Multiple Query Optimization for Wireless Sensor Networks

1
Multiple Query Optimization for Wireless Sensor
Networks

• Shili Xiang Hock Beng Lim Kian-Lee Tan
• (ICDE 2007)

Presented by Shan Bai
2
Highlight

• Introduction
• Background
• Challenge
• Goal of this paper
• Multiple Query Optimization
• Base station optimization
• In-network optimization
• Discussion
• References

3
Introduction

• Background
• WSN are deployed in many important applications
  to query the physical world.
• (environmental monitoring, healthcare monitoring,
  military surveillance, tracking of goods and
  manufacturing processes, traffic monitoring,
  etc.)
• The sensor network needs to support the efficient
  processing of multiple queries.

4
Introduction

• Users Requirement
• Users can issue declarative queries without
  having to worry about how the data are generated,
  processed, and transferred within the network,
  and how sensor nodes are (re)programmed to
  satisfy changing user interest. ( User
  transparency)
• The WSN should be able to concurrently handle
  several user requests through running multiple
  queries
• Challenge
• Various data from the network at the same time,
  and both users and their interests can change
  over time.

5
Introduction

• Current situation
• Several sensor data query processing systems,
  such as Cougar 16, 4 and TinyDB 9, have been
  developed by the database research community.
• However, most existing work on sensor data query
  processing has focused on the optimization and
  execution of a single long-running query.
• ---systems cannot amortize the data acquisition,
  computation and communication cost of fetching
  the common data for multiple queries.
• ---lead to bandwidth contention and even data
  loss as a result of transmission collisions
  (which may in turn require retransmission).

6
Introduction

• Goal of this paper
• To design a light-weight but effective scheme to
  support multiple data acquisition and aggregation
  queries in a wireless sensor network, in order to
  minimize the number of radio transmissions.
• Similar queries to share the limited
  communication and computational resources.

7
Highlight

• Introduction
• Background
• Challenge
• Goal of this paper
• Multiple Query Optimization
• Base station optimization
• In-network optimization
• Discussion
• References

8
Multiple Query Optimization Base station
optimization

• Base station optimization
• Use the base station as a filter to reduce
  duplicate data accesses from the sensor network,
  and as a screen to hide the query dynamics as
  much as possible.

9
Multiple Query Optimization Base station
optimization

• Given a set of queries Q that has been submitted
  to the base station, rewrite them into a new
  query set Q. The optimal situation is that data
  requested by queries in Q will be just enough to
  answer queries in Q, and the same data needed for
  various queries in Q will be acquired only once
  by queries in Q
• COST MODEL
• VOL cost of one query, the number of its result
  dissemination messages in a unit of time.
• C whole data space
• d the average depth of nodes in the network
• sel(p) the selectivity of predicates p
• As a result of multi-hop routing protocols, the
  cost of data acquisition query qi with sampling
  period si can be estimated as

10
Multiple Query Optimization Base station
optimization

• Define metric Benefit to quantify the cost
  savings by query rewriting.
• It is beneficial to write q1 and q2 into q if
  and only if Benefit12 0. We have the following
  theorem, the proof of which is omitted due to
  space constraint.
• Theorem 1. Benefit12 0 only if GCD(s1, s2)
  s1 or GCD(s1, s2) s2.
• GCD Greatest Common divisor of s1,s2 sample
  period

11
Multiple Query Optimization Base station
optimization

• to identify the most beneficial synthetic query
  qj to rewrite with this qi.

12
Multiple Query Optimization Base station
optimization

• it is possible that multiple synthetic
  queries can benefit from the newly integrated
  synthetic query.

13
Multiple Query Optimization Base station
optimization

• Iterative query Insertion algorithm
• The main idea whenever a synthetic query is
  updated, it is checked against the synthetic
  query list to see if it is beneficial to other
  synthetic queries if so, the most beneficial
  pairs are rewritten, and the newly updated
  synthetic query will be checked against the
  synthetic query list this process terminates
  when there is no further beneficial rewriting.
• To achieve this, after Integrate (qid, qi) in
  line 16 in Algorithm 1 has updated the synthetic
  query qid into a new one, Insert (qid,Qsyn).
• The iterative query insertion algorithm is
  expected to reduce more redundancy among the data
  requested by user queries

14
Multiple Query Optimization Base station
optimization

• To enable our multi-query optimization scheme to
  perform well for dynamic workloads where user
  queries are inserted at different frequency and
  run for various duration, introduce a parameter a
  to adjust our query termination algorithm
  according to the property of application workload.

15
Multiple Query Optimization Base station
optimization

• Summary
• When there are several queries in the system, and
  substantial similarities between queries, the
  query insertion and termination can most likely
  be handled at the base station, in terms of
  reduction in the number of radio messages and the
  scalability of the number of concurrent queries,
  without affecting the sensor network.

16
Highlight

• Introduction
• Background
• Challenge
• Goal of this paper
• Multiple Query Optimization
• Base station optimization
• In-network optimization
• Discussion
• References

17
Multiple Query OptimizationIn-network
optimization

• base station optimization cannot support sharing
  of the commonality among queries at the finest
  granularity.
• base station optimization cannot take advantage
  of special property of sensor nodes, such as the
  broadcast nature of radio transmission.
• Sensor nodes make local decisions themselves and
  adaptively handle the query workload with time.

18
Multiple Query Optimization

• In-network optimization
• Sharing over time.
• more progressive sharing over time by scheduling
  the data acquisition and transmission of all
  queries in a whole.
• At the end of a querys propagation phase,
  setSampleRate is triggered, which may start (or
  restart) the nodes clock to fire at the GCD of
  the epoch duration of all the queries. We set
  the epoch start time on sensor nodes to be
  divisible by the epoch duration instead of the
  arrival time of a new query (here we assume that
  every epoch duration is divisible by 2048ms).

19
Multiple Query Optimization

• In-network optimization
• Sharing over space
• Each sensor node dynamically selects a route
  (parent) that is aware of the query space in the
  meanwhile, it tries to take advantage of the
  broadcast nature of the radio channel to satisfy
  multiple queries in one message.

20
Multiple Query Optimization

• In-network optimization
• Sharing over space
• .

Base station
A
B
F
C
D
E
G
H
Routing tree in DAG (Directed Acyclic Graph)
21
Highlight

• Introduction
• Background
• Challenge
• Goal of this paper
• Multiple Query Optimization
• Base station optimization
• In-network optimization
• Discussion
• References

22
Discussion

• Pros
• Two optimization tiers are similar and
  complementary to each other. Both of them can
  eliminate the duplicate transmission of the same
  data for several data acquisition queries

23
Discussion

• Cons
• The base station optimization is somewhat more
  constrained by the granularity while the
  in-network optimization will result in a bigger
  result message size.
• In in-network optimization, aggregation queries
  can only benefit among themselves with semantic
  correctness guarantee.
• The authors should provide more detail evidence
  showed to prove the performance improvements over
  the traditional single query optimization
  technique -- design some real algorithms and
  simulation combine base-station and in-station in
  a whole.
• To achieve a robust data transmission for the
  data that is requested by many user queries.

24
References

• 1 A. Demers, J. Gehrke, R. Rajaraman, N.
  Trigoni, and Y. Yao. The cougar project A
  work-in-progress report. SIGMOD Record, 32(4),
  2003.
• 2 S. Madden, M. J. Franklin, J. M. Hellerstein,
  and W. Hong.TINYDB An acquisitional query
  processing system for sensor networks. ACM TODS,
  30(1), November 2005.
• 3 S. Xiang, H. B. Lim, and K. L. Tan. Impact of
  multi-query optimization in sensor networks. In
  Proc. of DMSN, 2006.
• 4 A. Demers, J. Gehrke, R. Rajaraman, N.
  Trigoni, and
• Y. Yao. The cougar project A
  work-in-progress
• 9 S. Madden, M. J. Franklin, J. M. Hellerstein,
  and W. Hong. TinyDB An acquisitional query
  processing system for sensor networks. ACM TODS,
  30(1), November 2005.
• 16 Y. Yao and J. Gehrke. Query processing for
  sensor networks. In Proc. of CIDR, 2003.

25
Questions/Comments?