Effects of mobility models and nodes distribution on wireless sensors networks

International Journal of Ad hoc, Sensor & Ubiquitous Computing (IJASUC) Vol.5, No.6, December 2014
DOI : 10.5121/ijasuc.2014.5601 1
EFFECTS OF MOBILITY MODELS AND NODES
DISTRIBUTION ON WIRELESS SENSORS
NETWORKS
Ali A.J.Al-Sabbagh1,3
, Ruaa A.S.Alsabah1,2
and Ahmed.Y.Mjhool1,4
1
Electrical and Computer Engineering, Florida Institute of Technology, Florida, USA
2
Department of Computer Science, University of Karbala, Iraq
3
Department of Babylon, SCIS, Ministry of Communication, Iraq
4
University of Kufa, Najaf, Iraq
ABSTRACT
Wireless sensor networks (WSN) is an important future technology, in several applications in military,
health, environment and industries. Currently the integration of social and sensor is very important by
considering the characteristics of social networks in designing wireless sensor networks WSN for
improvement such as (number of messages from source to destination, radius of coverage, connectivity, and
spreading). This area has not received much attention and few researches focus on the performance
evaluation. In this paper we have studied the impact of different mobility and distribution models which is a
variable one should define which model is best for the infrastructure given their differences, also study
include the exact effect of nodes distribution and analyzed by calculation the number of messages of 12
cases to get a real performance evaluation under different conditions and same routing techniques. This
work provides us a greater understanding and clear an idea of the effect of mobility plus distribution.
KEYWORDS
Wireless Sensor Networks, Social networks, Mobility Models, Node Distribution, Performance Analysis.
1. INTRODUCTION
Wireless Sensor Networks (WSNs) have become demand research area because significantly in
touch with several fields in life, WSN are self-configuring, networks containing of dynamic /
static sensor nodes in wireless connection to form an random topology. A typical WSN consists
of many small devices deployed over a geographical area, where each device is called a Sensor,
that can measure or sense an environmental or physical conditions (e.g., Pressure, recognition and
humidity) in a particular area. [1] Data forwarding (routing) in WSNs is the process of choosing
the best strategies for sending data the source to the final destination. WSN are not currently
deployed on a large scale, this research area is mostly simulation founded. Specifically mobile
nodes: Mobile wireless sensor networks owe its name as a mobile sink. An advantage of dynamic
WSN with respect to static WSN are well energy efficiency that’s make forwarding a sense data
from source to destination in first priority , improved coverage and improve target tracking and
larger channel capacity.[2,3] A social network is a description of the relations between a set of
actors. The integration of social networks can help to get better performance and understand the
relations among people and the behavior of actors in a network. Integration WSN with (Social
Network of Sensors) where this aspect comes from the movement of nodes with sensing

2
capability such as phones, navigation systems, and other smart objects brought around by
humans, vehicles, or even animals. A mobility model in a dynamic network describes the
movement of mobile nodes and how their positions, directions, and velocity variation by time [4].
In the context of social movement of humans, we need models that have the ability to describe
most of human mobility characteristics or any other models because most of the future sensors
will been carried by cars, human, animal and vehicle. In 2010 proposed a model for human
mobility to better describe by Song et al [5]. Integration with WSN makes them different from
traditional sensor network because some of these sensors can be mobile without needing energy
to achieve mobility. In this paper we have studied the performance of the networks under
different nodes distribution and mobility models as shown in figure (1):
Figure (1) Implementation diagram for multi scenarios
2. RELATED WORK
The most recent studies that related to sensor networks, focus on coverage, algorithms, and
protocols to reposition sensors in the environment. Typically, dynamic networks are the hottest
topic between wireless sensor networks and social networks. Since dynamic networks depends on
mobility models, there are many works discuss nodes mobility from difference perspective.
Nowadays, the general understanding of human mobility patterns are non-random; indeed that
sound completely true because usually the human move or travel with purpose [6] Moreover,
people have a plans before travel to anywhere. For example, going to school or work, the one
normally route specific ways with less distance and time to get our final destination frequently.
So it is clear that our mobility are random. Continuous-time random walk CTRW, in 2006, is a
random walk has been used by Brokmann et al. to explain the scaling laws for the flow of bank
notes and then to infer the dynamics of human travels. Mtibaa et al. in 2010 offered PeopleRank
as a new forwarding method. This method is based on the PageRank algorithm and the centrality
of nodes. Ranking nodes are considered to their social information, a node is given higher weight
if it is socially connected to the significant network nodes [7] . The results exposed that the
PeopleRank method outperforms the Epidemic approach. Song et al. in 2010, has suggest a
pattern model for human mobility according to their preferred locations. This model is based on
two mechanisms which are: exploration and preferential return. Li et al. in 2011, proposed that
mobility of sensors can be used to efficiently decrease the detection time of a stationary intruder.
Tomasini et al. in 2013 proposed that wireless sensor networks with social mobility patterns can

3
enhance the performance in terms of nodes’ coverage in addition to data delivery. The same
authors, but in another study, evaluated which mobility models are more convenient among
different mobility models in social networks of sensor for data delivery. The study results are like
that (1) not big changes in the performance when using different mobility models in dense
networks; (2) communication range plays important role in performance in terms of data delivery;
(c) delivery rate is a big issue when using individual mobility model in wireless sensor network.
De Melo et al. in 2013, suggest a new technique called RECAST to analyze individuals’
interactions in a dynamic network. This technique group individuals’ interactions to four groups:
Random, Bridges, Acquaintance and Friends. The results of the research showed friends group
was the highest used to deliver data message and random group was the lowest used. Li et al. in
2014, proposed a social group-based forwarding approach. The initial proposal is depends on the
history of encounters of individuals to choose the right relays to data delivery according to
diverse social relationships between network characters. [7, 8] In this paper have studied the
performance of WSN in both different mobility and different distribution.
3. MOBILITY MODELS
The node mobility to achieve big effect on the performance of mobile ad hoc networks and
wireless sensor networks (WSN). Because mobility is to help security, increase capacity and more
reliable of detection in mobile ad hoc networks and WSN. In contrast mobility can also lead to
reduce the unpredictable random topology and the change of links [2,12]. In this project, we use
two common mobility, which are Brownian motion, and Levy walk.
3.1. Brownian Motion
This type of mobility model is describe the motion of the atoms, which move randomly. Where
the time is divided with an interval of constant slot T. At every slot I .where the destination
(Xi+1, Yi+1) is selected depending on the current location, (Xi, Yi), i.e., [5].
(1)
Where W1 and W2 follows normal distribution.
η is the variance of Brownian motion. So, the jump distance during a slot can be obtained from
the distribution
(2)
3.2. Levy walk with cut off model
The levy model describe the most of human mobility characteristics for this reason it is used in
order to achieve the social movement of humans. This model is first described by Song et al. [9]
proposed a model for individual mobility (IM) to better describe human movements which is
Levy walk. It is a different type of random walk in which the jump distance is distributed
depending on a heavy-tailed distribution for every movement. It follows a power law form.
. (3)
In this model, the upper limit of the jump distance can be set as infinite. To overcome this
limitation, the model has been studied Levy extended so that the probability density cutoffs
should be applied on both small and large distances. So the CDF of the jump distance in the Levy
is expressed by:

4
(4)
Where γ and v are the minimum and maximum distances, respectively. This model is based on
two mechanisms which describe the human mobility [10, 12]:
Exploration: The tendency to explore new locations decreases with time. Where the next step is
completely independent of the previously visited locations. Preferential Return: the human's
behavior shows an important property, which is the tendency to return to the most visited
locations (e.g., home, work). Different scenarios can be achieve by the mobility models described
above. For example, in the Brownian motion by changing one parameter η, different levels of
randomness can be offered. On the other hand, the Levy walk can describe the motion
characteristics of human agents in various environments such as campus and downtown, by
altering the parameters α, γ, and v [11]. Figure (2) shoes the difference between the Brownian
motion and Levy Walk.
(a) (b)
Figure (2) mobility model (a) Brownian motion and (b) Levy Walk.
4. EPIDEMIC ROUTING (ER)
The ER is a flooding-based routing protocol mainly used in delay tolerant networks (DTNs). In
DTNs, the assumption of traditional routing of the end to end path from the source to the
destination always exists for successful transmission, is not always valid. Thus, ER rely on transit
transmission in its operation. In ER, if a source node generates a packet and needs to send it to a
destination node, it floods this packet. The source node and all neighboring nodes received this
packet will re-flood this packet whenever they meet a new neighboring node, results in multiple
replica of this packet throughout the network and consequently maximize the probability of one
node encounters the destination node. However, ER causes a massive energy consumption and
therefore, more dead nodes which inversely decrease the probability of packet delivery. [13,14]
If energy consumed by nodes is not an issue, the ER is considered a very efficient in packet
delivery perspective. In wireless sensor networks (WSNs), where the majority of nodes are
battery-powered and battery recharging or replacements is not applicable, renders the energy
consumption is of our highest priority. Each node has to have a buffer to store a created or
received messages destined to other nodes. Message indexing is performed as a way to efficiently
manage the messages. Also, summery vector is an array of bits maintained at each node indicates
the entries stored in hash table.

5
Figure (3) Epidemic Routing
The process of anti-entropy session initiated by the node with lowest id with the highest id,
through which the message are forwarded. Each node must have a list of nodes with which the
node has encountered recently to avoid multiple anti-entropy sessions with the same nodes during
a preconfigured period of time. To calculate the delivery ration DR(t), let N be the total number
of nodes moving within a square area L2, and the transmission range r, where r << L. The pair
wise meeting time between nodes is a random variable and it is exponentially distributed. Thus
the probability that two nodes can meet is defined as pairwise meeting rate and can be calculated
by equation (5).
ρ ≈ 2ωrE[v*]/L2 (5)
Where, ω, is a constant depends on the mobility model. E[v*], is the average relative speed
between two nodes. The number of nodes that received a packet after time t, denoted as I(t) can
be calculated by equation (6).
I(t) = N/[1+(N-1)e-ρNt] (6)
It is intuitive that the more nodes have received the packet, the large probability the message will
be delivered to the destination. The delivery ratio of a packet increases with time, so it is a
function of time DR(t) and equals to:
DR(t) = I(t)/N = 1/[1+(N-1)e-ρNt] (7)
5. PROBABILISTIC ROUTING (FLOODING)
The mobility of nodes in social networks is not completely random. Thus, the movement patterns
of nodes are most likely to be predictable. For example, if a location has been frequently visited,
it is probably to be visited again. Delivery predictability is the metric calculated based on node’s
movement pattern and used in probabilistic routing to improve performance. This metric is
calculated at each node for each known destination node mapping the probability of that node
delivering a message to each known destination node. To mitigate resources waste incurred in
ER, nodes exchange only information about delivery predictabilities and each node decide
whether to forward messages or not based on these probabilities. This Probability P ϵ [0, 1]
indicates the probability of encountering a certain node. Whenever a node is encountered,
corresponding P must be updated according eq. (8)
P(a, b) = P(a, b) old + (1-P(a, b) old )*Pinit (8)

6
Where, P(a, b) , delivery predictability nod (a) has for node (b) Pinit ϵ [0, 1], an initialization
constant If a node doesn’t encounter the destination node for a while, it is less likely to be a good
forwarder to that destination, thus the delivery predictability must be aged. Equation (9) is the
aging equation.
P(a, b) = P(a, b) old * γk (9)
Where, γ ϵ [0, 1), is the aging constant k, is the number of time units have elapsed since the last
time the metric was aged
6. SIMULATION RESULTS
In this section, we verify our findings by the implantation all cases mentioned in figure (1) to find
number of messages in each cases (source to distention i.e LBF case mean: Lattice, Brownian,
Flooding and ECE: Exponential, Levy with cut off, Epidemic ) and the representation for each
block as shown in table (1):
Table (1) Description for each simulation case
In this work we proposed to change some factor by keeping the other the general case generates
by 300 nodes as a mobile and 50 static for a specific location for both source and sink, then we
have repeated the simulation procedure under different distribution by keeping the mobility and
changing the routing tech. Obviously this work proposed 2 type of routing techniques, Two type
of mobility models and 3 kind of distributions (each option represented by a letter as shown in
figure above). The simulation result of the number of messages and spreading percentage for each
case as follow for example (LBF: Lattice-Brownian-Flooding Prophet) table (2):
Table (2) Simulation results for each case under different distribution, Mobility and Routing
RoutingMobilityNode Distribution
E: EpidemicB: Brownian motion
L: Lattice Uniform
E: Exponential
F: Flooding Prophet
C: Levy walk with
cut offP: Power low

7
Obviously simulation results shows that Epidemic Routing will arrive the distention (sink) by
number of message more than Flooding Prophet. Also the levy walk with cut off mobility motion
gives a better results compare with the Brownian motion because higher number of messages
mean higher radiated power. Integration in this paper clearly make the effect of mobility. In other
side node distribution play a main role as well for selection an appropriate routing in designing
WSN and compression of results for each same group case constant like changing only
distribution. For example at lattice distribution and Brownian motion we can implement the WSN
under Epidemic then under Flooding Prophet to get different results as shown in figure (4):
Figure (4) Result compression for each case – A) Messages number from source to destination, B)
Spreading percentage (i.e LBF: Lattice, Brownian, Flooding and ECE: Exponential, Leavy with cut off,
Epidemic )
7. CONCLUSIONS
WSN have greatly extended playing an important role for the data efficient delivery source to
destination and It is an important emerging communication system which takes the advantages of
any possible contact opportunities to deliver data among mobile devices. Integration WSN with
any other connection details like social network will lead to improve the network performance
and provide deep understanding. Routing and forwarding data to the sink are main challenges. In
this paper, we propose a good evaluation procedure to compare different cases in mobility and
distribution to get a full view of each factor how effect on WSN performance. In the previous
section, we analyzed our results for the scenario mentioned in figure (1) by calculation the
number of messages for each case. Now, we can conclude the following:

8
1) Mobility motion has a main effect on WSN for any individual application like power
consumption, spreading messages and directions.
2) Routing by probabilistic (PRophet) less than Epidemic by number of message which give
a good indication for power reduction by reduce transmission.
3) Static node has not relevant effect on WSN with compare with mobility nodes. Also
taking in consideration the nude distribution during design WSN.
4) Levy walk with cut off human mobility motion is better than Brownian motion for any
WSN application related to human activities.
5) If the goal is reducing the number of exchanged Messages within the network, the best
option is to select an appropriate routing and mobility techniques for each individual
application. Finally, in future we can mitigate the impact of integration using a different
technology for wireless communications to get better performance in WSN.
ACKNOWLEDGEMENT
Our Team greatly appreciate Mr.Basim Al-khashab who is a PhD student in Florida tech
(FIT), Florida, USA for discussions and valuable comments on the manuscript.
REFERENCES
[1] N. Pantazis, S. Nikolidakis, and D. Vergados, “Energy-efficient routing protocols in wireless sensor
networks: A survey,” Communications Surveys Tutorials, IEEE, vol. 15, no. 2, pp. 551–591,2013.
[2] R.Alagu Pushpa1 A.Vallimayil "Impact of Mobility models on Mobile Sensor Networks"
International Journal of Communication Network & Security, Volume-1, Issue-1, 2011
[3] Nikolaos A. Pantazis, Dimitrios D. Vergados, “A Survey on Power Control Issues in Wireless Sensor
Networks,” IEEE Communications Surveys, 2007, Vol. 9, Issue 4, pp. 86-107.
[4] Marcello Tomasini, Franco Zambonelli, "Evaluating the Performance of Social Networks of Sensors
under Different Mobility Models" biomedcomm,62 IEEE COMPUTER SOCIETY 2013
[5] Wang, Pu, and Ian F. Akyildiz. "Effects of different mobility models on traffic patterns in wireless
sensor networks." Global Telecommunications (GLOBECOM 2010), 2010 IEEE. IEEE, 2010..
[6] C. Song, T. Koren, P. Wang, and A.-L. Barab´asi, “Modelling the scaling properties of human
mobility,” Nature Physics, vol. 6,no. 10, pp. 818–823, 2010.
[7] G. Lin, G. Noubir, and R. Rajaraman, “Mobility models for ad hoc network simulation,” in
INFOCOM 2004. Twenty-third AnnualJoint Conference of the IEEE Computer and Communications
Societies, vol. 1. IEEE, 2004.
[8] S. Chellappan, W. Gu, X. Bai, D. Xuan, B. Ma, and K. Zhang, “Deploying wireless sensor networks
under limited mobility constraints,” Mobile Computing, IEEE Transaction 1142–1157, 2007.
[9] C. Song, T. Koren, P. Wang, and A.-L. Barab´asi, “Modelling the scaling properties of human
mobility,” Nature Physics, vol. 6, no. 10, pp. 818–823, 2010.
[10] R. M. M. Tomasini, F. Zambonelli, “Using patterns of social dynamics in the design of social
networks of sensors,” ACM, 2013.
[11] I. Rhee, M. Shin, S. Hong, K. Lee, and S. Chong. "On the levy-walk nature of human mobility". In
Proceeding of INFOCOM 2008, pages 924–932, 2008.
[12] Wang, Pu, and Ian F. Akyildiz. "Spatial correlation and mobility-aware traffic modeling for wireless
sensor networks." IEEE/ACM Transactions on Networking (TON) 19.6 (2011).
[13] Xiaofeng Lu, Pan Hui. An Energy-Efficient n-Epidemic Routing Protocol for Delay Tolerant
Networks, IEEE 2010
[14] Floriano De Rango, Salvatore Amelio, Peppino Fazio. University of Calabria, Italy, IEEE 3013.
[15] Anders Lindgrn, Avri Doria, Olov Schelѐn, Lulea University of Technology. Probabilistic Routing in
Intermittently Connected Networks. Maryland, USA 2003

Effects of mobility models and nodes distribution on wireless sensors networks

More Related Content

What's hot (17)

Viewers also liked (19)

Similar to Effects of mobility models and nodes distribution on wireless sensors networks (20)

Recently uploaded (20)

Effects of mobility models and nodes distribution on wireless sensors networks