Searching the Physical World: Distributed Protocols for Data Coverage and Caching in WSNs

1
Searching the Physical World Distributed
Protocols for Data Coverage and Caching in WSNs
_at_ Dept. of Computer Communication Engineering,
University of Thessaly
Dimitrios Katsaros, Ph.D.
Nicosia, June 17th, 2008
2
Outline of the talk

• WSNs A working reality
• What is the Sensory Web?
• Data Coverage issues in WSNs
• Cooperative Caching for WSNs
• Concluding remarks

3
Outline of the talk

• WSNs A working reality
• What is the Sensory Web?
• Data Coverage issues in WSNs
• Cooperative Caching for WSNs
• Concluding remarks

4
Wireless Sensor Networks (WSNs)

• Wireless Sensor Networks features
• Homogeneous devices
• Stationary nodes
• Dispersed network
• Large network size
• Self-organized
• All nodes acts as routers
• No wired infrastructure
• Potential multihop routes

5
WSNs - Applications
6
More exotic applications of WSNs
7
Whats special about WSNs ?

• Resource constraints
• sensor nodes are battery-, memory- and
  processing-starving devices
• Variable channel capacity
• multi-hop nature of WSNs implies that wireless
  link capacity depends on the interference level
  among nodes
• Multimedia in-network processing
• sensor nodes store rich media (image, video), and
  must retrieve such media from remote sensor nodes
  with short latency

8
Challenges

• Huge network size
• Unknown/variable network topology
• Agnostic users
• Fault tolerance
• Sensor readings are simply votes

9
Outline of the talk

• WSNs A working reality
• What is the Sensory Web?
• Data Coverage issues in WSNs
• Cooperative Caching for WSNs
• Concluding remarks

10
Research areas Ultimately ? ???
Sensory Web
Mobile/Pervasive Computing
Overlay Nets
Web
Mobile Ad Hoc
Wireless Sensors Networks
Information Retrieval
11
Search Engines for the Physical World

• Cooperating Sensors
• Distributed Protocols
• Energy-efficient Communication
• Short-latency Data Retrieval
• Unknown Network Topology
• Topology Control
• Storage in Flash Devices

12
Outline of the talk

• WSNs A working reality
• What is the Sensory Web?
• Data Coverage issues in WSNs
• Cooperative Caching for WSNs
• Concluding remarks

13
Querying WSNs

• Simple queries, e.g., Report the value of the
  humidity
• Aggregate queries, e.g., Report the average
  humidity of all sensors in region X
• Approximate queries, requiring data summarization
  to perform holistic data aggregation in the form
  of histograms, contour maps, e.g., Report the
  contour of toxic chemical gas in region X
• Complex queries, which, if expressed in SQL,
  would involve joins nested or conditioned-based
  sub-queries, e.g., Among regions X and Y, report
  the average humidity of the region with the
  highest temperature
• Advanced queries, such as top-k queries, e.g.,
  Report the k data objects with the highest
  temperature

14
Qyerying limitations (1/2)
Report the k smallest values of humidity within
region X along with the sensors that sensed them
What about sensor failures?
15
Qyerying limitations (2/2)
Report the k smallest values of humidity across
the whole sensornet along with the sensors that
sensed them
What about small shifts in the region boundaries?
16
The concept of Data Coverage
Report the sensor(s) whose humidity value is not
covered by any other humidity value across the
whole sensornet
Sensor with max humidity value
17
The concept of k-Data Coverage
Report the sensor(s) whose humidity value is
covered by at most k (e.g., k2) other humidity
values across the whole sensornet
Sensor with max value
Sensor with 2nd max value
Sensor with 3rd max value
18
Feature Distribution Maps
Still, we can not find out what happens in
neighborhoods, i.e., local minima, local maxima,
etc. These are not network-wide (global)
19
The concept of d-hop k-Data Coverage
Depict the points (i.e., sensors) with the
largest, relative to their neighboring sensors,
humidities

• localized definition of neighborhoods
• no region prespecification
• define d to be the sensornet diameter
• Network-wide k-coverage

20
The d-hop k-Data Coverage problem

• Generalizes
• The k-skyband query
• The top-k query
• The d-hop dominating set formation problem
• Deals with
• Any number of readings by a sensor node
• Any number of measured quantities, e.g.,
  humidity, temperature, etc.
• More generic predicates, not only maximum, minimum

21
Data Coverage in Neighborhoods-DaCoN

• Distributed protocol for processing d-hop k-data
  coverage queries in WSNs
• Runs localized in neighborhoods
• No network spanners, e.g., aggregation tree,
  spanning tree
• No demanding initialization phase to construct
  the spanner
• Uniform energy consumption, no hot spots of
  communication
• Runs in 3 phases

22
DaCoNs execution

• In a 2-dimensional space, assume that we wish the
  maximization of the first dimension and the
  minimization of the second one
• v_i.d_x denotes the x-th dimension of value v_i
• v_i covers a value v_j, if it holds
• v_i.d_1 gt v_j.d_1 and v_i.d_2 lt v_j.d_2

23
PHASE 1. First d-rounds

• Each sensor sends its k-th larger values to all
  its 1-hop neighbors
• It finds the k-th larger values taking account
  its own values and the values that has received
  from its neighbors
• It forms a message with these values and it
  stores the message into a buffer frb
• In the next d-1 rounds, the above procedure is
  repeated with the difference that now each sensor
  considers as its k-th larger values, the values
  of the last message of the frb

24
PHASE 2. Next d-rounds

• Similarly to the previous rounds, but
• Each sensor finds its k-th values by taking into
  account the previous message and the messages
  that has received from its neighbors as follows
  each v_i value (1 i k) is selected by
  keeping the smaller i-th value of these
  messages
• These values form a message that is stored into a
  buffer srb

25
PHASE 3. Answer of query

• Each value v_i (1 i k) of the answer is
  selected as follows the sensor compares
  the messages of frb and srb and tries to find
  pairs of values in the first i-th values of each
  message
• After the identification of all pairs of
  values, the sensor selects the minimum pair as
  the i-th value of its answer
• If a pair of values does not exist, the sensor
  selects the maximum of the first i-th values of
  the messages of frb

26
(No Transcript)
27
DaCoN evaluation

• No competing methods
• Network topologies,
• existence and strength of clusters of sensors
• density of sensor nodes, etc
• Sensor data generator

28
Impact of sensornet size messages
29
Impact of sensornet size activated sens
30
Impact of assortativity messages
31
Impact of assortativity activated sens
32
Impact of k (500 sensors) activated sens
33
Impact of k (1000 sensors) activated sens
34
d-hop k-data coverage

• Feature Distribution Maps
• Fully distributed solution DaCoN
• Little overhead
• Little storage
• Light computational load
• Few messages no hotspots in communication
• How do we improve upon the latency, when the
  sensors need data from other sensors?
• Cooperative Caching

35
Outline of the talk

• WSNs A working reality
• What is the Sensory Web?
• Data Coverage issues in WSNs
• Cooperative Caching for WSNs
• Concluding remarks

36
Our proposal

• Cooperative Caching NICOCA protocol
• multiple sensor nodes share and coordinate cache
  data to cut communication cost and exploit the
  aggregate cache space of cooperating sensors

• Each sensor node has a moderate local storage
  capacity associated with it, i.e., a flash memory
• Jim Gray predicted that flash memories will
  replace hard disks

37
Relevant work (1/2)

• Caching in OSs, DBMS, Web
• No extreme resource constraints
• Caching for wireless broadcast cellular networks
• more powerful nodes,
• one-hop communication with resource-rich base
  stations
• Most relevant research works
• cooperative caching protocols for MANETs
• GroCoca organize nodes into groups
• based on data request pattern mobility pattern)
• ECOR, Zone Co-operative, Cluster Cooperative
  form clusters of nodes
• based geographical proximity or adopting node
  clustering algorithms for MANETs

38
Relevant work

• Protocols that deviated from such approaches
• CacheData intermediate nodes cache the data to
  serve future requests instead of fetching data
  from their source
• CachePath mobile nodes cache the data path and
  use it to redirect future requests to the nearby
  node which has the data instead of the faraway
  origin node
• Amalgamation of them the champion HybridCache
  cooperative caching for MANETs

39
NICoCa consists of

• A metric for estimating the importance of a
  sensor node, which will imply short latency in
  retrieval
• A cooperative caching protocol which strives to
  achieve uniform energy consumption
• Datum discovery and cache replacement component
  subprotocols
• Performance evaluation of the protocol and
  comparison with the state-of-the-art cooperative
  caching for MANETs, with J-Sim

40
Terminology and assumptions

• WMSN is abstracted as a graph G(V,E)
• edge e(u,v) exists iff u is in the transmission
  range of v and vice versa (bidirectional links)
• The network is assumed to be connected
• N1(v) the set of one hop neighbours of v
• N2(v) the set of two hop neighbours of v
• N12(v) combined set of N1(v) and N2(v)
• LNv is the induced subgraph of G associated
  with vertices in N12(v)
• dG(v,u) distance between v and u

41
A measure of sensor importance

• suw swu number of shortest paths from u ? V to
  w ? V (suu0)
• suw(v) number of shortest paths from u to w
  that some vertex v ? V lies on
• Node importance index NI(v) of a vertex v is

42
The NI index in sample graphs
43
The NI index in sample graphs

• Nodes with large NI
• Articulation nodes (in bridges), e.g., 3, 4, 7,
  16, 18
• With large fanout, e.g., 14, 8, U

44
Centralized solution ???

• Create a broadcast tree to coordinate the
  identification of NIs
• lot of messages
• larger latency
• Hot-spots in communication (nodes with large NI)
• Localized Algorithms are preferable
• NIs in neighborhoods

45
The NI index in a localized algorithm
2-hop neighbors of node 8
node 8 calculates the NI of its 2-hop neighbors
46
The NI index in a localized algorithm
nodes 14 and 16 are more important than the
others from the viewpoint of node 8
Each node can identify its own important nodes
47
Housekeeping information in NICoCa

• Ultimate source of multimedia data Data Center
• Each node is aware of its 2-hop neighborhood
• Uses NI to characterize some neighbors as
  mediators
• Can be either a mediator or an ordinary node
• Each sensor node stores
• the dataID, and the actual datum
• the data size, TTL interval
• for each cached item
• characterized either as O (i.e., own) or H (i.e.,
  hosted)
• the timestamps of the K most recent accesses

48
The cache discovery protocol (1/2)

• A sensor node issues a request for a multimedia
  item
• Searches its local cache and if it is found
  (local cache hit) then the K most recent access
  timestamps are updated
• Otherwise (local cache miss), the request is
  broadcasted and received by the mediators
• These check the 2-hop neighbors of the requesting
  node whether they cache the datum (proximity hit)
• If none of them responds (proximity cache miss),
  then the request is directed to the Data Center

49
The cache discovery protocol (2/2)

• When a mediator receives a request, searches its
  cache
• If it deduces that the request can be satisfied
  by a neighboring node (remote cache hit),
  forwards the request to the neighboring node with
  the largest residual energy
• If the request can not be satisfied by this
  mediator node, then it does not forward it
  recursively to its own mediators, since this will
  be done by the routing protocol, e.g., AODV
• If none of the nodes can help, then requested
  datum is served by the Data Center (global hit )

50
The cache replacement protocol

• Each sensor node first purges the data that it
  has cached on behalf of some other node
• Calculate the following function for each cached
  datum i

• The candidate cache victim is the item which
  incurs the largest cost
• Inform the mediators about the candidate victim
• If it is cached by a mediator, the metadata are
  updated
• If not, it is forwarded and cached to the node
  with the largest residual energy

51
Evaluation setting (1/2)

• We compared NICOCA to
• Hybrid, state-of-the-art cooperative caching
  protocol for MANETs
• Implementation of protocols using J-Sim
  simulation library

52
Evaluation setting (2/2)

• Measured quantities
• number of hits (local, remote and global)
• residual energy level of the sensor nodes
• average latency for getting the requested data
• the number of packets dropped
• Present here only results for number of hits
• representative of latency, collisions and energy
  consumption
• A small number of global hits
• less network congestion, fewer collisions and
  packet drops.
• Large number of remote hits ? effectiveness of
  cooperation
• Large number of local hits ? effective
  cooperation
• the cost of global hits vanishes the benefits of
  local hits

53
Cache vs. hits (MB files uniform access) in a
sparse WMSN (d 4)
54
Cache vs. hits (MB files uniform access) in a
dense WMSN (d 7)
55
Cache vs. hits (MB files uniform access) in a
very dense WMSN (d 10)
56
Observe MB files uniform access

• For all network topologies (sparse, dense and
  very dense), NICoCa achieves more remote hits and
  less global hits than HybridCache
• This performance gap widens in favor of NICoCa as
  we move from sparse to denser WMSNs
• For very dense sensor deployments, NICoCa
  achieves double the remote hits of HybridCache
  and only half of its global hits
• For sparse WMSNs HybridCache achieves slightly
  more local hits than does NICoCa, but this gap
  vanishes completely when moving to denser network
• This small gain of HybridCache for sparse
  topologies is not advantageous at all, since it
  incurs global hits as many as twice the number of
  its local hits

57
Cache vs. hits (KB files Zipfian access) in a
sparse WMSN (d 4)
58
Cache vs. hits (KB files Zipfian access) in a
dense WMSN (d 7)
59
Cache vs. hits (KB files Zipfian access) in a
very dense WMSN (d 10)
60
Observe KB files Zipfian access

• For all network topologies (sparse, dense and
  very dense), NICoCa achieves more remote hits and
  less global hits than HybridCache
• For very dense WMSNs, the requests reaching Data
  Center for NICoCa are less than half those of
  HybridCache!
• NICoCa's global hits do not vary significantly
  with varying network topologies and varying local
  sensor storage
• Global hits of HybridCache are severely affected
  by the topology and the cache size
• For cache equal to 1 of the total data,
  HybridCache's global hits increase at a pace of
  50!
• The results for Zipfian access on megabyte-sized
  data more impressively in favor of NICoCa

61
Summary

• Wireless Sensor Networks (WSNs)
• Cooperation among sensors
• Distributed protocols
• A brand new world or Distributed Algorithms
  reloaded?
• Exploit the unknown network topology!
• Impresice/incomplete queries!
• New storage devices (flash)
• Minimize energy consumption
• Minimize latency

62
Thank you for your attention!

• Any questions?

63
Important references

1. N. Dimokas, D. Katsaros, Y. Manolopoulos.
   Cooperative caching in wireless multimedia sensor
   networks. ACM Mobile Networks and Applications,
   accepted, May 2008
2. M. Kontaki, D. Katsaros, Y. Manolopoulos. The
   d-hop k-data coverage query problem in wireless
   sensor networks. Submitted, June 2008
3. D. Katsaros, Y. Manolopoulos. Prediction in
   wireless networks by Markov chains. IEEE Wireless
   Communications magazine, (under second round
   review), April 2008
4. L. Yin and G. Cao. Supporting cooperative caching
   in ad hoc networks. IEEE Transactions on Mobile
   Computing, 5(1)77-89, 2006