Anda di halaman 1dari 28

Embedded Systems for Wireless Sensor


Rabi Mahapatra

• Advancement of integration between “tiny embedded

processors, wireless interfaces, and “micro-sensors” based
on MEMS led to emergence of wireless sensor network.

• Characterized by their ability to monitor the physical

environment through ad-hoc deployment of numerous tiny,
intelligent, wirelessly networked sensor nodes.

Mahapatra-Texas A&M-Spring'07 2
What is Wireless Sensor Networks

• Large number of heterogeneous sensor devices

– Ad Hoc Network
• complex sensor nodes with
– communication, processing, storage capabilities

Mahapatra-Texas A&M-Spring'07 3
Emerging applications

• Indoor Settings: condition based maintenance of

equipment in factory
• Outdoor environment:
– Monitor natural habitats
– Remote ecosystems
– Forest fires
– Disaster sites
– Defense armaments
– Spy microsats

Mahapatra-Texas A&M-Spring'07 4
Challenges of WSN
• Requirements: small size, large number, tetherless and low
cost. Hence constrained by
– Energy, computation and communication
• Small form factors => prohibits large long lasting batteries
• Cost & energy => low power processors, small radios with
minimum bandwidth & small transmission ranges.
• Ad-hoc deployment => no maintenance and battery
• Increase NW lifetime => No raw data to gateway for

Mahapatra-Texas A&M-Spring'07 5
Topics to be discussed
• Simulation tools on WSN
• A simulation architecture overview
• Sensor node model & framework of SN
• Battery model
• Case studies
• Bonus points

Mahapatra-Texas A&M-Spring'07 6
Existing Simulators
• Pros
– Very modular
– Easy to use
• Cons
– Geared for wired inter-networks
– No wireless support, not efficient due to overhead
SSFNet: a parallel simulator for wireless
• Specific for mobile wireless networks.
• Built as a set of libraries. The libraries are built in Parsec( a C-based
discrete event simulation language).
• Layered architecture with easy plug-in capability.

SSFNet and Glomosim are not better than NS-2 in terms of design and

Mahapatra-Texas A&M-Spring'07 7
Existing Simulators
• NS-2: De facto standard for network simulations
– Does support wireless simulations
– A primitive energy model is present.
– Object oriented design and Lots of documentation.
– Uses Tcl to specify the Components, and Otcl to glue them
– Difficult to use and learn
– Interdependency among modules pose difficult to implement new
– Originally built for wired networks, later extended for wireless.
– Supposedly, does not work well for large topologies.

Mahapatra-Texas A&M-Spring'07 8
More Sensor Network Simulator
– Based on Low Energy Adaptive Clustering Hierarchy
(LEACH) protocol developed by Dr. Wendi
– Has included Network Preserving Protocol
(NPP) for better performance along with
– Not completed for robustness

Mahapatra-Texas A&M-Spring'07 9
Sensor Network Simulator
• Component Features: (2004)
– Battery Model: Linear Battery, Discharge Rate
Dependent and/or Relaxation Battery
– Application Layer : Random Neighbor; Constant Bit
– Network Layer: Simple Flooding; a simplified verion of
ADOV without route repairing, a simplified version of
DSR without route repairing
– MAC Layer: NullMAC; IEEE 802.11
– Physical Layer: Duplex Transceiver; Wireless Channel
– Simulation Engine: CostSimEng (sequential)
Mahapatra-Texas A&M-Spring'07 10
Status of simulators

• Other simulator: OpNET

• All these tools are not equipped to capture all the aspects
of interests in sensor networks.

Mahapatra-Texas A&M-Spring'07 11
Simulator: SensorSim from UCLA

• Extension to NS - 2.
• Provides battery models, radio propagation models and
sensor channel models.
• Provides a lightweight protocol stack.
• Has support for hybrid simulation.
• Must be integrated with NS - 2.

Mahapatra-Texas A&M-Spring'07 12
SensorSim Architecture
monitor and control
hybrid network
real sensor apps on
(local or remote)
virtual sensor nodes

GUI app
socket serial
comm comm

GUI Interface HS Interface ns RS232

V R Gateway Machine

modified event scheduler

Proxies for real
Mahapatra-Texas A&M-Spring'07
sensor nodes 13
Simulation Machine
SensorSim Architecture Overview
• Sensor NW has three types of nodes:
– Sensor nodes: monitor immediate environment, with many
– Target nodes: generates various stimuli for sensor nodes
– User nodes: client and administration of sensor network
• Separate channels:
– Sensor channels: communication among sensor nodes and target
– Network channels: to user node or gateways and onward
transmission to other network.
– Concurrent transmission possible
– Easier to model complex behavior of sensor nodes, reaction to
multiple sensor signals.

Mahapatra-Texas A&M-Spring'07 14
Sensor Network Model architecture

Sensor channel

sensor sensor sensor

Wireless channel


Mahapatra-Texas A&M-Spring'07 15
SensorSim Model
• Sensor node => one wireless NW protocol stack, one or
more sensor stack corresponds to as many transducers
– Sensor stack detects stimuli, process it and forward them to
application layer, which in turn process and send them to user node
through wireless channel
– A power model corresponding to energy producing-consuming
hardware components is also provided. These component can stay
at different power saving and performance states.
– The algorithm in both the stacks control the mode of power states
of hardware components. Also, performance of the algorithm
depends on the mode.

Mahapatra-Texas A&M-Spring'07 16
Sensor Node Model in SensorSim

Node Function Model

Sensor Node

Middleware Power Model

State (Energy Consumers and Providers)
Sensor Stack 1 Change
Network Sensor Radio Model
Protocol Stack Protocol Stack
Sensor Layer
Network Layer Sensor Layer CPU Model
Physical Layer Status Model
MAC Layer Physical Layer Check Sensor #1 Model

Physical Layer
Sensor #2 Model

Wireless Channel Sensor Channel 1

Sensor Channel 2
Mahapatra-Texas A&M-Spring'07 17
User Node
Target Node

User Application
Target Application

Network Sensor Stack

Protocol Stack
Network Layer Sensor Layer

MAC Layer Physical Layer

Physical Layer

Sensor channel
Wireless Channel

Mahapatra-Texas A&M-Spring'07 18
Framework of Sensor Network Simulation
• Node Placement & traffic generation
– Performance of WSN is affected when topology of node
distribution changes
– Application requires a typical distribution (uniform for forest fire,
Gaussian for perimeter defense)
– Three types of traffics: user-to-sensor (command & queries),
sensor-to-user (sensor reporting to user) and sensor-to-sensor
(collaborative signal processing before reporting)

Mahapatra-Texas A&M-Spring'07 19
Sensor Stack & Channel
• Sensor stack is a signal sink that is responsible for
triggering the application layer every time a sensing event
• Simple sensing scheme to elaborate signal processing can
be implemented on sensor stack
• Sensor stack acts as a signal source in Target Node and
contains signatures unique to the model
• Sensor channel model the medium of signal transmission
(e.g. ground to carry seismic events).
• A good simulation tool should model varieties of mediums
and type of sensors ( acoustic, infra red, ultrasonic)

Mahapatra-Texas A&M-Spring'07 20
Battery Model
• Goal: increase the battery life time
• Need to study how different aspects of real battery
behavior can affect the energy efficiency of applications
• T = C/I, C is capacity in Ah. I is discharge current
• Linear Model:
– Linear storage of current. Assumes the maximum capacity is
unaffected by discharge rate.
– Allows user to see efficiency of user application by providing how
much capacity is consumed. The remaining capacity after td can be
expressed as C = C’ - I(t)dt integral taken over period t = 0 to td
– It assumes that the I(t) will stay same during the period, if
operation mode does not change (radio switching from Tx toRx)
– Remaining capacity is computed when discharge rate is changed

Mahapatra-Texas A&M-Spring'07 21
Battery Model
• Discharge rate dependent model:
– Considers the effect of battery discharge rate on maximum
– Battery capacity efficiency factor K is introduced. K = Ceff /Cmax
– Capacity C = K.C’ – I . Td
– K varies with current I and is close to 1 when discharge rate is low
and approaches 0 when discharge rate is high.

A Popular Battery Model: Dual Foil from UC Berkeley.

Mahapatra-Texas A&M-Spring'07 22
Battery Model
• Relaxation Model:
– Real-life battery exhibit a phenomenon called “relaxation”.(Fuller
94, Linden 95, Chiasserini 99)
– When battery discharge rate is high, diffusion rate of active
ingredients through the electrode & electrolyte falls behind. If high
discharge is sustained, the battery reaches its end even if the active
materials are still available.
– However, if discharge current is either cut-off or reduced during
the discharge, active materials catches up with depletion of the
materials. It gives battery to recover the capacity lost at high
discharge rate.
– An analytical model has been used for SensorSim

Mahapatra-Texas A&M-Spring'07 23
Case Studies
• Low rate/low power vs. high rate/ high power (Fig. 7 in the

• Monitoring a moving vehicle in a sensor field

– Study the effect of traffic on the sensing and communication traffic
and evaluate the power management.

Mahapatra-Texas A&M-Spring'07 24
Power Management Model

BZR event
BZR BZR event
BZR Off event
Transmit transmit BZR event Receive
Transmit BZR Receive transmit
event done receive
transmit Idle
transmit receive
Idle done timeout receive
done Sleep timeout(3 sec)

Without Power Management With Power Management

Mahapatra-Texas A&M-Spring'07 25
Some Important Studies

• Utility-based decision-making in WSN.

• Upper bound on network life-time
• Impact of mobility on capacity and life-time
• Coverage and Density
– Criticality threshold, scalability, integration etc.
• Security
• Ease of Deployment
• Synchronization

Mahapatra-Texas A&M-Spring'07 26

We have looked at some of the issues with SN and discussed

the Sensor Network model developed at UCLA.
• Assignments:
– Read the reference papers and look for more titles
– Prepare a bibliography on each topic mentioned in the previous
– Consider one topic as assigned to you in the class and read apex
papers on that topic.
– Summarize and comment on the contributions and shortcomings.
– Due Tuesday morning by e-mail with file name as (your name-

Mahapatra-Texas A&M-Spring'07 27
• Sung Park, A savvides & M B Srivastava, “Simulating
Networks of Wireless Sensors”, Proceedings of the 2001
Winter Simulation Conference.

Mahapatra-Texas A&M-Spring'07 28