5Deployment Parameters Diffraction: passing the signal through small opining and spreading it after passing the openingScattering: scatter the coming signalReflection : send the signal back to towards the sender
10Virtual Force Algorithm Sensors are initially deployed randomlyObjective:To maximize the CoverageAssume no prior knowledge about the monitored fieldAll nodes are mobileEnergy and obstacles might present in the field
11Virtual Force Algorithm (Cont.) Attractive and Repulsive forcesSensors do not physically moveA sequence of virtual motion paths is determined for the randomly placed sensors.Once the effective sensor positions are identified, a one-time movement is carried out to redeploy the sensors at these positions.
12Virtual Force Algorithm (Semi Distributed.) Assumptions:Clustered networkAll clustered heads are able to communicate with the sink nodeThe cluster head is responsible for executing the VFA and managing the one-time movement of sensors to the desired locations.
13Virtual Force Algorithm (Cont.) Each sensor behaves as a “Source of force” for all other sensors.This force can be either positive (Attractive) or negative (Repulsive).The closeness and wide distance between two sensors are measured using a predefined threshold.
14Virtual Force Algorithm (Cont.) Sensor Binary ModelConsider an n by m sensor field grid and assume that there are k sensors deployed in the random deployment stage.Each sensor has a detection range r. Assume sensor si is deployed at point (xi , yi ).For any point P at (x, y), we denote the Euclidean distance between si and P as d(si , P),The coverage of a Grid Point P can be expressed by:
15Virtual Force Algorithm (Cont.) Virtual ForcesAttraction force F12Repulsive force F13Zero Force F14Obstacle Force preferential coverageForce Total Force on node i =
16Virtual Force Algorithm (Cont.) Energy ConstraintsUsing such forces , the cluster head runs the VFAAfter stability occurs , Sensors are ordered to move to the new positionsEnergy and Obstacles might be problemsAny sensor will not be able to move the required distance , the moving order is discardedObstacles need an obstacle avoidance algorithm
17Think…..If some sensors are stationary, does this affect the virtual force algorithm?
19The problemA set of sensors S is deployed in a monitored field F(A)for a period of time T.The field is divided into a grid of cells A.Each cell is assigned a weight where represents the importance of the cell i.The location of each sensor is assumed known;More than one sensor could be deployed in one cell.Sensors are assumed heterogeneous in terms of their energy and mobility.
20Assumptions A sensor could be in different states; it could have its sensing off or on based on the field monitoring requirements.Sensing off, radio off – (sleep mode)Sensing off, radio receiving – (Receiving mode)Sensing off, radio transmitting – (Routing mode)Sensing on, radio receiving – (Sensing and Receiving mode)Sensing on, radio transmitting – (Sensing and Transmitting mode)Sensing on, radio off (Sensing mode)
21The main idea Knowing the energy map of the network : May lead to early detection to the uncovered areas.Redeploy new sensorsTurn off some of the sensors due to their coverage redundancyWake up some of the nodes when neededMove one or mobile nodes to cover the required uncovered spots
22Redeployment based Energy map Step 1: Energy dissipation rate predictionEach sensor predicts its own energy rate based on its history (e.g. Markov Chain ..)Step 2: sensors send their initial energy and the location, predicted energy dissipation rate to the sink node through a cluster head.Sensors update their energy dissipation rate based on a specific threshold (if the new dissipation rate increased more than the given threshold , the node sends the new dissipation rate)
23Redeployment based Energy map Step 3: the sink node constructs the energy map based on the received dissipated energy rate from the sensors.The sink may move one of the mobile sensors to the uncovered spot or wake up one of the sleeping sensors
24Think ……. What are the disadvantages of energy mapping algorithm ? - Sensor network is an event based network . Therefore , events are not frequently or based on specific pattern. Thus, the amount of messages to be transmitted to report the energy mapping will not be expected and might play a role in sensors energy dissipation.Centralized algorithm
26The problem of sensor deployment Given the target area, how to maximize the sensor coverage with less time, movement distance and message complexityThe importance of the problemDistributed instead of centralized
27Voronoi Diagram Definition: Every point in a given polygon is closer to the node in this polygon than to any other node.
28Overview of the proposed algorithm Sensors broadcast their locations and construct local Voronoi polygonsFind the coverage holes by examine Voronoi polygonsIf holes exist, reduce coverage hole by movingVOR : VORonoi-basedPull sensors to the sparsely covered area
29Part of Assignment 2Implement both Virtual Force algorithm and Voronoi based algorithm ? Report your experience and algorithms efficiency?Given a set of sensors with limited amount of energy. Some of these sensors are assumed mobile and others are assumed stationary. Assume similar sensing and communication ranges for all sensors. Sensors are allowed to move from one place to another iff they have enough energy to move to the required destination. In addition , the borders of the monitored area is assumed known in terms of 2D coordinates. Borders may be found in the monitored area. Advice a suitable deterministic deployment algorithm for efficient deployment to the sensors given that the deployed sensors have to be connected and important areas in the field are covered. In addition , your algorithm must guarantee the coverage of the monitored field for certain period of time.You may look for an already given solution or come up with a convincing one .
30Deterministic Deployment Deployment Using Circle Packing
31Deployment Using Circle Packing Deployment of homogenous sensorsFull Coverage DeploymentDeployment of connected heterogeneous sensors31
32Deployment of homogenous sensors Sensing rangeDensity1234567891416253632
33Full Coverage Deployment sSensor’s sensing range (r)11621731841952062172282392410251126122713281429153033
34Sequential Packing-based Deployment Algorithm (SPDA) GivenSensors Sensing RangesSensors Communication RangesBounded Monitored FieldObjectiveBest Connected Deployment SchemeMax. Coverage.Min. Overlapped AreasBenefit from the properties learned from the optimal deployment34
41Potential benefits of sink relocation Increased network longevity: shortened data paths can safe the total energy consumed to data collection and extend the life of relaying nodes.Improved timeliness: involves fewer relays leading to avoidance of large packet backlogsEnhanced safety: moves the sink away from harmful events without damaging network performance
42Energy-Based Relocation -- Motivation Normal Operational Mode:Sensors pursue multi-hop paths to communicate with the sink nodeIssues:When the sink is stationary, nearby sensors get involved in heavy packet forwarding and die quicklySink nodeInactive SensorActive SensorOne hop SensorDead SensorCan repositioning the sink node help?To where ?Nodes further away are picked as substitute relaysConsequence:Increase in total transmission power rapid energy depletionEffect grows spirally outward
43Moving the Sink Where to go Towards the region, whose sensors generate the most number of packetsCentroid of the last-hop nodes that route the largest traffic (use a distance * traffic metric)
44Think….What about putting the sink node initially in the center of all nodes? Does this will be the best position for the sink node?No , because sensor networks again are event based networks
45Part of your assignment Device an algorithm for Multiple Sink Network Design Problem in Large Scale Wireless Sensor Networks?You may look at :E. Ilker Oyman and Cem Ersoy, Multiple Sink Network Design Problem in Large Scale Wireless Sensor Networks,, IEEE International Conference on Communications, 2004
46Relay Node Placement in WSN Clustering Algorithms
47Clustering FactsClustering plays a dominant role in delaying the first node death, while aggregation plays a dominant role in delaying the last node deathIn each cluster one node acts as a cluster head which is in charge of coordinating with other cluster heads
48LEACH Algorithm LEACH is considered as clustering and routing protocol The LEACH Network is made up of nodes, some of which are called cluster-headsThe job of the cluster-head is to collect data from their surrounding nodes and pass it on to the base stationLEACH is dynamic because the job of cluster-head rotatesLEACH is considered as clustering and routing protocolLEACH (Low-Energy Adaptive Clustering Hierarchy)
49The Amount of Energy Depletion This is the formula for the amount of energy depletion by data transfer:
50LEACH’s Two PhasesThe LEACH network has two phases: the set-up phase and the steady-stateThe Set-Up PhaseWhere cluster-heads are chosenThe Steady-StateThe cluster-head is maintainedWhen data is transmitted between nodes
51Stochastic Threshold Algorithm Cluster-heads can be chosen stochastically (randomly based) on this algorithm:If n < T(n), then that node becomes a cluster-headThe algorithm is designed so that each node becomes a cluster-head at least once.R round number
52Deterministic Threshold Algorithm A modified version of this protocol is known as LEACH-C (or LEACH Centralized)This version has a deterministic threshold algorithm, which takes into account the amount of energy in the node…
53Think more …..How to modify LEACH to include more parameters such as node degree?
54HEED: Hybrid Energy Efficient Distributed Clustering HEED was designed to select different cluster heads in a field according to the amount of energy that is distributed in relation to a neighboring node.Four primary goals:prolonging network life-time by distributing energy consumptionterminating the clustering process within a constant number of iterations/stepsminimizing control overheadproducing well-distributed cluster heads and compact clusters.
55Heed AlgorithmEach node performs neighbor discovery, and broadcasts its cost to the detected neighbors.Each node sets its probability of becoming a cluster head, Chprob , as follows:Where, Cprob is the initial percentage of cluster heads among n nodes (it was set to 0.05),Eresidual and Emax are the residual and the maximum energy of a node (corresponding to the fully charged battery), respectively.The value of CHprob is not allowed to fall below the threshold pmin .
56Disadvantage (LEACH and HEED) – think…. Nodes’ score is computed based on node identifiers, and each node holds its message transmission until all its neighbors with lower IDs have done so.Each node stops its protocol execution if it knows that every node in its neighborhood has transmitted.It is assumed that the network topology does not change during the algorithm execution, and it is thus valid for each node to wait until it overhears every higher-scored neighbor transmitting.
58HEED AssignmentPrevious Algorithm is used with homogenous sensors (all have the same characteristics ).Device another clustering algorithm for heterogeneous WSN (nodes with different capabilities) .You may have a look at the following paperHarneet Kour and Ajay K. Sharma, “Hybrid Energy Efficient Distributed Protocol for Heterogeneous Wireless Sensor Network, ” International Journal of Computer Applications (0975 – 8887) Volume 4 – No.6, July 2010the score is computed based on node identifiers,and each node holds its message transmission until all itsneighbors with lower IDs have done so. Each node stopsits protocol execution if it knows that every node in itsneighborhood has transmitted. It is assumed that thenetwork topology does not change during the algorithmexecution, and it is thus valid for each node to wait untilit overhears every higher-scored neighbor transmitting.
60DECAEach node periodically transmits a Hello message to identify itself, and based on such Hello messages, each node maintains a neighbor list.Define for each node the score function as:Where E stands for the node residual energy, C stands for the node connectivity, I stands for the node identifier, and the weights followThe computed score is then used to compute the delay for this node to announce itself as the cluster head. The higher the score, the sooner the node will transmit.The computed delay is normalized between 0 and a certain upper bound Dmax
63MFLC for single and multimodal sensor networks A single feature sensor network is a network with each sensor node reports only one feature.Multimodal sensor network is a network with nodes report more than one feature.MFLC adapts LEACH clustering technique to support the multimodal sensor networks.MFLC differs from the LEACH on the criteria used for a node to decide to be a cluster head or not.
64MFLC single and multimodal sensor networks Score Equation :
65Data Similarity Clustering Based Fuzzy Logic (DSBF)
66DSBF Phase One: Computing Node Degrees Phase Two: Cluster Head ElectionPhase Three: Data Reporting
67Phase One: Computing Node Degrees The node degree based similarity feature is computedThe node degree in this context means the number of similar sensors around
71Flat Routing Each node plays the same role Data-centric routing Due to not feasible to assign a global id to each nodeSave energy through data negotiation and elimination of redundant dataProtocolsSensor Protocols for Information via Negotiation (SPIN)Directed diffusion (DD)Rumor routingMinimum Cost Forwarding Algorithm (MCFA)Gradient-based routing (GBR)Information-driven sensor querying/Constrained anisotropic diffusion routing (IDSQ/CADR)COUGARACQUIREEnergy-Aware RoutingRouting protocols with random walks
72Sensor protocols for information via negotiation (SPIN) FeaturesNegotiationto operate efficiently and to conserve energyusing a meta-dataResource adaptationTo extend the operating lifetime of the systemmonitoring their own energy resourcesSPIN MessageADV – new data advertisementREQ – request for ADV dataDATA – actual data messageADV, REQ messages contain only meta-data
73Sensor protocols for information via negotiation (SPIN) Operation processStep1ADVStep3DATAStep2REQStep4Step5Step6
74Sensor protocols for information via negotiation (SPIN) Resource adaptive algorithmWhen energy is plentifulCommunicate using the 3-stage handshake protocolWhen energy is approaching a low-energy thresholdIf a node receives ADV, it does not send out REQEnergy is reserved to sensing the eventAdvantageSimplicityEach node performs little decision making when it receives new dataNeed not forwarding tableRobust to topology changeDrawbackLarge overheadData broadcasting
75Think…. In SPIN What about mobile nodes? What about the multimodal Wireless nodes?
76Directed Diffusion (DD) FeatureData-centric routing protocolA path is established between sink node and source nodeLocalized interactionsThe propagation and aggregation procedures are all based on local informationFour elementsInterestA task description which is named by a list of attribute-value pairs that describe a taskGradientPath direction, data transmission rateData messageReinforcementTo select a single path from multiple paths
78Directed Diffusion (DD) AdvantageSmall delayAlways transmit the data through shortest pathRobust to failed pathDrawbackImbalance of node lifetimeThe energy of node on shortest path is drained faster than anotherTime synchronization techniqueTo implement data aggregationNot easy to realize in a sensor networkThe overhead involved in recording informationIncreasing the cost of a sensor node
79Think…. In DD What about mobile nodes? What about the multimodal Wireless nodes?
80Comparison between SPIN, LEACH & Directed Diffusion OptimalRouteNoYesNetworkLifetimeGoodVery goodResourceAwarenessUse of meta-data
81Rumor Routing Feature Combine query flooding and event flooding Discovering arbitrary paths instead of the shortest pathRumor routing is attractive only whenThe number of queries is larger than a thresholdThe number of events is smaller than another thresholdAssumptionThe network is composed of densely distributed nodesOnly short distance transmissionsImmobile nodes
82Rumor Routing Basic scheme Each node maintains A lists of neighbors An event tableWhen a node detects an eventGenerate an agentLet it travel on a random pathThe visited node form a gradient to the eventWhen a sink needs an eventTransmit a queryThe query meets some node which lies on the gradientRoute establishment
83Rumor RoutingThe node sensing an event probabilistically generates an agent.In order to propagate directions to the event as far as possible in the network, a straightening algorithm is usedThe agent maintains a list of recently seen nodes.When picking its next hop, it will first try nodes not in the list.
84Think…. In Rumor Routing What about mobile nodes?What about the multimodal Wireless nodes?
85Minimum Cost Forwarding Algorithm (MCFA) ObjectiveEstablish the cost fieldTransmit the data through the minimum-cost pathFeatureOptimalityMinimum cost path criteria : hop count, energy consumption, delay etc.SimplicityNeed not to maintain forwarding tableNeed not to know an ID for a neighbor node
86Minimum Cost Forwarding Algorithm (MCFA) Operation processEach node stores its cost to the sinkThe sink broadcasts an ADV messagecontaining its own cost (0 initially)Each node receiving the message transmits neighbor nodeAdd the cost in ADV message to its own costThe cost field is set upafter the ADV message propagates through the networkThe source transmits an information through cost fieldDrawbackLimited network sizeThe time to set the cost field is directly proportional to the size of the networkLoad is not balanced
87Think…. In Rumor Routing What about mobile nodes?What about the multimodal Wireless nodes?
88Geographic Adaptive Fidelity (GAF) Forms a virtual grid of the covered areaEach node associates itself with a point in the grid based on its locationNodes associated with same point in grid are considered equivalentSome nodes in an area are kept sleeping to conserve energyNodes change state from sleeping to active for load balancing
89Creating a Virtual Grid Use location information (GPS) to create a virtual gridAll nodes in a grid are equivalentOnly one node from a grid point is active at a timeAll nodes in a grid point is within the radio range of nodes in adjacent gridsVirtual grid results in hierarchical clusters of nodes
90Think once more …. What are the problems of GAF? What about mobile nodes?What about the multimodal Wireless nodes?Depends on location information in grid creation.Scheduling and node synchronization are required