Designing challenges in WSN

 Wireless Sensor Network : WSN is special case of ad-hoc networks with

reduced or no mobility and are known as “Data Centric”. This means, unlike
traditional ad hoc network where data is requested from specific node or location,
data is requested based on sensed attributes.
 The use of particular type of query might depend on the application requirements.
The major factors that need to considered while designing sensor network are listed
below :
 Scalable and flexible architecture : In the sensor network the number of
sensor nodes deployed may be order of hundred, thousands or millions so
that we can easily extend the network size. The communication protocols
must be designed in such a manner that deploying many nodes in the
network does not affect clustering and routing. In other words, the network
must preserve its stability. Introducing more nodes into the network means
that additional communication messages will be exchanged, so that these
nodes are integrated into the existing network.
 Error-prone wireless medium : Since sensor networks can be deployed in
different situations, the requirements of each different application may vary
significantly. We should consider that the wireless medium can be greatly
affected by noisy environments. An attacker interferes knowingly and causes
enough noise to affect the communication.
 Fault tolerance and adaptability : Fault tolerance means to maintain sensor
network functionalities without any interruption due to failure of sensor node
because in sensor network every node have limited power of energy so the
failure of single node doesn’t affect the overall task of the sensor network.
Adaptable protocols can establish new links in case of node failure or link
congestion. Network can able to adapt by changing its connectivity in case
of any fault. In
 that case, well-efficient routing algorithm is applied to change the overall
configuration of network.
 Infrastructure : Sensors network are infrastructure less in which nodes
can communicate directly with base station. It utilizes multi-hop radio
relaying and number of base station depends upon area covered by node
and its radio range.
 Node Deployment : Sensor network can be deployed randomly in
geographical area. After deployment, they can be maintained automatically
without human presence. In sensor network node deployment falls into two
categories either a dense deployment or a sparse deployment. In dense
deployment we have relatively high number of sensor nodes in the targeted
field while in a sparse deployment we have fewer nodes and it is used when
the cost of sensor nodes increases and prohibited the use of dense deployment.
The dense deployment is used when it is important to detect the every moment
or when we have multiple sensors for covering an area.
 Real–time : Achieving Real-time in WSN is difficult to maintain. It must
support maximum bandwidth, minimum delay and several QOS
parameters. This issue can affect time synchronization algorithm.
 Dynamic changes : As in sensor network nodes are deployed without any
topology and they are adaptable to changes due to addition of new nodes or
failure of nodes. Thus, unlike traditional networks, where the goal is to maximize
the channel throughput or minimize the node deployment, but in a sensor
network focus is to extend the system lifetime and the system robustness.
 Power Consumption : Wireless sensor node is microelectronic device means
it is equipped with a limited number of power source. Nodes are dependent on
battery for their power. Hence power conservation and power management is
an important issue in wireless sensor network. Due to this reason researchers
are focusing on the design of power aware protocols and algorithm for
sensors network.
 Production cost : As the name suggests production cost, we know that in the
sensor network we have large no of nodes deployed, so if a single node will
be very high then the cost of over all network will be very high. As a result
the cost of each sensor node has to be kept low. In order to make sensor
network feasible the cost of sensor node should be less. As a result the cost of
sensor node will be a very challenging issue.
 Short Range Transmission : In WSNs we should consider the short
transmission range in order to reduce the possibility of being eavesdropped.
 As in long range transmission we need high transmission power due to the
point to point transmission between the nodes to reach the destination which
increases the chance of being eavesdropped.
 Hardware design : While designing any hardware of sensor network, it
should be energy- efficient. Hardware such as microcontroller, power
control, and communication unit should be design to consume less energy.
 Limited computational power and memory size : It is another factor that
affects WSN in the sense that each node stores the data individually and
sometime more than one node stored same data and transferred to the base
station which waste the power and storing capacity of nodes so we must
develop effective routing schemes and protocols to minimize the redundancy
in the network.
 Operating environment : Sensor nodes are deployed densely either very
closed or inside the phenomenon which is to be observed. These nodes may
work under-busy interaction, at the bottom of an ocean, in the interior of a
large machinery, on the surface of an ocean during a tornado, in a home or
large building and in a large warehouses.
 Simplicity : Simplicity is an important point in the wireless sensor network
since sensor nodes are small and there is restriction on the utilization of energy
as they are energy dependent so the
 computing and communicating software used in the nodes should be
computation efficient and less in size than the traditional software in the
network.
 Quality of Service : It means data should be delivered within time period.
Some real time sensor applications are based on time means if data should not
be delivered on time from the moment it is sensed; the data will become
unusable for e.g. fire detection requires good quality of services.
 Security : Security is very important parameter in sensor network since
sensor networks are data centric so there is no particular id associated with
sensor nodes and attacker can easily inserted himself into the network and
stole the important data by becoming the part of network without the
knowledge of sensor nodes of the network. So it is difficult to identify
whether the information is authenticated or not.
Comparison between MANET, WSN AND IOT
Sr. Wireless Sensor
Feature Ad hoc Network IOT
No. Network
Number of sensor
1. Large in quantity Medium in quantity Less in quantity
nodes or motes
2. Deployment type Very much dense Scattered Scattered
3. Rate of failure Prone to failures Very rare Less
Change in network
4. Dynamically *Dynamically Less
topology
 Point to Point
(P2P)
5. Communicat ion Peer to Peer Peer to peer and end to
 Star
mode end
 Mesh
 Hybrid [24]
Not replaceable/ not
6. Battery Replaceable Replaceable
rechargeable
Identifiers (IDs) used
7. No unique IDs Unique IDs Unique IDs
in the network
8. Centric mode Data centric Address centric Both
9. Fusion/Aggregat ion Possible Not suitable Possible
Computational
10. capacities & memory Limited Not limited Limited
requirement
Data rate support
11. Lower Higher Moderate
provided
12. Redundancy High Low Low
ZigBee, IEEE 802.15.4, ZigBee, RFID, Bluetooth
13. Standards IEEE 802.11
ISA100, IEEE 1451 and BACnet
 Absence of access to
Needed only if nodes the running code,
Needed as mobility
14. Fault tolerance exhaust available
increases Denial-of-service for
energy or are moved heterogeneous
hardware [30]
15. Commu nicat io n Short Long Long
 Range
Close to humans
Focus on interaction e.g. laptops, PDAs, Machine-to Machine
16. Interaction
with the environment mobile radio Interaction
terminals

Comparison of WSN and MANET :

Sr.
Parameters Wireless Sensor Networks Ad Hoc Networks
No.
1. Number of sensor nodes Large Medium
2. Deployment Densely deployed Scattered
3. Failure rate Prone to failures Very rare
4. Topology Dynamic Dynamic
5. Commu nicat io n paradigm Peer to Peer Peer to peer and end to
end
6. Battery Not replaceable /not Replaceable
rechargeable
7. Identifiers No unique identifiers Unique identifiers
8. Centric Data centric Address centric
9. Fusion / aggregation Possible Not suitable
10. Computational capacities and Limited Not limited
memory
11. Data rate Low High
12. Redundancy High Low
13. Routing protocols Flooding, Gossiping, Flat Routing, Pro-active, Reactive,
Hierarchical, Location based Hybrid

14. Standards ZigBee, IEEE 802.15.4, ISA100, IEEE 802.11

IEEE 1451
15. Fault tolerance Needed only if nodes exhaust Needed as mobility
available energy or are moved increases
16. Communicat ion Range Short Long
Classification of clustering algorithms

 There have been several different methods to classify the algorithms used for
WSNs.
 Clustering. Four of the most common classifications are shown in Figure.
Heterogeneous Algorithms
Homogeneous Algorithms Centralized
Algorithms
Clustering Distributed Algorithms
Algorithms
Static Algorithms Dynamic
Algorithms Probabilistic
Algorithms
Non-Probabilistic Algorithms

Common Classification Algorithm

 Clustering Algorithms Homogeneous or Heterogeneous Networks : This
classification according to the characteristics and performance of sensor nodes in
a cluster. In heterogeneous sensor networks, all nodes have the same
specifications, hardware and processing capabilities. In these networks, which
are common in nowadays applications, each node can be a CH. In addition to
these networks, the CH role can be replaced between the nodes periodically (for
the creation of better and more integrated load balancing energy). Against in
heterogeneous sensor networks, generally, there are two types of sensor, the first
type sensors with more processing abilities and complex hardware. These
sensors predetermined as a CH node. The other type conventional sensors, with
lower abilities, which in fact, used to sense the environment properties.
 Centralized or Distributed Clustering Algorithms : These clustering
algorithms imply on the method utilized for shaping clusters. A distributed CH
selection and shaping process are the most suitable way to obtain enhanced
flexibility and faster convergence times independent of the number of nodes of
the WSN. This approach is the most efficient method, particularly for large
networks. Also, there are a few approaches using centralized or hybrid method,
which one or more coordinate nodes or base stations (sink), responsible for the
break up into detached and control of all network cluster members. These
networks are not appropriate for overall objective large-scale practical
application WSNs. They may be appropriate just for specific targets bounded-
scale applications in which high-quality connectivity and network separation is
needed.
 Static and Dynamic Clustering Algorithms : Other conventional classification
is static or dynamic clustering. Process shaping clusters is dynamic (otherwise
 as static) when it contains regular (periodic or event-oriented) CH re-election or
includes cluster reorganization routine, these procedures may be effective in
order to respond to changes in network topology and only accurately the cluster
topology or proper movement with the purpose CH role between the nodes to
obtain in energy saving. Dynamic cluster architectures make a better use of the
sensors in a WSN and naturally result in improved energy consumption
management and network lifetime.
 Probabilistic and Non-Probabilistic Clustering Algorithms : This classification
based on cluster shape parameters can be used to select the CH. These clustering
algorithms are divided into two categories : probabilistic (random or hybrid) and
non-probability (deterministic). Most clustering algorithms are known, they can
be divided into two main categories. In the probabilistic clustering
algorithm for determining the initial CH, a probability assigned to each
node. Probabilistic clustering algorithms, beyond the more energy-
efficient, often running faster at the convergence and reduce the volume
of messages they exchange. In the non-probabilistic clustering
algorithms basically criteria (deterministic) more specific for CH
selection and cluster formation, take into consideration. These criteria
are essentially based on the proximity of adjacent nodes like
(connectivity, degree, etc. ) and information received from other closely
located nodes.