WIRELESS SENSOR NETWORKS UNIT I

OVERVIEW OF WIRELESS SENSOR NETWORKS

KEY DEFINITIONS OF SENSOR NETWORKS
Sensor networks is an interdisciplinary research area that draws on contributions from signal
processing, networking and protocols, databases and information management, distributed
algorithms, and embedded systems and architecture. In the following, we define a number of key
terms and concepts that will be used throughout the book as we develop techniques and
examples for sensor networks.
Sensor: A transducer that converts a physical phenomenon such as heat, light, sound, or
motion into electrical or other signals that may be further manipulated by other apparatus.
Sensor node: A basic unit in a sensor network, with on-board sensors, processor,
memory, wireless modem, and power supply. It is often abbreviated as node. hen a
node has only a single sensor on board, the node is sometimes also referred to as a
sensor, creating some confusion.
Network topology: A connectivity graph where nodes are sensor nodes and edges are
communication links. In a wireless network, the link represents a one-hop connection,
and the neighbors of a node are those within the radio range of the node.
Routing: !he process of determining a network path from a packet source node to its
destination.
Date-centric: Approaches that name, route, or access a piece of data via properties, such
as physical location, that are external to a communication network. !his is to be
contrasted with address-centric approaches which use logical properties of nodes related
to the network structure.
Geographic routing: "outing of data based on geographical attributes such as locations
or regions. !his is an example of date-centric networking.
In-network: A style of processing in which the data is processed and combined near
where the data is generated.
Collaborative processing: Sensors cooperatively processing data from multiple sources
in order to serve a high-level task. !his typically requires communication among a set of
nodes.
State: A snapshot about a physical environment #e.g., the number of signal sources, their
locations or spatial extent, speed of movement$, or a snapshot of the system itself #e.g.,
the network state$.
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae !
WIRELESS SENSOR NETWORKS UNIT I
Uncertainty: A condition of the information caused by noise in sensor measurements, or
lack of knowledge in models. !he uncertainty affects the system%s ability to estimate the
state accurately and must be carefully modeled. &ecause of the ubiquity of uncertainty in
the data, many sensor network estimation problems are cast in a statistical framework.
'or example, one may use a covariance matrix to characteri(e the uncertainty in a
)aussian-like process or more general probability distributions for non-)aussian
processes.
ask: *ither high-level system tasks which may include sensing, communication,
processing, and resource allocation, or application tasks which may include detection,
classification, locali(ation, or tracking.
Detection: !he process of discovering the existence of a physical phenomenon. A
threshold-based detector may flag a detection whenever the signature of a physical
phenomenon is determined to be significant enough compared with the threshold.
Classi!ication: !he assignment of class labels to a set of physical phenomena being
observed.
"ocali#ation and tracking: !he estimation of the state of a physical entity such as a
physical phenomenon or a sensor node from a set of measurements. !racking produces a
series of estimates over time.
$alue o! in!or%ation or in!or%ation utility: A mapping of data to a scalar number, in the
context of the overall system task and knowledge. 'or example, information utility of a
piece of sensor data may be characteri(ed by its relevance to an estimation task at hand
and computed by a mutual information function.
Resource: "esources include sensors, communication links, processors, on-board
memory, and node energy reserves. "esource allocation assigns resources to tasks,
typically optimi(ing some performance ob+ective.
Sensor tasking: !he assignment of sensors to a particular task and the control of sensor
state #e.g., on,off, pan,tilt$ for accomplishing the task.
Node services: Services such as time synchroni(ation and node locali(ation that enable
applications to discover properties of a node and the nodes to organi(e themselves into a
useful network.
Data storage: Sensor information is stored, indexed, and accessed by applications.
Storage may be local to the node where the data is generated, load-balanced across a
network, or anchored at a few points #warehouses$.
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae "
WIRELESS SENSOR NETWORKS UNIT I
&%bedded operating syste% '(S): !he run-time system support for sensor network
applications. An embedded -S typically provides an abstraction of system resources and
a set of utilities.
Syste% per!or%ance goal: !he abstract characteri(ation of system properties. *xamples
include scalability, robustness, and network longevity, each of which may be measured by
a set of evaluation metrics.
&valuation %etric: A measurable quantity that describes how well the system is
performing on some absolute scale. *xamples include packet loss #system$, network
dwell time #system$, track loss #application$, false alarm rate #application$, probability of
correct association #application$, location error #application$, or processing latency
#application,system$. An evaluation method is a process for comparing the value of
applying the metrics on an experimental system with that of some other benchmark
system.
ADVANTAGES OF SENSOR NETWORKS
.etworked sensing offers unique advantages over traditional centrali(ed approaches. /ense
networks of distributed communicating sensors can improve signal-to-noise ratio #S."$ by
reducing average distances from sensor to source of signal, or target. Increased energy efficiency
in communications is enabled by the multihop topology of the network. 0oreover, additional
relevant information from other sensors can be aggregated during this multihop transmission
through in-network processing. &ut perhaps the greatest advantages of networked sensing are in
improved robustness and scalability. A decentrali(ed sensing system is inherently more robust
against individual sensor node or link failures, because of redundancy in the network.
/ecentrali(ed algorithms are also far more scalable in practical deployment and may be the only
way to achieve the large scales needed for some applications.
*nergy Advantage
&ecause of the unique attenuation characteristics of radio-frequency #"'$ signals, a multihop "'
network provides a significant energy saving over a single-hop network for the same distance.
1onsider the following simple example of an N-hop network. Assume the overall distance for
transmission is .r, where r is the one-hop distance. !he minimum receiving power at a node for
a given transmission error rate is P
receive
, and the power at a transmission node is P
send
. !hen,
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae #
WIRELESS SENSOR NETWORKS UNIT I
the "' attenuation model near the ground is given by
here r is the transmission distance and is the "' attenuation exponent. /ue to multipath and
other interference effects, is typically in the range of 2 to 3. *quivalently,
!herefore, the power advantage of an N-hop transmission versus a single-hop transmission over
the same distance Nr is
'igure 4.4 illustrates the power attenuation for the multihop and single-hop networks. A
larger N gives a larger power saving due to the consideration of "' energy alone. 5owever, this
analysis ignores the power usage by other components of an "' circuitry. 6sing more nodes
increases not only the cost, but also the power consumption of these other "' components. In
practice, an optimal design seeks to balance the two conflicting factors for an overall cost and
energy efficiency. 7atency and robustness considerations may also argue against an unduly large
number of relay nodes.
Figure 1.1 !he power advantage of using a multihop "' communication over a distance
of Nr.
Detection Advntge
*ach sensor has a finite sensing range, determined by the noise floor at the sensor. A denser
sensor field improves the odds of detecting a signal source within the range. -nce a signal source
is inside the sensing range of a sensor, further increasing the sensor density decreases the average
distance from a sensor to the signal source, hence improving the signal-to-noise ratio #S."$. 7et
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae $
WIRELESS SENSOR NETWORKS UNIT I
us consider the acoustic sensing case in a two-dimensional plane, where the acoustic power
received at a distance r is
hich assumes an inverse distance squared attenuation. !he S." is given by
Increasing the sensor density by a factor of k reduces the average distance to a target by a factor
of

4
K
,
!hus, the S." advantage of the denser sensor network is
!herefore, an increase in sensor density by a factor of k improves the S." at a sensor by 48 log
k db.
!NI"!E #ONSTRAINTS AND #$ALLENGES
6nlike a centrali(ed system, a sensor network is sub+ect to a unique set of resource constraints
such as finite on-board battery power and limited network communication bandwidth. In a
typical sensor network, each sensor node operates untethered and has a microprocessor and a
small amount of memory for signal processing and task scheduling. *ach node is also equipped
with one or more sensing devices such as acoustic microphone arrays, video or still cameras,
infrared #I"$, seismic, or magnetic sensors. *ach sensor node communicates wirelessly with a
few other local nodes within its radio communication range.
Sensor networks extend the existing Internet deep into the physical environment. !he
resulting new network is orders of magnitude more expansive and dynamic than the current
!19,I9 network and is creating entirely new types of traffic that are quite different from what
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae %
WIRELESS SENSOR NETWORKS UNIT I
one finds on the Internet now. Information collected by and transmitted on a sensor network
describes conditions of physical environments:for example, temperature, humidity, or vibration
: and requires advanced query interfaces and search engines to effectively support user-level
functions. Sensor networks may internetwork with an I9 core network via a number of gateways,
as in 'igure 4.2. A gateway routes user queries or commands to appropriate nodes in a sensor
network. It also routes sensor data, at times aggregated and summari(ed, to users who have
requested it or are expected to utili(e the information. A data repository or storage service may
be present at the gateway, in addition to data logging at each sensor.
Figure 1.% Sensor networks significantly expand the existing Internet into physical spaces. !he
data processing, storage, transport, querying, as well as the internetworking between the !19,I9
and sensor networks present a number of interesting research challenges that must be addressed
from a multidisciplinary, cross-layer perspective.
!he repository may serve as an intermediary between users and sensors, providing a persistent
data storage. Additionally, one or more data storage devices may be attached to the I9 network,
to archive sensor data from a number of edge sensor networks and to support a variety of user-
initiated browsing and search functions.
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae &
WIRELESS SENSOR NETWORKS UNIT I
!he current generation of wireless sensor hardware ranges from shoe-box-si(ed Sensoria
I.S .) sensors with an S5-; microprocessor to matchbox-si(ed &erkeley motes with an <-bit
microcontroller. A few samples of sensor hardware are shown in 'igure 4.=> their corresponding
capabilities are summari(ed and compared in !able 4.4. It is well known that communicating 4
bit over the wireless medium at short ranges consumes far more energy than processing that bit.
'or the Sensoria sensors and &erkeley motes, the ratio of energy consumption for
communication and computation is in the range of 4888 to 48,888. /espite the advances in
silicon fabrication technologies, wireless communication will continue to dominate the energy
consumption of networked embedded systems for the foreseeable future.
Figure 1.& Samples of wireless sensor hardware? #a$ Sensoria I.S .) 2.8 sensor node> #b$ 59
i9A@ with <82.44b and microphone> #c$ &erkeley,1rossbow sensor mote, alongside a 6.S.
penny> #d$ An early prototype of Smart /ust 0*0S integrated sensor, being developed at 61
&erkeley. #Picture courtesy of Kris Pister and Jason Hill$.
!hus, minimi(ing the amount and range of communication as much as possible:for
example, through local collaboration among sensors, duplicate data suppression, or invoking
only the nodes that are relevant to a given task:can significantly prolong the life of a sensor
network and leave nodes free to support multiuser operations. In addition, the shorter "'
transmission range improves spectrum usage and increases throughput for a sensor network.
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae '
WIRELESS SENSOR NETWORKS UNIT I
!he information management and networking for this new network will require more
than +ust building faster routers, switchers, and browsers. A sensor network is designed to collect
information from a physical environment. .etworking will be intimately coupled with the needs
of sensing and control, and hence the application semantics. !o optimi(e for performance and
resources such as energy, one has to rethink the existing !19,I9 stack and design an appropriate
sensor network abstraction to support application development. 'or example, in many
applications, it is more appropriate to address nodes in a sensor network by physical properties,
such as node locations or proximity, than by I9 addresses. 5ow and where data is generated by
sensors and consumed by users will affect the way data is compressed, routed, and aggregated.
&ecause of the peer-to-peer connectivity and the lack of a global infrastructure support, the
sensors have to rely on discovery protocols to construct local models about the network and
environment. 0obility and instability in wireless links preclude the use of many existing edge-
network gateway protocols for internetworking I9 and sensor networks.
!able 4.4 1omparison of the four sensor platforms shown in figure 4.=
!o summari(e, the challenges we face in designing sensor network systems and applications
include?
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae (
WIRELESS SENSOR NETWORKS UNIT I
Limited hardware: *ach node has limited processing, storage, and communication
capabilities, and limited energy supply and bandwidth.
Limited support for networking: !he network is peer-to-peer, with a mesh topology and
dynamic, mobile, and unreliable connectivity. !here are no universal routing protocols or
central registry services. *ach node acts both as a router and as an application host.
Limited support for software development: !he tasks are typically real-time and
massively distributed, involve dynamic collaboration among nodes, and must handle
multiple competing events. )lobal properties can be specified only via local instructions.
&ecause of the coupling between applications and system layers, the software
architecture must be codesigned with the information processing architecture.
DRIVING A''LI#ATIONS
!he applications of S.s are innumerable, since each sensor node is capable of monitoring a
wide variety of ambient conditions such as temperature, humidity, lightning condition, pressure
and noise levels. &elow, we have given some typical application areas for S.s.
1. (i)itr* ++)iction,- Sensor nodes can be spread across a battlefield or enemy area and
be programmed to track and monitor enemy troop movements or movement of terrorists
and can be used to locate their exact positions. !hey can also be used instead of mines to
immediately detect any movement, thus safeguarding human lives.
Sensor networks can also be used to detect the use of biological or chemical weapons,
and relay this information to commanders, so as to allow sufficient time for soldiers to
take defensive measures in the field.
Improved battlefield communication is another benefit of S. for military applications.
5ere, the soldier with a personal digital assistant #9/A$ acquires an extended sense by
interacting with the surrounding S.. !arget field imaging, security and tactical
surveillance, and intrusion detection are other similar applications.
&ut military applications impose several special requirements on the S.. 'irstly, auto-
deployment and self-organi(ation should be supported. Secondly, detection of a sensor
node should be difficult, otherwise an adversary may determine the location of sensor
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae )
WIRELESS SENSOR NETWORKS UNIT I
nodes easily, and compromise them. In such a situation, information with the
compromised node can be stolen, or invalid information can be in+ected into the system.
!hus, secure communication is vital but may not be feasible in the form of large keys and
robust protocols because of the limited computation power available. !his is a ma+or
challenge in the deployment of S. for military applications.
2. Environ.ent) ++)iction,- Sensor networks are being increasingly used for
environmental concerns. *xamples include tracking the nesting habits of seabirds by
monitoring a large geographic region with human presence, or attaching the sensors
directly to large mammals to monitor their behavior.
0onitoring of river currents is another application of S.s, to measure their water
inflow and mixture from various sources. ater-quality monitoring may also be useful to
determine contamination with bacteria or other harmful pollutants.
A ma+or application in this category is the spread of sensor nodes across a forest to
monitor temperatures and give early warnings of fire outbreaks. eather prediction,
climate monitoring, distributed computing, pollution tracking, seismic detection,
detecting ambient conditions such as temperature, movement, sound, light, or the
presence of certain ob+ects, inventory control and disaster management are other similar
applications.
!he ma+or requirement of environment sensors is the need for rugged operation in hostile
surroundings and extended sleep periods to maximi(e the lifetime of the network.
3. (edic) ++)iction,- S.s can be used in medical applications by using the sensor
node as a device that can reside on or within the human body and perform tasks that are
currently done by costly machines. !hese include glucose monitors for continuous
reading of insulin levels in diabetic patients> heart monitors for keeping track of the
functioning of the heart, especially for patients with irregular heartbeats or coronary
diseases> and artificial retinal and cortical implants to electronically transmit information
to visually impaired persons.
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae !*
WIRELESS SENSOR NETWORKS UNIT I
Another example of the use of S.s in medical applications is the vital statistics
repository, which takes the form of a smart card holding medical information on persons
with severe allergies to certain substances or medications.
-ther medical applications include use in telemonitoring of human physiological data,
tracking and monitoring doctors and patients inside a hospital and insurance cards.
0edical applications using sensors also have special requirements. !hey must be safe and
biocompatible to merit continuous functioning inside the human body and not damage the
tissues. !he sensors must be designed for long-term operation and have enough power so
that frequent surgery is avoided. !his means that they must be highly fault-tolerant and
provide redundancy and graceful degradation in failure scenarios.
4. Indu,tri) ++)iction,- 'or use in industrial applications, low-cost sensor nodes could
be attached to equipment to monitor performance. !hey could also be attached to parts as
they move through an assembly pipeline on the shop floor. !hus, inefficiencies in plant
process flow can be recogni(ed quickly, rush orders could be expedited more easily and
customer queries could be answered faster and more accurately.
"adio frequency identity #"'I/$ tags are already being placed on items like gasoline and
other merchandise to allow fast and accurate scanning at checkout and for inventory
tracking. !hese tags could be replaced by wireless sensors at fixed locations and used for
tracking.
!he use of sensor nodes for industrial and commercial applications requires their cost to
be made very low so that they can be used in bulk. !he protocols in use should, therefore,
also be highly scalable.
/ !r0n ++)iction,- S.s can be used for various urban applications like transportation
and traffic systems, auto-identification by driving license, parking availability, security
monitors in shopping malls, parking garages, city streets and home security.
ENA1LING TE#$NOLOGIES FOR WIRELESS SENSOR NETWORKS
&uilding such wireless sensor networks has only become possible with some fundamental
advances in enabling technologies. 'irst and foremost among these technologies is the
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae !!
WIRELESS SENSOR NETWORKS UNIT I
miniaturi(ation of hardware. Smaller feature si(es in chips have driven down the power
consumption of the basic components of a sensor node to a level that the constructions of S.s
can be contemplated. !his is particularly relevant to microcontrollers and memory chips as such,
but also, the radio modems, responsible for wireless communication, have become much more
energy efficient. "educed chip si(e and improved energy efficiency is accompanied by reduced
cost, which is necessary to make redundant deployment of nodes affordable.
!hese three basic parts of a sensor node have to accompanied by power supply. !his
requires, depending on application, high capacity batteries that last for long times, that is, have
only a negligible self-discharge rate, and that can efficiently provide small amounts of current.
Ideally, a sensor node also has a device for energ* ,cvenging, recharging the battery with
energy gathered from the environment A solar cells or vibration-based power generation are
conceivable options. Such a concept requires the battery to be efficiently chargeable with small
amounts of current, which is not a standard ability. &oth batteries and energy scavenging are still
ob+ects of ongoing research.
!he counterpart to the basic hardware technologies is software. !he first question to
answer here is the principal division of tasks and functionalities in a single node A the
architecture of the operating system or runtime environment. !his environment has to support
simple retasking, cross-layer information exchange, and modularity to allow for simple
maintenance. !his software architecture on a single node has to be extended to a network
architecture, where the division of tasks between nodes, not only on a single node, becomes the
relevant question A for example, how to structure interfaces for application programmers. !he
third part to solve then is the question of how to design appropriate communication protocols.
QISCET ECE Dept V. JAI KUMAR
Assoc Professor
Pae !"