SNoD Sensor Networks on Defense Team Members: Kaustubh Supekar Gaurav Sharma Deepti Agarwal Aditya B

1
SNoDSensor Networks on DefenseTeam
MembersKaustubh SupekarGaurav SharmaDeepti
AgarwalAditya BarveBrijraj VaghaniSeema
JoshiDebashis HaldarGautam Kaushal
2
Smart Dust eh!!!

• The particles of dust that could be watching you
  .
• Smart Dust is the brainchild of Associate
  Professor Kris Pister and Professor Randy H.
  Katz, who are currently working at the
  University of California, Berkeley

3
Techie definition

• Smart Dust Sensor (also known as a sensor mote)
  is a tiny wireless micro-electromechanical sensor
  (MEMS) packed into a cubic millimeter speck that
  can detect anything from light to vibrations.

4
Mote Constraints

• Power, size and cost
• These get translated to
• Slow clock cycles of the micro controller.
• Less memory.
• Smaller number of hardware controllers.

5
MOTE Specifications

• Two Board Sandwich
• CPU/Radio board
• Sensor Board temperature, light
• Size
• Mote 1?1 in
• Pocket PC 5.2?3.1 in
• CPU
• Mote 4 MHz, 8 bit
• Pocket PC 133 MHz, 32 bit
• Memory
• Mote 512 B RAM 8K ROM
• Pocket PC 32 MB RAM 16 MB ROM
• Radio
• 900 Hz, 19.2 kbps
• Bluetooth 433.8 kbps
• Lifetime (Power)
• Mote 3-65 days
• Pocket PC 8 hrs

6
Software challenges

• Power efficient
• Less memory
• Application specific
• Concurrency-intensive operations
• Multiple, high-rate data flows (radio, sensor,
  actuator)
• TinyOS must process a bit every 100 µs
• Real-time
• Real-time query and feedback control of physical
  world
• Little memory for buffering data must be
  processed on the fly
• TinyOS No buffering in radio hw missed
  synchronization results in lost data.

7
Sensor Network

• A collection of sensor motes that perform
  autonomous sensing to form the basis of a
  massively integrated sensor network which sends
  some useful information to a base station.
• Some examples
• Environmental sensor networks to detect and
  monitor environmental changes.
• Wireless surveillance sensor networks for
  providing security in a shopping mall, parking
  garage, etc.
• Military sensor networks to detect enemy
  movements, the presence of hazardous material.

8
Sensor Vs Ad-Hoc Networks

• The number of sensor nodes in a sensor network
  are much more than in an ad-hoc network.
• Sensors act with other sensors in a restricted
  vicinity.
• The topology of a sensor network changes more
  frequently.
• Sensor nodes mainly use a broadcast paradigm,
  whereas most ad-hoc networks are based on a
  point-to-point communication
• Sensors are more constrained in memory and energy
  compared to ad-hoc networks.
• Sensors interact with physical environment, they
  experience Task Dynamics.

9
How do they work?

• Several sensors are deployed in a remote terrain.
• The sensors organize themselves to establish a
  communication network. (Clustering and Routing)
• Divide the task of mapping and monitoring the
  terrain in an energy efficient manner.
• Adapt their overall sensing accuracy according to
  the total remaining resources.
• Re-organize themselves upon sensor failure or
  addition.

10
Our Aim

• Deploy a large number of sensor motes to form a
  network to monitor movements of human beings in a
  terrain.
• The sensor motes communicate and co-ordinate
  efficiently to establish an approximate count of
  the number of human beings in that terrain.

11
Requirements

• Large number of sensors which remain stationary
  after random deployment.
• Low energy usage (Computationally light execution
  and minimal amount of energy transmission)
• Self-organizing network
• Collaborative signal processing (Data
  aggregation)
• Security

12
Human detection

• PIR sensors
• Passive Infrared Sensors detect infra-red heat
  energy emitted by humans. Triggering occurs when
  they detect a change in infrared levels, as and
  when a warm object moves in or out of range of a
  sensor. They are quite resistant to false
  triggering.

13
Salient Features

• A localized clustering algorithm that contributes
  to scalability, robustness and efficient
  utilization of resources.
• A routing algorithm that adapts itself to the
  dynamics of the network and changes in the
  clusters.
• A data aggregation algorithm to fuse data from
  multiple sensors within a cluster.
• Security features that ensure only authorized
  users are granted access to the network and only
  those messages that werent altered in transit
  are accepted.

14
Clustering

• Allows to efficiently co-ordinate local
  information and thus contribute to scalability,
  robustness and efficient utilization of
  resources.
• Each cluster is formed during bootstrap and all
  the members of a cluster elect a cluster head,
  which performs data aggregation and routing for
  the cluster.

15
After PROMOTION TIMER expires, if a sensor
hasnt received any other sensors advertisement
declaring itself as the Cluster Head, then the
sensor promotes itself as the cluster head.
All sensors start advertising presence within a
predefined broadcast range and start WAIT TIMER.
All non-cluster heads choose their cluster head
and form clusters.
Cluster heads
All cluster heads send advertisements to the
sensors from which they received presence
advertisements.
By the end of the WAIT TIME all the sensors
would have received advertisements from the
sensors within its broadcast range. All sensors
then start their PROMOTION TIMER, which is
inversely proportional to the energy level of the
sensor and number of advertisements it received.
16
Routing
Base station
w01
w04
w02
w03
Level 0
w11
w12
Level 1
(W01w11 w02w12) / 2
All the cluster heads receiving this message set
themselves as Level 0 members and assign weight
to themselves based on the received RF signal
strength from the base station and send back
acknowledgements.
Only the clusters not within Level 0 listen to
those messages and again calculate their own
weight based on the received messages signal
strength. These clusters are at Level 1.
Level 0 cluster heads now propagate another
message within a limited range for other cluster
heads.
Base station sends a message within a limited
range so that only nearest cluster heads receive
it.
17
Communication amongst Sensors

• Since all sensors within a single cluster operate
  at the same frequency, provisions have to be made
  such that the signals do not interfere with each
  other .We overcome this problem by using CSMA/CA.
• Similarly even the cluster heads when sending
  information to a lower level or to the base
  station use the same technology

18
Future Work

• Implementation of Security Features
• Data Aggregation of information from various
  sensors.
• Fine tuning of routing and clustering algorithms.
• Parallel processing of Clustering and Routing
  algorithms, the current version works
  sequentially
• Simulation of the algorithms to test their
  validity.

19
Applications

• Location and quantity detection in a Warehouse.
• Disaster detection and recovery which today by
  comparison is very human intensive.

20
QA
21
(No Transcript)