Download event storage node

Event Query Processing
Based on Data-Centric Storage
in Wireless Sensor Networks
Longjiang Guo, Yingshu Li and
Jianzhong Li
Globalcom IEEE 2007
Outline
Introduction
Event formulation and storage
Event query processing
Choosing optimal storage strategy
Simulation
Conclusion
INTRODUCTION
interested in the events, instead of the
sensors
A event is a fusion of multiple sensed data
satisfying some conditions.
event fire: temperature>90 AND smoke>100L/mol,
find all the locations where ET satisfied the
pre-defined observation attributes.
EVENT FORMULATION AND STORAGE
preprocessing of the observed data :
includes a check on whether the data satisfies
the event conditions
If it does, the event information will be stored in
the sensor network.
EVENT FORMULATION AND STORAGE
A. Event Formulation




Definition 1: Observation attribute Ai (i=1…n)
Definition 2: Observation value
Definition 3: Round
Definition 4: Event type ETi (A1, A2, …, Aiki )
 a group of pre-defined observation attributes
 ET(Tmpt, Wnd, Hudit) = Tmpt>t1 and Wnd>w2 and Hudit>h3
 Definition 5: Observation zone
divided into m×n grids ,each grid is an observation
zone.
 Definition 6: Event e(ET, g, t)
 Definition 7: Event query Q{[t1, t2],ETi}
EVENT FORMULATION AND STORAGE
A. Event Formulation
The observation nodes of ETi : Si1, Si2, …, Siki ,
where node Sij observes attribute Aij.
W: types of events ET1 , ET2 ,…, ETw
EVENT FORMULATION AND STORAGE
A. Event Formulation
Every node Sij broadcasts a message
within its observation zone.
ETi , Sij’s ID , Sij’s position
When a node Sij receives a message,
it firstly checks whether the event type is the
same as its own
forwards the message to its neighbors
EVENT FORMULATION AND STORAGE
A. Event Formulation
For each observation zone, in the kth round,
Sij is selected as the event fusion node
 j=(k mod ki )+1
the event fusion node
routes the event to
storage node
Si2 ,…, Si6 route IDs
and values to Si1
Broadcast Si1 ‘s position
Si2
Si1
k=1, Si1 as event fusion node
Si6
Si5
Si3
Si4
EVENT FORMULATION AND STORAGE
B. Event Storage
 External storage:
the sink is the event storage node
an event fusion node will route the events directly to the
sink
 Local storage:
the event storage node is the same as the event fusion
node.
 Data centric storage:
hash to an in-network position (x, y) according to the
event type.
nearest to that position, called Home Node
EVENT FORMULATION AND STORAGE
B. Event Storage
 More energy can be saved to put Home Node at
the center of a network.
 Center Mapping Data Centric Storage (CM-DCS)
 Events of the same type will be hashed to an
observation zone lying at the center of the
network.
 A node in the observation zone nearest to the
center of the network serves as an event storage
node
EVENT FORMULATION AND STORAGE
B. Event Storage
 Lemma 1:
the total energy consumption for routing events
is proportional to the distance between the
event fusion node and the event storage node.
 Proof: This is obvious.
EVENT FORMULATION AND STORAGE
B. Event Storage
 Lemma 2:
 If observation nodes are distributed uniformly
in an observation zone,
event fusion node is the center of the
observation zone.
 Proof:
event fusion node’s
(c, d)
(E(Xf), E(Yf))
(a, b)
position
(Xf, Yf) : node (X, Y)
an observation
is a two dimensional
E(Xf)=0.5(a+c)
random variable.
E(Yf)=0.5(b+d)
EVENT FORMULATION AND STORAGE
B. Event Storage
 Theorem 1:
If event storage nodes are located near the center of the
sensor network,
energy consumption for routing events from an event
fusion node to an event storage node is minimized.
 Proof:
(X1, Y1), (X2, Y2), …, (Xmxn, Ymxn) :centers of the
observation zones (fusion nodes).
 (X, Y) : event storage node


EVENT FORMULATION AND STORAGE
B. Event Storage
Mapping:
{ET1→Z6； ET2→Z10；
ET3→Z11；ET4→Z7}
: storage node
EVENT FORMULATION AND STORAGE
C. Storage at an Individual Event Storage
Node
each event storage node is responsible for
event type ETi
time-stamped vector-based storage
strategy:
 If (j1, j2, …, jk of I) = 1 ,
observation zones
event of type ETi.
detected the
EVENT QUERY PROCESSING
extract information from the sensor
network for query Q{[t1, t2], ETi}.
Event query processing based on CMDCS
Event query processing based on Local
Storage
EVENT QUERY PROCESSING
A. Event query processing based on CM-DCS
 Phase 1: Deciding the routing destination.
 Phase 2: Routing query Q{[t1, t2], ETi}. from the
sink to the event storage node p
 Phase 3: Answering query Q{[t1, t2], ETi}.
 Phase 4:Routing Ip back to the sink from node p.
A={(I,t)} | t  [t1 , t2 ],
then
Ip 

( I ,t ),t[ t1 ,t 2 ]
sink
I
Ip (x , y )
i
i
p
ETi
Q{[t1 , t2], ETi}.
EVENT QUERY PROCESSING
B. Event query processing based on Local
Storage
 Phase 1:Query dissemination.
sink broadcasts the query, add its ID to query packet.
p replaces the ID in the query packet with its own ID
 Phase 2:Collection of children’s IDs.
p broadcasts <p,fp> , where fp is p’s parents
p receives {<q1, fq1>, < q2, fq2 >, …, < qm, fqm >},
 Children(p)={qi| <qi, fqi>, where fqi =p}.
 Phase 3:Combination of query results.
event storage node p receives an event query
A={(I,t)} | t [t1 , t2 ], then p computes I p   I
( I ,t ),t[ t ,t ]
 Phase 4:Routing Ip back to the sink from node p.
1 2
CHOOSING OPTIMAL STORAGE
STRATEGY
provide a guideline of choosing a correct
storage strategy for different applications.
CHOOSING OPTIMAL STORAGE
STRATEGY
CHOOSING OPTIMAL STORAGE
STRATEGY
A. Estimation of Energy Consumption For
Initialization
broadcast a message and forward a message
to its neighbors
Receive from neighbors/broadcast to neighbors
CHOOSING OPTIMAL STORAGE
STRATEGY
B. Estimation of Energy Consumption For
Event Formulation
The expected number of hops (nodefusion
node) =sqrt(N/(8mn))
(A random node fusion node)
CHOOSING OPTIMAL STORAGE
STRATEGY
C. Estimation of Energy Consumption For
External Storage
The expected number of hops (nodesink)
=(2N)/2
#events
(node fusion node)
(fusion node sink)
CHOOSING OPTIMAL STORAGE
STRATEGY
D. Estimation of Energy Consumption For
CM-DCS
The expected number of hops (node storage
node) =(2N)/4
(node fusion node)
(fusion node storage )
Query: (sink storage node)
Answer: (storage node sink)
CHOOSING OPTIMAL STORAGE
STRATEGY
E. Estimation of Energy Consumption For
Local Storage
Query
Answer
CHOOSING OPTIMAL STORAGE
STRATEGY
F. Comparing the Energy Consumptions
Assume: Eb =Eu , Er =1.5Eb , Sevent =Squery =0.25Sresult
Observation 1: if Nq >N·prob, external storage
Observation 2：
ifρ and Nq increase,CM-DCS
if ρ is a constant and N increases, local storage
SIMULATION RESULTS
SIMULATION RESULTS
SIMULATION RESULTS
SIMULATION RESULTS
SIMULATION RESULTS
SIMULATION RESULTS
CONCLUSION
propose a data centric storage strategy
CM-DCS
event query processing algorithms:
EP-CM-DCS and EP-LS.
compare the energy consumptions
users can have a guideline of choosing a
correct storage strategy for different
applications.