task type = four-legged animal //detect animal location interval = 20 The Animal Tracking Task type = four-legged animal //detect animal location interval = 20 ms // send back events every 20 ms duration = 10 seconds // … for the next 10 seconds Rect = [-100, 200, 400] // from sensors within rectangle The task description specifies an interest for data matching the attribute. 58

A Sensor that detects an animal might generate the following data: type = four-legged A Sensor that detects an animal might generate the following data: type = four-legged animal // type of animal seen instant = elephant // instant of this type location = [125, 220] // node location Intensity = 0. 6 // signal amplitude measure Confidence = 0. 85 // confidence in the match Timestamp = 01: 20: 40 // event generation time 59

Interests and Gradients Interests n An interest is usually injected into the network at Interests and Gradients Interests n An interest is usually injected into the network at some (possibly arbitrary) node in the network n We use the term sink to denote this node n The task state is purged from the node after the time indicated by the duration attribute 60

How Interests are Diffused Through the Sensor Network? n For each active task, the How Interests are Diffused Through the Sensor Network? n For each active task, the sink periodically broadcasts an interest message to each of its neighbors n Interest is periodically refreshed by the sink – The refresh rate is a protocol design parameter that trades off overhead for increased robustness to lost interests. 61

Interests - Cont’d n The initial interest takes the following form: type = four-legged Interests - Cont’d n The initial interest takes the following form: type = four-legged animal interval = 1 s rect = [-100, 200, 400] Timestamp = 01: 20: 40 // hh: mm: ss expires. At = 01: 30: 40 62

How interests are processed? Ø Every node maintains an interest cache Ø When a How interests are processed? Ø Every node maintains an interest cache Ø When a node receives an interest, it checks to see if the interest exists in the cache Ø If no matching entry exists, the node creates an interest entry Ø This entry has a single gradient towards the neighbor from which the interest was received, with the specified event data rate 63

How interests are processed? Ø When a gradient expires, it is removed from its How interests are processed? Ø When a gradient expires, it is removed from its interest entry Ø When all gradients for an interest entry have expired, the interest entry itself is removed from a cache Ø All received interests are re-sent. A node may suppress a received interest if it recently re-sent a matching interest. 64

Interests and Gradients Every pair of neighboring nodes establishes a gradient towards each other Interests and Gradients Every pair of neighboring nodes establishes a gradient towards each other 65

Gradients – Cont’d n When a node receives an interest from its neighbor, it Gradients – Cont’d n When a node receives an interest from its neighbor, it has no way of knowing whether that interest was in response to one it sent out earlier, or is an identical interest from another sink on the “other side” of that neighbor n Such two-way gradients can cause a node to receive one copy of low data rate events from each of its neighbors n This technique can enable fast recovery from field paths or reinforcement of empirically better paths n A gradient specifies both a data rate and a direction in which to send events. 66

Data Propagation n A sensor node that detects a target searches its interest cache Data Propagation n A sensor node that detects a target searches its interest cache for a matching interest entry n A matching entry is one whose rect encompasses the sensor location and the type of the entry matches the detected target type 67

Event Description The source sends to each neighbor for whom it has a gradient, Event Description The source sends to each neighbor for whom it has a gradient, an event description every second of the form: type = four-legged animal // type of animal seen instance = elephant // instance of this type location = [125, 220] // node location intensity = 0. 6 // single amplitude measure confidence = 0. 85 // confidence in the match timestamp = 01: 20: 40 // local time when event was generated 68

Data Propagation – Cont’d n A node that receives a data message from its Data Propagation – Cont’d n A node that receives a data message from its neighbor attempts to find a matching interest entry in its cache n If no matching exist, the data message is silently dropped n Otherwise, the received message is added to the data cache and the message is re-sent to the node’s neighbors 69

Data Propagation – Cont’d n By examining its data cache, a node can determine Data Propagation – Cont’d n By examining its data cache, a node can determine the data rate of received events n To re-send a received data message, a node needs to examine the matching interest entry’s gradient list n If all gradient have a data rate that is greater than or equal to the rate of incoming events, the node may simply send the received data message to the appropriate neighbors 70

Reinforcement n The sink initially diffuses an interest for a low event-rate notification (1 Reinforcement n The sink initially diffuses an interest for a low event-rate notification (1 event per second) n Once sources detect a matching target, they send lowrate events, possibly long multiple paths, towards the sink n After the sink starts receiving these low data rate events, it reinforces one particular neighbor in order to “draw down” higher quality (higher data rate) events. 71

Reinforcement – Example n Reinforce any neighbor from which a node receives previously unseen Reinforcement – Example n Reinforce any neighbor from which a node receives previously unseen event n To reinforce this neighbor, the sink re-sends the original interest message but with a smaller interval (higher data rate) 72

Example – cont’d Type = four-legged animal Interval = 10 ms Rect = [-100, Example – cont’d Type = four-legged animal Interval = 10 ms Rect = [-100, 200, 400] Timestamp = 01: 22: 35 expire. At = 01: 30: 40 Ø When the neighboring node receives this interest, it will reinforce at least one neighbor. 73

Reinforcement – cont’d The local rule we described, then, selects an empirically low delay Reinforcement – cont’d The local rule we described, then, selects an empirically low delay path Whenever one path delivers an event faster than others, the sink attempt to use this path to draw down high quality data 74

Basic Directed Diffusion Setting up gradients Source Sink Interest = Interrogation Gradient = Who Basic Directed Diffusion Setting up gradients Source Sink Interest = Interrogation Gradient = Who is interested 75 Source: http: //www. isi. edu/scadds

Basic Directed Diffusion Sending data and Reinforcing the best path Source Sink Low rate Basic Directed Diffusion Sending data and Reinforcing the best path Source Sink Low rate event Reinforcement = Increased interest 76 Source: http: //www. isi. edu/scadds

Directed Diffusion and Dynamics Source Sink Recovering from node failure Low rate event High Directed Diffusion and Dynamics Source Sink Recovering from node failure Low rate event High rate event Reinforcement 77 Source: http: //www. isi. edu/scadds

Directed Diffusion and Dynamics Source Sink Stable path Low rate event High rate event Directed Diffusion and Dynamics Source Sink Stable path Low rate event High rate event 78 Source: http: //www. isi. edu/scadds

Conclusion n We have described directed diffusion paradigm for designing distributed sensing algorithm n Conclusion n We have described directed diffusion paradigm for designing distributed sensing algorithm n Directed diffusion has the potential for significant energy efficiency n Directed diffusion are stable under the ranges of network dynamics n Directed diffusion has some novel features – Data-centric dissemination – Reinforcement-based adaptation to the empirically best path – In-network data aggregation and caching 79

Thanks 80 Thanks 80