Sharing data using Distributed HashTables on MobileAd-hoc Networks
ProjectThor
PrasannaGautam ‘11
Advisor: Dr.Timothy Richards
Computer Science Department
 Number of Internet enabled mobile devices is
increasing
(Source: ITU, Mark Lipacis, Morgan Stanley Research)
 Mobile Ad-hoc Networks (MANET)
 Self configuring wireless network
 Great interest in research and defense industry
0
50
100
150
200
250
300
350
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Research Papers trend (1995-2009)
Adhoc Network MANET "Adhoc Network" and MANET
(Source: mendeley.com)
 Pros
 Can be used in
▪ Disaster Scenarios
▪ When Infrastructure is unavailable
▪ Deploying infrastructure is too expensive
 Existing implementations in major OSes (802.11
ad hoc mode)
▪ Heterogeneous setup possible
 Quick setup
 Cheap to deploy
 Cons
 Power hungry
 Limited network access
 Churn
▪ Nodes coming in and out of network
 Difficult network management
 POSIT – search and catalog application
running on Android
 Started in 2008
 RWG Adhoc protocol implemented in
collaboration with Linköping University, Sweden
 RandomWalk Gossip
 Phones send messages randomly
 until a specified number of phones have received
it
 POSIT is deployed in a location without
network access
 RWG is used
 After a few hours, they start draining out
 Some phones will never see some data
 Power Usage
 Churn
 Nodes coming in and out of the network/range
 Without using network (ADP1)
 Phone lasted for ~132 hours = 5.5 days
0
20
40
60
80
100
120
0:00:00 24:00:00 48:00:00 72:00:00 96:00:00 120:00:00 144:00:00
Battery Level (no network usage) on HTC ADP1
Battery
 Sending ~50 bytes every second
 Phone lasted for 12.8 hours
▪ 14999 messages = 14999*50 bytes = 732.37 KB
0
20
40
60
80
100
120
0:00:00 2:24:00 4:48:00 7:12:00 9:36:00 12:00:00 14:24:00
battery performance when sending 50 bytes
per second
level
 Sending 1K file every 5 seconds
 Phone lasted for 13.5 hours
▪ 6998 messages = 6998 KB = 6.83 MB
0
20
40
60
80
100
120
0:00:00 2:24:00 4:48:00 7:12:00 9:36:00 12:00:00 14:24:00
battery performance when sending 1K of data
every 5 seconds
level
 Sending 1K file every second
 Phone lasted for 8 hours
▪ 14696 messages = 14696 KB = 14.35MB
0
20
40
60
80
100
120
0:00:00 1:12:00 2:24:00 3:36:00 4:48:00 6:00:00 7:12:00 8:24:00 9:36:00
battery performance when sending 1K of data
every second
level
 Talk less
 When you do talk, say more
 Nodes can come in an out of the network
 Could have multiple sub-networks
 Not necessarily defined structure/topology
 Distributed HashTables (DHTs)
 Distributed system provides lookup similar to
hash table
 Information is distributed and replicated among
nodes
 Used extensively by peer-to-peer applications
like Bittorrent
 A Churn and Mobility ResistantApproach for DHTs (Landsiedel et. al.)
 r being the communication range
 Data can be stored within r/2 distance to be reachable by a node n
 Study algorithms based on DHT on Mobile
Ad-hoc Networks
 Design and implement a specialized DHT
simulator program
 Test various conditions
▪ Power usage
▪ Topologies
▪ Effects of churning
▪ Network Partition
 RWG
 constantly communicating/sending data
 Random routing
 Can’t handle incomplete downloads
 DHT
 Minimal communication
 Intelligent routing
 Even incomplete data is useful
 DHTs have shown to
 Deal with churn effectively
 Reduce overall communication
 Using DHTs for communication on Adhoc
Networks should
 Improve power usage
 Handle churn better
Project thor

Project thor

  • 1.
    Sharing data usingDistributed HashTables on MobileAd-hoc Networks ProjectThor PrasannaGautam ‘11 Advisor: Dr.Timothy Richards Computer Science Department
  • 2.
     Number ofInternet enabled mobile devices is increasing (Source: ITU, Mark Lipacis, Morgan Stanley Research)
  • 3.
     Mobile Ad-hocNetworks (MANET)  Self configuring wireless network
  • 4.
     Great interestin research and defense industry 0 50 100 150 200 250 300 350 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Research Papers trend (1995-2009) Adhoc Network MANET "Adhoc Network" and MANET (Source: mendeley.com)
  • 5.
     Pros  Canbe used in ▪ Disaster Scenarios ▪ When Infrastructure is unavailable ▪ Deploying infrastructure is too expensive  Existing implementations in major OSes (802.11 ad hoc mode) ▪ Heterogeneous setup possible  Quick setup  Cheap to deploy
  • 6.
     Cons  Powerhungry  Limited network access  Churn ▪ Nodes coming in and out of network  Difficult network management
  • 7.
     POSIT –search and catalog application running on Android  Started in 2008  RWG Adhoc protocol implemented in collaboration with Linköping University, Sweden  RandomWalk Gossip  Phones send messages randomly  until a specified number of phones have received it
  • 8.
     POSIT isdeployed in a location without network access  RWG is used  After a few hours, they start draining out  Some phones will never see some data
  • 9.
     Power Usage Churn  Nodes coming in and out of the network/range
  • 10.
     Without usingnetwork (ADP1)  Phone lasted for ~132 hours = 5.5 days 0 20 40 60 80 100 120 0:00:00 24:00:00 48:00:00 72:00:00 96:00:00 120:00:00 144:00:00 Battery Level (no network usage) on HTC ADP1 Battery
  • 11.
     Sending ~50bytes every second  Phone lasted for 12.8 hours ▪ 14999 messages = 14999*50 bytes = 732.37 KB 0 20 40 60 80 100 120 0:00:00 2:24:00 4:48:00 7:12:00 9:36:00 12:00:00 14:24:00 battery performance when sending 50 bytes per second level
  • 12.
     Sending 1Kfile every 5 seconds  Phone lasted for 13.5 hours ▪ 6998 messages = 6998 KB = 6.83 MB 0 20 40 60 80 100 120 0:00:00 2:24:00 4:48:00 7:12:00 9:36:00 12:00:00 14:24:00 battery performance when sending 1K of data every 5 seconds level
  • 13.
     Sending 1Kfile every second  Phone lasted for 8 hours ▪ 14696 messages = 14696 KB = 14.35MB 0 20 40 60 80 100 120 0:00:00 1:12:00 2:24:00 3:36:00 4:48:00 6:00:00 7:12:00 8:24:00 9:36:00 battery performance when sending 1K of data every second level
  • 14.
     Talk less When you do talk, say more
  • 15.
     Nodes cancome in an out of the network  Could have multiple sub-networks  Not necessarily defined structure/topology
  • 16.
     Distributed HashTables(DHTs)  Distributed system provides lookup similar to hash table  Information is distributed and replicated among nodes
  • 17.
     Used extensivelyby peer-to-peer applications like Bittorrent
  • 18.
     A Churnand Mobility ResistantApproach for DHTs (Landsiedel et. al.)  r being the communication range  Data can be stored within r/2 distance to be reachable by a node n
  • 19.
     Study algorithmsbased on DHT on Mobile Ad-hoc Networks  Design and implement a specialized DHT simulator program  Test various conditions ▪ Power usage ▪ Topologies ▪ Effects of churning ▪ Network Partition
  • 20.
     RWG  constantlycommunicating/sending data  Random routing  Can’t handle incomplete downloads  DHT  Minimal communication  Intelligent routing  Even incomplete data is useful
  • 21.
     DHTs haveshown to  Deal with churn effectively  Reduce overall communication  Using DHTs for communication on Adhoc Networks should  Improve power usage  Handle churn better