2004/2/10 2004/2/10jenchi minimum-energy asynchronous dissemination to mobile sinks in wireless...

2004/2/102004/2/10JenchiJenchi

Minimum-Energy Minimum-Energy Asynchronous Dissemination Asynchronous Dissemination to Mobile Sinks in Wireless to Mobile Sinks in Wireless

Sensor NetworksSensor Networks

ACM Conference on Embedded Networked SensorACM Conference on Embedded Networked SensorSystems (SenSys 2003), Los Angeles, USA, Nov 2003 Systems (SenSys 2003), Los Angeles, USA, Nov 2003

Hyung Seok Kim, Tarek F. Abdelzaher, and Wook Hyun Kwon


OutlinesOutlines

IntroductionIntroduction Assumptions and Basic ModelAssumptions and Basic Model SEADSEAD Related WorksRelated Works EvaluationEvaluation ConclusionConclusion


IntroductionIntroduction

Sinks Sinks – gathering the sensor readingsgathering the sensor readings– may be mobile PDAs carried by users or may may be mobile PDAs carried by users or may

be static access pointsbe static access points– have different service requirementshave different service requirements

desired data refresh ratedesired data refresh rate end-to-end delay between the source and the siend-to-end delay between the source and the si

nknk


Introduction (cont.)Introduction (cont.)

Energy is identified as the most crucial Energy is identified as the most crucial resource in sensor networksresource in sensor networks– The energy consumption of each sensor node The energy consumption of each sensor node

the cost of transmitting and receiving messagesthe cost of transmitting and receiving messages

When sinks are mobile in sensor networksWhen sinks are mobile in sensor networks– Communication consists of three main partsCommunication consists of three main parts

Building the dissemination tree (d-tree)Building the dissemination tree (d-tree) Disseminating dataDisseminating data Maintaining linkage to mobile sinksMaintaining linkage to mobile sinks


Introduction (cont.)Introduction (cont.)

A Scalable Energy-efficient Asynchronous A Scalable Energy-efficient Asynchronous Dissemination protocol (SEAD)Dissemination protocol (SEAD)– A distributed self-organizing protocol that saves A distributed self-organizing protocol that saves

communication energycommunication energy– Sinks are mobileSinks are mobile– Does not use mobile sinks as intermediate Does not use mobile sinks as intermediate

members of the treemembers of the tree This precludes frequent changes of the dissemination This precludes frequent changes of the dissemination

path due to sink mobilitypath due to sink mobility

– A stationary sensor node takes the mobile sink’s A stationary sensor node takes the mobile sink’s place for building an optimal dissemination treeplace for building an optimal dissemination tree


Assumptions and Basic Assumptions and Basic ModelModel

Basic AssumptionsBasic Assumptions– Each sensor node is aware of its own Each sensor node is aware of its own

geographic locationgeographic location Not require GPS at every nodeNot require GPS at every node

– After having been deployed, sensor nodes After having been deployed, sensor nodes remain stationary at their initial locationsremain stationary at their initial locations

– The sensor nodes are homogeneousThe sensor nodes are homogeneous– Sensor nodes communicate with sinks by Sensor nodes communicate with sinks by

delivering data across multiple hopsdelivering data across multiple hops


Assumptions and Basic Model Assumptions and Basic Model (cont.)(cont.)

Overview of the algorithmOverview of the algorithm– One source generates the sensory update One source generates the sensory update

traffic possibly on behalf of a group of local traffic possibly on behalf of a group of local sensorssensors

– The data updates are disseminated along a The data updates are disseminated along a tree to the mobile sinks in an asynchronous tree to the mobile sinks in an asynchronous mannermanner

– Each branch of the tree may have its own Each branch of the tree may have its own update rate update rate depending on the desired refresh rate of the depending on the desired refresh rate of the

downstream observersdownstream observers



Overview of the algorithm (cont.)Overview of the algorithm (cont.)– UU : the average update rate of the source : the average update rate of the source– UUmm : a minimum update rate, : a minimum update rate,

to detect failures or packet loss in the sensor to detect failures or packet loss in the sensor networknetwork

If a source has no new sensor readingsIf a source has no new sensor readings– It disseminates idle messages along the tree at rate UIt disseminates idle messages along the tree at rate Umm

– If a node in the tree receives no messages If a node in the tree receives no messages including idle messages from the source for a including idle messages from the source for a period longer than period longer than 1/ U1/ Umm The node contacts its parentThe node contacts its parent If its parent has failed and gives no response, If its parent has failed and gives no response,

– The node asks for a new parent by sending an error The node asks for a new parent by sending an error message to the source of the d-treemessage to the source of the d-tree



When a mobile sink wants to join the d-When a mobile sink wants to join the d-treetree– It selects one of its neighboring sensor nodes It selects one of its neighboring sensor nodes

to send a to send a join queryjoin query to the source of the tree to the source of the tree– the selected node is called the sink’s the selected node is called the sink’s access nodeaccess node

The access nodeThe access node– is used to represent the moving sink when the is used to represent the moving sink when the

optimal d-tree is builtoptimal d-tree is built– Amortize the overhead in the presence of Amortize the overhead in the presence of

mobilitymobility



The tree delivers data to the fixed access The tree delivers data to the fixed access node. In turn, the access node delivers the node. In turn, the access node delivers the data to the sink without exporting the sink’s data to the sink without exporting the sink’s location information to the rest of the treelocation information to the rest of the tree

The tree is updated only when the access The tree is updated only when the access node changesnode changes

As the sink moves, no new access node is As the sink moves, no new access node is chosen until the hop count between the chosen until the hop count between the access node and the sink exceeds a access node and the sink exceeds a thresholdthreshold– Trade-off between path delay and energy spent Trade-off between path delay and energy spent

on reconstructing the treeon reconstructing the tree



Source data is replicated at selected Source data is replicated at selected nodes between the source and sinksnodes between the source and sinks

We define a We define a replicareplica as a sensor node as a sensor node that stores a copy of the source datathat stores a copy of the source data– It temporarily stores the latest data It temporarily stores the latest data

incoming from the source and incoming from the source and asynchronous disseminate it to others asynchronous disseminate it to others along the treealong the tree



DefinitionDefinition– VV_ _ the sensor nodes in the sensor networkthe sensor nodes in the sensor network– BB_ _ sinks indexed by sinks indexed by n=1,2,…,Nn=1,2,…,N and and m=1,2,…,Mm=1,2,…,M

– AA__A={AA={A11,A,A22,…,A,…,AMM}} V⊂ be the set of M access nodes for mobile sinks B={B{B11,B,B22,…,B,…,BMM}} which request which request data from the source at refresh rates data from the source at refresh rates R={RR={R11,R,R22,…,R,…,RMM}}

– Energy_cost(a,b) Energy_cost(a,b) ∝ d(a,b)Pab

PPabab— packet sending rate— packet sending rate d(a,b)d(a,b)—distance between a and b—distance between a and b


SEADSEAD Subscription QuerySubscription Query

– Sink directs a join query to source via its access nodeSink directs a join query to source via its access node Gate replica searchGate replica search

– A gate replica is determined, which serves as the A gate replica is determined, which serves as the grafting point from which a branch to the new access grafting point from which a branch to the new access point is extendedpoint is extended

Replica placementReplica placement– locally readjusts the tree in the neighborhood of the gate

replica to further reduce communication energy D-tree managementD-tree management

– The constructed tree is managed to accommodate mobile sinks or defective regions such as a group of congested or failed nodes


SEAD –SEAD –Subscription QuerySubscription Query

Mobile sinks beacon periodically to determine Mobile sinks beacon periodically to determine their neighborstheir neighbors

A mobile sink A mobile sink BBii selects the nearest of its adjac selects the nearest of its adjacent nodes as the access node ent nodes as the access node AAii

– BBii sends a sends a join queryjoin query to a source via to a source via AAii

– The The join queryjoin query message contains the location of the message contains the location of the AAii

– and the sink’s desired updated rate and the sink’s desired updated rate RRii

The access node directly sends the The access node directly sends the join queryjoin query to to the source via the underlying routing protocolthe source via the underlying routing protocol


SEAD—SEAD—Gate Replica SearchGate Replica Search

Each node Each node nn in the tree has a set in the tree has a set C(n)C(n) of children of children It maintains a downstream rate It maintains a downstream rate QQnn

cc for each chil for each child d cc ∈ C(n)C(n)

The algorithm starts when a source receives a qThe algorithm starts when a source receives a query indicating a sink’s desired refresh rate, uery indicating a sink’s desired refresh rate, RRii

Each level of replica Each level of replica rr, including the source rece, including the source receiving that message, runs a recursive search as fiving that message, runs a recursive search as followsollows– If If QQnn

c c < R< Ri i ,then change ,then change QQnnc c value tovalue to R Ri i


SEAD—SEAD—Gate Replica Search (cont.)Gate Replica Search (cont.)

When a replica r is connected to the access node When a replica r is connected to the access node AAii, th, th

e additional cost, K(r), is calculated ase additional cost, K(r), is calculated as


SEAD—SEAD—Gate Replica Search (cont.)Gate Replica Search (cont.) Consider the incremental cost, Consider the incremental cost, K(r)-K(c)K(r)-K(c) where where rr is is cc’s parent ’s parent Which relates the cost at node r to that at its child c (4)Which relates the cost at node r to that at its child c (4)


SEAD—SEAD—Gate Replica Search (cont.)Gate Replica Search (cont.)


SEAD—SEAD—Replica PlacementReplica Placement Locally adjusts the tree around the gate

replica to produce an optimal d-tree There are two ways to connect the access There are two ways to connect the access

node to the gate replicanode to the gate replica– Non-replica modeNon-replica mode

Connect it as a child of the gate replicaConnect it as a child of the gate replica Adds no replicas to the treeAdds no replicas to the tree

– Junction modeJunction mode Create a child for the gate replica to feed the access Create a child for the gate replica to feed the access

node and some of the gate replica’s original childrennode and some of the gate replica’s original children The new child replica is called a junction replicaThe new child replica is called a junction replica


SEAD—SEAD—Replica Placement (cont.)Replica Placement (cont.)

The replica placement phase The replica placement phase – compares the non-replica mode cost compares the non-replica mode cost UUnon_replicnon_replic

aa to a junction replica mode cost to a junction replica mode cost UUjreplicajreplica

– Selects the better option so that the access Selects the better option so that the access node joins the tree in a way that minimizes tnode joins the tree in a way that minimizes the energy costhe energy cost



The gate replica The gate replica gg calculates the cost of the non-replica mode UUnon_replicanon_replica(c) for each child c C(g)∈


SEAD—SEAD—Replica Placement (cont.)Replica Placement (cont.) The gate replica g finds the neighbor node n among its a

djacent nodes within a singe hop range. Then it calculates the energy cost UUjreplicajreplica(c) for each child c



After getting both After getting both UUnon_replicanon_replica(c)(c) and and UUjreplicajreplica(c),(c), the g the gate replica finds one of its children ate replica finds one of its children cc ∈ C(g)C(g) to to maximizemaximize


SEAD—SEAD—D-tree managementD-tree management

Sink mobilitySink mobility


SEAD—SEAD—D-tree management (cont.)D-tree management (cont.)

Sink mobilitySink mobility



Leaving D-TreeLeaving D-Tree– A sink A sink BBii leaves the d-tree by sending a leaves the d-tree by sending a

leave message to its access nodeleave message to its access node

– The access node The access node AAii requests its parent to requests its parent to delete delete AAii from its list of children and stop from its list of children and stop forwarding data to itforwarding data to it



System LifetimeSystem Lifetime Avoiding defective points or areasAvoiding defective points or areas Complexity and resource Complexity and resource

requirementsrequirements


EvaluationEvaluation Simulation ： NS-2 A popular sensor node prototype ： MICA2 Power is supplied via disposable AA batteries.

– About 0.080W for transmitting and About 0.025W for receiving. The transceiver has a 250m radio range at 433 MHz, which is the case wit

h the radio transceiver of a MICA2 mote. The sensor network consists of (300 ≤ N ≤ 500) sensor nodes in a 2000

m × 2000m grid or 3-5 nodes per 200m × 200m use the two-ray ground model as the radio propagation model and an o

mni-directional antenna having unity gain in the simulation Each query packet is 36 bytes long and the data packet has 64 bytes The default number of sinks is 8. Three different sources generate different data at an average interval of

6 seconds The desired update rates are generated at random The energy consumption is measured in terms of Joules per node. The default mobile sink speed is set to 10 m/sec (i.e., the fastest human

speed)


Evaluation (cont.)Evaluation (cont.) Compare SEAD to

– Directed Diffusion (DD) [15] A data-centric communication paradigm specifically desig

ned for sensor networks Subscribers use flooding to spread interests to the sensor

network– TTDD [25]

Multicast protocol Exploits local flooding within a local cell of a grid which so

urces build proactively Each source disseminates data along the nodes on the gri

d line to the sinks– ADMR [16]

Mobile ad hoc network


Evaluation—Evaluation—Asynchronous dissemination (cont.)Asynchronous dissemination (cont.) Energy consumed for data packetsEnergy consumed for data packets


Evaluation—Evaluation—Asynchronous dissemination (cont.)Asynchronous dissemination (cont.) Average delayAverage delay


Evaluation—Evaluation—Sink mobilitySink mobility Energy consumption for the number of sinksEnergy consumption for the number of sinks


Evaluation—Evaluation—Sink mobility (cont.)Sink mobility (cont.) Energy consumption with different sink speedsEnergy consumption with different sink speeds


Evaluation—Evaluation—Sink mobility (cont.)Sink mobility (cont.) Remaining energy distribution of the sensor nodesRemaining energy distribution of the sensor nodes


Evaluation—Evaluation—Impact of the node densityImpact of the node density Energy per node with different sensor node densityEnergy per node with different sensor node density


Evaluation—Evaluation—Impact of the node density (cont.)Impact of the node density (cont.) Average delay with different sensor node densityAverage delay with different sensor node density


Evaluation—Evaluation—End-to-end delay of SEADEnd-to-end delay of SEAD Average delay with different number of sinksAverage delay with different number of sinks


Evaluation—Evaluation—End-to-end delay of SEAD (cont.)End-to-end delay of SEAD (cont.) Average delay with different sink speedsAverage delay with different sink speeds


ConclusionConclusion

SEADSEAD– Saves energy consumption in both Saves energy consumption in both

building the d-tree and maintaining building the d-tree and maintaining linkage to mobile sinkslinkage to mobile sinks

– Strikes a balance between end-to-end Strikes a balance between end-to-end delay and power consumption that delay and power consumption that favors power savings over delay favors power savings over delay minimizationminimization

– Is insensitive to changes in node densityIs insensitive to changes in node density

2004/2/10 2004/2/10jenchi minimum-energy asynchronous dissemination to mobile sinks in wireless...

Documents

sensor readingsmay

sensor networkscommunication

sensor networkif

basic model cont

neighboring sensor nodes

mobile sinksintroduction

mobile sinks place

access node changesas