Sunteți pe pagina 1din 14

Mobility Models Comparison for Vehicular Adhoc Network

1. Introduction:

The issue of link duration in Ad-hoc networks has gained much importance in the community of
researchers and engineers working in wireless networks of moving objects due to its applications
in Smartphones and other user mobile devices, robots, cars, unmanned aerial vehicles, sensors,
and actuators, etc in the last decade. These networks consist of a number of autonomous or semi-
autonomous wireless nodes/objects moving with diverse patterns and speeds while
communicating via several radio interfaces simultaneously. Here every object acting as a
networking node generating, relaying, and/or absorbing data, these networks may serve as a
supplementary infrastructure for the provision of smart, ubiquitous, highly contextualized and
customized services and applications available anytime-anywhere-anyhow. The advent of the
state of art entity mobility models based on random way point, random direction and free way
addresses the shortcomings and challenges of mobility of nodes in these networks.

The applicability of these mobility models have their own strengths and weakness depending
upon link duration, node structure, network size, convergence of nodes and some other necessary
parameters. The dynamical behavior of nodes, their connectivity and topology characteristics
constraints effects the link dynamics of the said problem, moreover the node mobility is one of
the factors which heavily influences the links dynamics, accurate representation of the user
mobility in the analysis of wireless networks is a crucial element for both simulation and
numerical/analytical modeling. The analysis of the impact of mobility on the performance of
routing protocols has been thoroughly investigated in while the traffic pattern, node density and
initial placement pattern of nodes that always affect the routing performance and link duration is
lacked. Similar work has been done in by encapsulating the movement pattern by classifying the
direct mobility and derived mobility metrics while the routing protocol have not been taken into
the account that is a serious concern in link duration.

The investigation has been made on the issue of link duration by considering the relative speed
and active distance between two nodes with the assumption that node movement follows the
random way point model; however later on it has been proved that the random way point model
could not provide a steady state where the average speed of nodes consistently decreases over
time, and therefore should not be used directly in simulations. The impact of mobility models on
link stability is explored by taking the statistics of multi-hop path duration, probability density
function parameters including mobility model, relative speeds, number of hops, and
communication range, however the computational cost of these networks increase exponentially
with the increase in number of nodes and varying speed. As an alternate approach, the
exponential distribution method was found to be a good approximation of path duration
distribution for nodes moving at moderate and high velocities in contrast of other theoretical
works suggesting inter contact time between nodes in ad hoc networks is light tailed through
numerical simulations based on the random way point model. Although lots of research has been
done in this area but one should investigate the link duration based on entity mobility models like
in the randomly moving objects.

In this article a detailed analysis of the probability density function of link duration, one of the
derived metric in mobility modeling technique, in the networks of randomly moving objects has
been presented. Three well known mobility models random way point and random direction
along with a freeway model without its spatial and temporal dependencies has been analyzed.
Our framework use two different velocity models, constant and variable velocities to simulate
large number of ad-hoc networking scenarios by varying different parameters such as diameter
of the network in hops, the node density , the dynamicity in term of speed and the diameter of the
network. We find that Link duration of ad-hoc networks with these type of movement behaviors
which shows no spatial or temporal dependencies, is determined by their speed, providing that
the nodes should be in transmission range of each other.

The remainder of the paper is structured as follows: In the section 1, the mobility models and
their formulation have been narrated along with the necessary diagrams. Section 2 presents the
an overview of the proposed scheme and necessary logical steps involved for getting optimistic
link duration via probability of link connectivity . The detailed simulation and discussion on the
results obtained is revealed in Section 3. In the last section the conclusions have been drawn
along with the direction for the future research.

1.Mobility Models

Mobility models play a key part when simulating ad hoc networks. A mobility model is used to
describe the movements of the nodes in statistical terms. Mobility models are broadly divided
into two types 1) Synthetic Models: wrapping all models based on mathematical models, 2)
Trace based Models: generating mobility patterns from real mobility traces. In this article entity
based synthetic mobility models are used

2.1. Random way point mobility model

Random way point mobility model is one of the most commonly used mobility model to depict
mobility patterns of nodes in many networks such as Ad-hoc networks. Simplicity in terms of
independent node movement description is the main characteristic of the model. This model
includes pause (silence) times between changes in destination and speed. A mobile node initiates
by residing in one location for a particular period of time which is also known as pause/silence
time. Once this pause duration elapses, the mobile node selects a destination randomly in the
simulation area and a speed which is evenly distributed between a minimum speed and a
maximum speed. The mobility node then moves toward the newly selected destination at the
desired speed. Upon arrival at the selected destination, the mobility node pauses for a pre-defined
time period before repeating the process, the basic mobility pattern is shown in Fig.1.
Fig.1. Sample movement pattern of a mobility node using the Random Waypoint

Along with the strengths the major weakness is clustering of the simulation area of nodes at the
central part that may lead towards the stucking in the local area.

1.2. Random direction mobility model

The RDM model was created to overcome density waves in the average number of neighbors
produced by the random waypoint mobility model. In this model, mobile nodes choose a random
direction in which to travel similar to the random walk mobility model. A mobile node then
travels to the border of the simulation area in that direction. Once the simulation boundary is
reached, the mobile node pauses for a specified time, chooses another angular direction (between
0 and 180), and continues the process. The structure and pattern of the mobility is shown below
in Fig.2.

Fig2. Sample movement pattern of a mobility node using the Random Waypoint

1.3. Free mobility model

Mobility model used in this paper is a variation of free way mobility without any spatial or
temporal dependencies, this model is proposed as to observe the probability of link duration in
the bi directional ad-hoc networks of objects, moving straight without any angle or direction
change.
In our proposed mobility model simulation area is divided into two half with respect to Y-axis.
All nodes will select their initial position then according to their Y-axis selection they will start
to move in right or left direction till the end of the simulation. If boundary reached then they will
leave the simulation area, and another node will be generated from the next extreme of the
boundary as to keep the density of the network constant. The behavior pattern in bi-directional
line following the mobility model is presented in Fig. 3.

Fig3. Sample movement pattern of a mobility of freeway mobility model

2. Simulation and Results

A scenario based simulations has been performed on the machine with AMD, Athlon Dual-Core
QL-62 2.00 GHz with 3 GB RAM and 32 bit operating System with MATLAB version 12a.
Mobility generator have been developed to create different mobility pattrns.

Fig4 (a) Fig4 (b)


Freeway model Random direction model
Fig 4(c) Random way point model

Fig 4:showing the moving patterns generated by the matlab simulator for all three models

There are generic parameter values or setting involves no of iterations, fitness value and
computational time complexity while the specific parameters involves network dimension size,
network density, mobility velocity etc. The parameter setting or values used in simulations are
tabulated in Table.1.
Table1. Simulation parameters/ setting values

Parameters Value/ Setting


Network Dimension 100100 m2
Vmax, Vmin Problem Specific
Start Position along x-point & y-axis [10 100]
Probability Function Gaussian
Direction Range [0-180]
Simulation Time 100, 200, 500, 1000
Time Step 1.0 seconds

3.1Comparative Study for Entity Models:

The comparative study of three models is examined based on the computational complexity in
term of time . entity models has been observed by four statistical parameters based on maximum
value (Max) which occurs for high node density, minimum value of the (Min) which occurs for
low node density, standard deviation in the (STD) and mean value of the time taken to run the
model on each independent run.

Table.2 Computational complexity in term of time

Model 1 Mean Max Min STD


Scenario 1 11.58 27.14 2.009 11.438
Case 1 Scenario 2 10.46 24.31 1.83 10.21
Scenario 3 11.44 26.76 1.94 11.26
Scenario 4 11.2 26.1 1.5 11.1
Scenario 1 11.52 26.8 2.01 11.25
Case 2 Scenario 2 10.46 24.3 1.86 10.18
Scenario 3 10.72 24.9 1.94 10.42
Scenario4 10.83 25.4 1.84 10.5
Model 2
random
direction
Scenario 1 9.85 24.29 1.21 10.5
Scenario 2 9.28 22.38 1.24 9.62
Scenario 3 10.42 25.3 1.38 10.88
Scenario4 10.8 24.8 1.5 10.2
Case 2 Scenario 1 10.4 25.2 1.4 10.81
Scenario 2 9.15 22.13 1.23 9.5
Scenario3 9.39 22.5 1.35 9.61
Scenario4 9.45 23.3 1.42 9.8
Model 3
High way
Scenario 1 10.6 25.8 1.3 11.13
Scenario 2 10.22 24.37 1.64 10.45
Scenario 3 11.27 27.3 1.58 11.78
Case 2 Scenario 1 13.86 33.57 1.63 14.46
Scenario 2 9.14 21.93 1.31 9.37
Scenario 3 9.05 21.7 1.3 9.28

3.2. Case 1: Constant Node Speed:


The comparison of three mobility models has been carried out with constant node speed while
other generic specification is same as mentioned in Table1. The following scenarios have been
evaluated in case 1.

Scenario: 1 Very Low mobility and low transmission Range

a. b.
c. d.
Fig.5. Behavior of varying node density with fix values of velocity and transmission range

In scenario 1,With the constant speed of 1m/s having transmission range 15 with different
density values 15, 25, 50 and 75, simultaneously whose graphs at semi-log scale to present the
clear difference are presented in Fig.5 (a) to Fig.5 (d), respectively. Random Waypoint shows
3% - 4.5 % chances of establishing connection for around 15 sec. it also shows more than 1%
chances of establishing connection for a longer period of time i.e. 20-30 sec. Random Direction
shows relatively high probability 11% of establishing connection for a time period of 15 seconds
for density value 15. For other density values its shows less probability 3% - 5% for time period
of 15 20 seconds. Highway scenario shows scattered behavior for establishing connection for
3% - 4.5% for different time periods

Scenario: 2 Low mobility and low transmission Range


,

a. b.
c. d.
Fig.6. Behavior of varying node density with low velocity and low transmission range

In scenario 2,With the constant speed of 10 m/s having transmission range 15 with different
density values 15, 25, 50 and 75, simultaneously whose graphs at semi-log scale to present the
clear difference are presented in Fig.5 (a) to Fig.6(d), respectivelyRandom Waypoint: have low
chances 10% - 15 % of establishing connection for 1-5 seconds. Random Direction: have low
chances 15% - 18% of establishing connection for 1-4 seconds.Highway scenario: have fair
chances 20% - 35% of establishing connection for 1-2 second but also have low chances 7% for
connecting for 6-8 seconds. For all three scenarios chances for establishing connection for a
longer period of time are negligible i.e. 0.1 %.

Scenario: 3 Moderate mobility and moderate transmission Range

a. b.
c. d.
Fig.7. Behavior of varying node density with moderate velocity and moderate transmission range
In scenario 3 of case 1, the low mobility v=20m/s with moderate transmission range of 25 has
been simulated while the node density is varying from low to high with numerical values 15, 25,
50 and 75, simultaneously whose graphs at semi-log scale to present the clear difference are
presented in Fig.7(a) to Fig.7 (d), respectively. Random Waypoint: have fair chances 15% - 19 %
of establishing connection for 1-2 seconds. Random Direction: have moderate chances 20% - 25
% of establishing connection for 1-2 seconds. Highway scenario: have fewer chances 15% - 19
of establishing connection for 1 second but also have low chances 7% for connecting for 6-7
seconds. For all three scenarios chances for establishing connection for a longer period of time
are negligible i.e. 0.1 %.

Scenario: 4 High mobility and moderate transmission Range

a. b.
c. d.
Fig.8. Behavior of varying node density with high velocity and moderate transmission range
In scenario 4 of case 1, the high mobility v=30m/s with moderate transmission range of 25 has
been simulated while the node density is varying from low to high with numerical values 15, 25,
50 and 75, simultaneously whose graphs at semi-log scale to present the clear difference are
presented in Fig.8 (a) to Fig.8 (d), respectively Random Waypoint: have fair chances 32 - 30 %
of establishing connection for 1-2 seconds. Random Direction: have moderate chances 35% - 25
% of establishing connection for 1-2 seconds. Highway scenario: have higher chances 32 - 30 of
establishing connection for 1 second but also have low chances 7% for connecting for 6-7
seconds. For all three scenarios chances for establishing connection for a longer period of time
are negligible i.e. 0.1 %.

3.1. Case 2: Variable Node Speed:


The comparison of three mobility models has been carried out with variable node speed while
other generic specification is same as mentioned in Table1. The following scenarios have been
evaluated in case 2

Scenario: 1 very low mobility moderate transmission Range


In scenario 1 of case 2, the low mobility v ranges from 1m/s to 4 m/s with low transmission
range of 15 has been simulated while the node density is varying from low to high with
numerical values 15, 25, 50 and 75, simultaneously whose graphs at semi-log scale to present the
clear difference are presented in Fig.9 (a) to Fig.9 (d), respectively
a. b.

c. d.
Fig.9. Behavior of varying node density with varying low velocity and moderate transmission range

Scenario: 2 low mobility low transmission Range

a. b.

c. d.
Fig.10. Behavior of varying node density with varying low velocity and low transmission range
In scenario 2 of case 2, the low mobility v ranges from 5m/s to 10 m/s with low transmission
range of 15 has been simulated while the node density is varying from low to high with
numerical values 15, 25, 50 and 75, simultaneously whose graphs at semi-log scale to present the
clear difference are presented in Fig.10 (a) to Fig.10 (d), respectively. Random Waypoint: have
very low chances 4 % - 6 % of establishing connection for 1-2 seconds. And less 4 % for
establishing connection for 4-22 seconds. Random Direction: have very low chances 7% - 15%
of establishing connection for 1-2 seconds. . And less 4 % for establishing connection for 3-20
seconds. Highway scenario: have very low chances 7 % of establishing connection for 1 second.
And less 4%for establishing connection for 2-35 seconds. For all three scenarios chances for
establishing connection for a longer period of time are very low i.e less than 1 % for more than
10 sec time
Scenario: 3 moderate velocity low transmission Range
In scenario of case 2, the low mobility v ranges from 11m/s to 20 m/s with low transmission
range of 15 has been simulated while the node density is varying from low to high with
numerical values 15, 25, 50 and 75, simultaneously whose graphs at semi-log scale to present the
clear difference are presented in Fig.11 (a) to Fig 11(d), respectively

a. b.

c. d.
Fig.11. Behavior of varying node density with moderate velocity and low transmission range

Scenario4: high velocity low transmission Range


In scenario 4 of case 2, the high mobility v ranges from 20m/s to 30m/s with low transmission
range of 15 has been simulated while the node density is varying from low to high with
numerical values 15, 25, 50 and 75, simultaneously whose graphs at semi-log scale to present the
clear difference are presented in Fig.12 (a) to Fig.12(d), respectively

a. b.

c. d.
Fig.12. Behavior of varying node density with high velocity and low transmission range
12(a) has been scaled for having a more cleared vision of probability density function

Conclusion
Diversity of application area of adhoc networks had made this area a very interesting topic for
researchers, Mobility models play a vital role in simulating the link duration between nodes used
to create ad hoc networks, In this paper monte carlo simulation based study has been performed
to analyze the overall connectivity of the network by creating large number of scenarios of adhoc
networks, probability of occurrence of different time duration has been calculated by using three
different mobility models. Based on the simulation and results the following conclusion can be
drawn:

The three entity models are capable enough to find the link duration with a reasonable
PDF values and computational derive.
Link duration is mainly dependent on the mobility speed , transmission range used, and
the pattern of movement
We had analyzed that the density of nodes in the network may affects the quantity of
links formed, but link duration is less affected by this factor, as we can see in all the scenarios
that by changing the density of nodes , the probability function is not affected by this variation,
we can justify the statement as on the bases of entity based modeling
of the movement pattern used. This factor highly effects the computational complexity in terms
of time.
Mobility speed effects the link duration probability a lot, in low mobility scenarios like
1 of both cases we can observe that probability of occurrence of any duration of link time is
present even though the density function shows a very less value for them, but this effect
diminish with increasing the node speed .The probability of having connection for shorter time
period increases with the increase of speed of mobile nodes in both cases, this factor has low
effect on computational complexity in terms of time
With the increase of transmission range diversity of occurring of long durations increase
in all mobility models, this factor has also low effect on computational complexity in terms of
time
Mobility pattern effects link duration probability significantly
Random way point show less probability of link duration as compare to random
direction and freeway mobility models in most of the scenarios, in nearly all scenarios this
mobility pattern, have highest probability function for 0 to 2 seconds while for all other
possibilities it shows very low value of link duration. This can be justified on the bases of walk
interval parameter setting of this particular mobility model
Random Direction mobility model shows diversity in its probability density function in
very low mobility cases, but it converge to some specific value of probability density function
with the increase of speed, it shows higher values of probability function as compare to random
direction mobility model in nearly all scenarios.

Most amazing results have been achieved for free way model, This model shows
probability of the occurrence of any particular event of link duration in all scenarios with wide
range , this effect converges toward uniformity in high mobility scenarios but probability of
having diversity is still present in freeway model as compare to random direction and random
way point mobility models (fig 6), to enhance this feature 10( a) has been scaled.
Random way point shows highest computer complexity as compare to the remaining two
models, random direction and high way scenarios shows moderate complexity, with a little
fraction. Random direction has the lowest computational complexity in terms of time
Highway mobility model show dominancy in producing the scatterings and Gaussian
behavior while the other two models are positively skewed.

S-ar putea să vă placă și