Analyze traffic forecast for decentralized multi agent system using I-ACO routing algorithm

Full text

Author's personal copy
Vol.:(0123456789)
1 3
Journal of Ambient Intelligence and Humanized Computing
https://doi.org/10.1007/s12652-018-0981-2
ORIGINAL RESEARCH
Analyze traffic forecast fordecentralized multi agent system using
I-ACO routing algorithm
V.GokulaKrishnan1 · N.SankarRam2
Received: 25 April 2018 / Accepted: 11 August 2018
© Springer-Verlag GmbH Germany, part of Springer Nature 2018
Abstract
Traffic congestion is a condition on road networks that may cause the vehicle to move very slow, longer trip timing changes,
and vehicle queue length increase. We need accurate predictions which require accurate status information about vehicles—
the fact that the vehicles are distributed over large-scale road infrastructure that makes mostly challenging one. Advanced
vehicle guidance systems use real time traffic information but unfortunately can only react upon the presence of traffic.
Anticipatory vehicle routing is promising approach, accounting for traffic forecast information. This concept presents an
efficient decentralized approach for anticipatory vehicle routing that is particularly useful in large-scale dynamic environ-
ments with some additional techniques and experiments. The approach is based on delegate multi agent systems (MAS),
i.e., an environment-centric coordination mechanism that is, in part, inspired by ant behavior. This paper mainly focus on
provide the traffic forecast among the road network is very efficient to minimize the traffic.
Keywords Traffic forecasting· Multi-agent systems· Anticipatory vehicle routing· Intelligent based ACO
1 Introduction
People use vehicles for making trips using the road infra-
structure. The large number of vehicles today and the limited
capacity of the road networks make routing traffic a particu-
larly challenging problem. Not only does a vehicle need to
reach its destination, it is desired that the trip can be per-
formed in a timely and comfortable fashion. Besides basic
SatNav devices, which use static maps for fastest path rout-
ing, more advanced devices exploit broadcast traffic infor-
mation (e.g. through Traffic Message Channel or TMC). An
accident causing a traffic jam on the route of a vehicle can
trigger the vehicle to reroute and bypass the traffic jam. This
mechanism allows a substantial performance gain already.
One disadvantage of these state-of-the-art approaches lies in
the fact that these allow only to react upon traffic jams after
they occurred, and hence are already propagating delays in a
typically substantial part of the traffic network. Anticipatory
vehicle routing aims to encompass this by using forecast of
traffic density. Forecast information can either be extracted
from historical data, or can directly rely on the individual
planned routes of the vehicles. Besides obtaining and dis-
seminating forecast information, major challenges are: (1) to
cope with the large scale of traffic—consisting of huge num-
bers of vehicles residing on large road networks, (2) to cope
with dynamics—accidents, road blocks, demand peaks, have
local effects with potentially far-reaching consequences, (3)
stability—reactions of vehicles to traffic information must be
managed to avoid unstable system behavior due to vehicles
rerouting continually (Claes etal. 2011).
In a broad sense, traffic forecasting is the process of esti-
mating the number of vehicles or people that are likely to use
different transportation facilities in the future. For instance,
a forecast may estimate the number of vehicles on a planned
road or bridge, the expected ridership on a railway/metro
line, the number of passengers visiting an airport, the num-
ber of ships arriving at/leaving from a seaport or may esti-
mate the expected future traffic levels for the whole country.
This process begins with the collection of data on current
* V. Gokula Krishnan
gokul_kris143@yahoo.com
N. Sankar Ram
n_sankarram@yahoo.com
1 Faculty ofComputer Science andEngineering, Sathyabama
Institute ofScience andTechnology, Rajiv Gandhi Salai,
Chennai, TamilNadu, India
2 Department ofComputer Science andEngineering, Sri Ram
Engineering College, Veppampattu, Tiruvallur, TamilNadu,
India

Author's personal copy
V.Gokula Krishnan, N.Sankar Ram
1 3
traffic. This traffic data is combined with other known data,
such as population and economic growth rates, employment
rate, trip rates, travel costs etc., to develop a traffic demand
model for the current situation. Knowledge of future traffic
flow is an essential input in the planning, implementation
and development of a transportation system.
Figure1 shows the traffic congestion in the road net-
work. Traffic congestion is one of the leading causes of
lost productivity and decreased standard of living in urban
settings. Recent advances in artificial intelligence suggest
that autonomous vehicle navigation will be possible in the
near future. Individual cars can now be equipped with fea-
tures of autonomy such as cruise control, GPS-based route
planning and autonomous steering (Schonberg etal. 1995).
Once individual cars become autonomous, it is inevitable
that before long all, or most, of the cars on the road will
have such capabilities, thus opening up the possibility of
considering autonomous interactions among multiple vehi-
cles. Multi-agent Systems (MAS) is the subfield of AI that
aims to provide both principles for construction of complex
systems involving multiple agents and mechanisms for coor-
dination of independent agents’ behaviors. Stone and Veloso
(2000) proposed an MAS-based approach for improving traf-
fic congestion, specifically at intersections. In a MAS with
decentralized coordination each agent has the capability to
coordinate, as well as their functional (problem solving)
capabilities.
Agent technologies can be applied into the building envi-
ronment to control environmental parameters and solve pos-
sible conflicts arising between energy efficiency and user’s
satisfaction. Generally, an agent is defined as a software (or
hardware) entity that is situated in a certain environment
and is able to autonomously react to changes in that envi-
ronment. An agent has three basic characteristics: reactiv-
ity, pro-activeness and social ability. Reactivity refers to the
ability of agent in reacting to the changes in its external envi-
ronment; pro-activeness indicates the goal-directed behavior
of an agent; social ability indicates an agent should be able
to interact with other agents based on an agent communica-
tion language, which allows agents to converse rather than
simply passing data (Priyanka and Sharma 2014).
2 Literature review
In VANET, the routing protocols are classified into various
categories: topology based routing protocols, position based
routing protocols, cluster based routing protocols, geo-cast
routing protocols and broadcast routing protocols. The topol-
ogy based routing protocols use links information that exists
in the network to perform packet forwarding. Hybrid proto-
cols: The hybrid protocols try to find the best compromise
between proactive and reactive routing protocols. E.g., Zone
Routing protocol (ZRP) (Zhao etal. 2016).
Karthikeyan etal. (2012) proposed the cluster based rout-
ing protocols, in which the network is divided into small
partitions called clusters. The main goal of clustering is
to find an interconnected set of clusters. E.g., Hierarchical
cluster based routing (Rakesh and Mayank 2011). Daeinabi
etal. (2011) proposed an efficient clustering algorithm for
clustering in VANETs. The authors get different factors such
as entropy, direction of vehicle and number of neighbors to
perform the clustering of vehicles in an exacting area.
Kumar etal. (2012) proposed an agent learning—based
clustering algorithm in vehicular ad hoc networks. The
authors evaluated the performance by taking different met-
rics like node participation, percentage of connectivity, clus-
ter head period, connectivity preservation ratio and mes-
sage transmission ratio. Tyagi and Vatsa (2011) proposed
architecture of VANET based on clusters that’s designed by
Fig. 1 Traffic occur in road

Author's personal copy
Analyze traffic forecast fordecentralized multi agent system using I-ACO routing algorithm
1 3
mobile agents. The authors create a new clustering concept
with the help of mobile agents among the VANET nodes.
However, in those studies one of the vehicles which is clus-
ter head takes all the decisions on behalf of the other vehi-
cles. And as the topology changes, these decisions may be
not be valid.
A lightweight anonymous method based on bilinear pair-
ing to resolve the privacy preserving problem of vehicular ad
hoc network (VANET) is proposed in INCoS international
conference. This scheme can protect drivers’ privacy not
only between vehicles, but also among vehicles and road
side units (RSUs). Meanwhile, this scheme does not ask for
vehicles to conduct any bilinear pairing procedures, thus
significantly ease the computational complexity of vehicles.
Authors demonstrate this scheme through protection anal-
ysis and evaluate its computational performance by simu-
lations. The results show that this work scheme performs
well in providing secure communications and anonymous
authentication among RSUs and vehicles.
Daeinabi etal. (2011) proposed three clustering algo-
rithms, in first algorithm author proposed a novel cluster-
ing algorithm vehicle clustering algorithm VWCA that take
into consideration the number of neighbor base on dynamic
transmission range the direction of vehicles, entropy, and the
distrust value parameters. These parameter increase stability
and connectivity and can reduce overhead in network .on
the other hand, transmission range of vehicles is important
to forward and receiving the massage .when fixed transmis-
sion range mechanism is used in VANET. It’s likely that
vehicles are not located in range of neighbor. This is because
of the high rate change in topology and high variability in
vehicle density .thus author proposed the allocation and
transmission rage (AATR). This technique is used as sec-
ond algorithm and finally author proposed the monitoring
of malicious vehicle (MMV) algorithm as third algorithm
to determine distrust value for each vehicle used in VWCA.
Over dynamically changing networks, finding best routes
can also be fulfilled through using Swarm Intelligence (SI)
based methods. One of the most advantageous SI methods
for exploring optimal solutions at low computational cost
is ant routing algorithm. Caro and Dorigo (2011) presented
AntNet which is a routing algorithm inspired by the natural
ants’ behavior and operates based on distributed agents.
An adaptive connectivity aware routing in VANETs was
proposed by Qing etal. (2010). The authors demonstrated
the choice of the optimal path based on collecting data from
diverse regions. Due to the mobility of nodes in VANETs,
the backbone must be continuously reconstructed in a timely
fashion, so the research on more distributed, adaptive and
intelligent protocol becomes critical. That is why; agent
technology has become an exciting and promising research
area for vehicular ad-hoc networks. Kumar etal. (2012)
proposed an agent learning—based clustering algorithm in
vehicular ad hoc networks. The authors evaluated the per-
formance by taking different metrics like node participation,
percentage of connectivity, cluster head period, connectivity
preservation ratio and message transmission ratio.
Philips etal. (2013) presented that traffic radar is a hol-
onic traffic coordination system that also uses the delegate
Multi-agent pattern to provide anticipatory route guidance
to users. Within the MODUM project Traffic Radar is used
to provide multimodal routing information to end users
(Namoun etal. 2013). As such many of the characteristics
in Traffic Radar are also present in AntTIS: swarm based
interactions and refresh-and-evaporate. Instead of vehicle
agents, traffic radar features vehicle holons and infrastruc-
ture agents have two counterparts that are called Node and
Link Holon’s. Their responsibilities are very similar. Vehi-
cle Holon’s initiate the exploration and intention propagat-
ing dMAS’s. They are also responsible for calculating the
routes. Link and Node Holon’s maintain the information on
the vehicle intentions and maintain accurate traffic models
describing the entity in the traffic network they represent.
The main aim of a system (Vijayakumar and Arun 2017a,
b) is to constantly scan the applications that are deployed in
the cloud and check for vulnerabilities as part of the continu-
ous integration and continuous deployment process based on
Hashing and NLP technique. This process of checking for
vulnerabilities after every update in the application should
be included in the software development lifecycle proposed
(Manoj etal. 2018) feature selection algorithm for data
based on ACO–ANN.
Junction based geographic routing (Tsiachris etal. 2013)
is one which make the use of selective greedy forwarding
up to the node that is located at a junction and is closer to
the destination. To alleviate issue of local optimum, a junc-
tion based multipath secure routing algorithm is proposed
(Sermpezis etal. 2013). The impact of traffic light on routing
protocol design was investigated in based on an intersection
based routing protocol designed for vehicular stateless rout-
ing propose a typical position based routing. It uses greedy
forwarding to forward packets initially. When a packet
reaches a local optimum it switches to the perimeter mode.
3 Problem identication
The communication of moving vehicles with roadside infor-
mation infrastructure relies on expensive and fragile wireless
network connections. Tasks need a continuous connection
for moving vehicle and a traffic information network are
probably not economically or technically feasible. Need for
manual interference to guide traffic. For example, human
manipulation of phasing lights because ability to reacts
independently. Once autonomous vehicles are common, this
mechanism may be useful for managing real traffic.

Author's personal copy
V.Gokula Krishnan, N.Sankar Ram
1 3
Lack of synchronization between differing modes of
transportation because there is no agent cooperative. Traffic
disturbances happen unexpectedly (e.g. traffic jam follow-
ing an accident); it cannot able to react to stimulus in the
environment in a timely manner due to degree of dynamism.
Group of agent that performs together, they consume more
power than a single agent because the information becomes
incomplete and insufficient resource to solve the problem
(see Fig.2).
4 Research methodology
In this proposed methodology to avoid the traffic in road
network intelligent based ant colony optimization method
is proposed to route the agents in the road network. The
AntTIS system is designed as decentralized MAS. Agents
are able to learn from the environment in which they are
operating and perform the task of CH selection. Agents are
deployed at different road junctions for monitoring the activ-
ities of the vehicles. The Multi-agent system (MAS) that
forms the AntTIS consists of two main agent types namely
the Vehicle Agent and the Infrastructure Agent. Every vehi-
cle is represented by a situated vehicle agent deployed on
(a smart device within) the vehicle. A vehicle agent is able
to access information about that vehicle’s intended desti-
nation and state, including location and speed. A vehicle
agent guides the driver by providing information on routing
towards its destination.
The autonomous architecture is followed by a decentralized
architecture in which real-time information is broadcasted to
vehicles allowing them to adjust their routing to current traf-
fic densities. If the vehicle makes a routing decision on its
own based on the information broadcasted by an Independent
Fig. 2 Architecture of proposed
system

Author's personal copy
Analyze traffic forecast fordecentralized multi agent system using I-ACO routing algorithm
1 3
Service Provider, the architecture is said to be decentralized.
The vehicles share the information to road side unit. In the
network, the routing takes place the share their data from one
vehicle to another. Traffic congestion monitoring system is
used to monitor the traffic load in the road network. While
monitoring the traffic load signal, traffic raise it generate alarm
to the vehicle about traffic information otherwise alarm will
be silent. Then the vehicle agent is rerouting to reach the des-
tination in optimal way. The proposed system provides the
best way to reach the destination without any congestion also
effective power consumption.
At time ‘t’, the length of queue ‘Q’ in n number of vehicle
agents, ‘Qin’ is the input of the vehicle which is arrived and
‘Qout’ is the output of the vehicle which is dispatched.
5 I‑ACO routing algorithm
The I-ACO (Intelligent Based Ant Colony Optimization) algo-
rithm is inspired by the behavior of the ants in real world; ant
colony algorithm is a multi-agent system, in which each single
agent is called an artificial ant. It has been applied to solve
many different types of problems, including the classical trave-
ling salesman problem, path planning and network routing.
In the ant colony algorithm, artificial ants search the solu-
tion space probabilistically to create candidate solutions. These
candidate solutions are then evaluated and updated, based on
the pheromone associated with each one of them. It should be
noticed that over time, certain amount of pheromone concen-
tration may evaporate. Finally, the one with the highest value
of pheromone is considered to be the optimal solution of the
problem. In this research, a new approach to find the optimal
signal timing plan for a traffic intersection is investigated using
ACO algorithm. Traffic signal problem is a complex combi-
natorial optimization problem which fits the nature of ACO
very well. The intelligence based is the algorithm take decision
to reroute the vehicle if traffic occurs. The traffic is occur in
the particular zone the RSU give instruction about the traffic
forecast and also generates the alarm.
Degree of dynamism (DoD) of the vehicle routing problem
can be solved by
The algorithm of intelligent based routing algorithm is.
Q(t)=Q(t−1)+Qin(t)−Qout (t).
DoD
=
V
Dest
V
Source
+V
Dest
.
Input:List of Vehicles
Output
:Optimal path to move vehicle from traffic area
Total Si
mulation time T
Begin
Simulation Time = 0
for each vehicle do
initialize vehicle cost
Cost of vehicle=
// is weight of the vehicle
// is the speed of the vehicle
end for
while (Sim_Time <= T) do
for each vehicle do
calculate Total number of packet received
and
transmitted
compute vehicle position
estimate future position of vehicle
compute η=[−]/[ + ]
broadcast ηvalue and estimate future position
of vehicle
end for
while all messages not received do
wait
end while
=1
=1
initiate traffic =1
if (traffic ≠ 0)
alarm generated
then take another route
else alarm silent
end if
if(S_Ant best> EMap_ best)
map best← map of antbest
end if
Simulation Time ++
end while
end

Author's personal copy
V.Gokula Krishnan, N.Sankar Ram
1 3
6 Performance analysis
The performance analysis of the proposed method is
compared with the existing method that can be follow-
ing below with graph and respective tabular column. The
simulation parameters are number of nodes connected in
a network at a time is varied and thus varying the number
of connections (see Table1).
6.1 Congestion avoidance
The congestion avoidance is the traffic flow occurs in
the road network. The existing methods have congestion
avoidance is low when compared to proposed method. To
overcome the existing issue move to proposed method
I-ACO algorithm that can provide the decision support to
take reroute possibilities. Figure3 show that the blue color
line indicates existing system and red color line indicate
proposed system (see Table2).
6.2 Energy consumption
The energy consumption of existing system become high
because it takes more time to elect the cluster head and
then vehicle moves from one region to another. The pro-
posed system have take decision while routing itself they
have other possible route also so energy consumption is
low. The blue color line shows the existing system which
achieve high energy and red color line indicate proposed
system that shows minimum energy shown in below Fig.4.
6.3 Resource constraint
When compared to existing system the resource that means
information accessibility is low. The proposed system is
high to access. The graph shows the packet accessibility
in particular time slot in Fig.5 show that the blue color
line indicates existing system and red color line indicate
proposed system.
Table 1 Various parameters used
Parameters Value
Number of nodes 50,150
Simulation time 500s
Traffic type CBR
Transmission range 300m
Simulation Area 500 × 500m
Node Speed 18m/s
Pause Time 00s
Packet size 512MB
Queue length 50
15 20 25 30 35 40
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
No. of Vehicles
Time(min)
Existing
Proposed
Fig. 3 Congestion avoidance graph
Table 2 Proposed method performance analysis compared with exist-
ing method
Factors Existing ACO based
clustering algorithm
(%)
Proposed Intelligent
based ACO algorithm
(%)
Congestion avoidance 75 89
Energy consumption 84 68
Resource constraint 69 90
Power consumption 85 72
Average delay 66 47
Simulation time 62 87
510 15 20 25 30 35 40 45
5
10
15
20
25
30
35
40
45
No. of agent
Message transmission
Existing
Proposed
Fig. 4 Energy consumption graph

Author's personal copy
Analyze traffic forecast fordecentralized multi agent system using I-ACO routing algorithm
1 3
6.4 Power consumption
The power consumption of the existing system is high when
compared to proposed system. Also the power is reduced.
It can be shown in below graph. Figure6 show that the blue
color line indicates existing system occupy more power
and red color line indicate proposed system have minimum
power usage only.
6.5 Average delay
The average delay is the time to reduce traffic and the exist-
ing method delay is high, the proposed system delay can
be minimized that can be shown in below graph. Figure7
show that the blue color line indicates existing system and
red color line indicate proposed system.
7 Conclusion
In this proposed methodology to avoid the traffic in road
network, Intelligent based ant colony optimization method
is proposed for routing the agents in the road network. Accu-
rate predictions require accurate status information about
vehicles—the fact that the vehicles are distributed over
large-scale road infrastructure makes this particularly chal-
lenging. Advanced vehicle guidance systems use real time
traffic information but unfortunately can only react upon the
presence of traffic. Anticipatory vehicle routing is prom-
ising approach, accounting for traffic forecast information.
Intelligent based ant colony optimization is efficient way to
reroute and minimize the power consumption, simulation
time and important aspect is to minimize the traffic delay. In
future, we can calculate the signal based climate forecasting
that can be efficient to avoid the traffic. Traffic signal can be
changed with respect to the changes in the climatic condi-
tions in the environment.
References
Claes R, Holvoet T, Weyns D (2011) A decentralized approach for
anticipatory vehicle routing using delegate multi-agent systems.
IEEE Trans Intell Transp Syst 12(2):364–373
Daeinabi A, Rahbar AGP, Khademzadeh A (2011) VWCA: an efficient
clustering algorithm in vehicular ad hoc networks. J Netw Comput
Appl 34:207–222
Di Caro G, Dorigo M (2011) Antnet: distributed stigmergetic control
for communications networks. arXiv preprint arXiv:1105.5449
Karthikeyan K, Appalabatla S, Nirmala M, Tesfazghi T (2012) Proto-
cols for mobile adhoc networks. Int J Adv Res Comput Sci Softw
Eng 2:12
Kumar N, Chilamkurti N, Park JH (2012) Agent learning—based clus-
tering algorithm in vehicular adhoc networks. Pers Ubiquitous
Comput 17:1683–1692
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
2
3
4
5
6
7
8
9
10
11
12
Time
Data Transmission
Existing
Proposed
Fig. 5 Resource constraint graph
15 20 25 30 35 40 45
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Vehicle
Time
Pwower Consumption
Existing
Proposed
Fig. 6 Power consumption graph
0.511.5 2 2. 533.5
5
10
15
20
25
30
35
40
45
50
Time
Traffic Load
Existing
Proposed
Fig. 7 Average delay graph

Author's personal copy
V.Gokula Krishnan, N.Sankar Ram
1 3
Manoj JR, Praveena AMD, Vijayakumar K (2018) An ACO–ANN
based feature selection algorithm for big data, Cluster comput-
ing. Springer, Berlin
Namoun A, Marín CA, Germain BS, Mehandjiev N, Philips J (2013)
A multi-agent system for modeling urban transport infrastructure
using intelligent traffic forecasts in industrial applications of hol-
onic and multi-agent systems. Springer, Berlin, pp175–186 (140)
Philips J, Germain BS, Van Belle J, Valckenaers P (2013) Traffic radar:
a holonic traffic coordination system using prosa + + and d-mas.
In: Industrial applications of holonic and multi-agent systems, pp.
138, 140, 154, 163–174, Springer, 2013
Priyanka T, Sharma TP (2014) A survey on clustering techniques
used in vehicular ad hoc networks. In: Proceedings of 11th irf
international conference, 15th June-2014, Pune, India, ISBN:
978-93-84209-27-8
Qing Y, Alvin L, Shuang L, Jian F, Prathima A (2010) ACAR: adaptive
connectivity aware routing for vehicular ad hoc networks in city
scenarios. Mob Netw Appl 15(1):36–60
Rakesh K, Mayank D (2011) A comparative study of various routing
protocols in VANET. JCSI Int J Comput Sci 8(4):643–648
Schonberg T, Ojala M, Suomela J, Torpo A, Halme A (1995) Posi-
tioning an autonomous off-road vehicle by using fused DGPS
and inertial navigation. In: 2nd IFAC conference on intelligent
autonomous vehicles, pp. 226–231
Sermpezis P, Koltsidas G, Pavlidou F-N (2013) Investigating a junc-
tion-based multipath source routing algorithm for VANETs, IEEE
Commun Lett 17(3):600–603
Stone P, Veloso M (2000) Multiagent systems: A survey from a
machine learning perspective. Auton Robots 8(3):345–383
Tsiachris S, Koltsidas G, Pavlidou F-N (2013) Junction-based geo-
graphic routing algorithm for vehicular ad-hoc networks. Wirel
Pers Commun 71(2):955–973 (Springer)
Tyagi H, Vatsa AK (2011) Seamless handoff through information
retrieval in VANET using mobile agent. IJCSI Int J Comput Sci
28(2):634–640
Vijayakumar K, Arun C (2017a) Continuous security assessment
of cloud based applications using distributed hashing algo-
rithm in SDLC. Cluster Comput. https ://doi.org/10.1007/s1058
6-017-1176-x
Vijayakumar K, Arun C (2017b) Automated risk identification using
NLP in cloud based development environments. J Ambient Intell
and Hum Comput
Zhao H-T, Wang H-M, Zhu H-O, Li D-P (2016) Routing optimization
algorithm based on multi lanes in VANET. In: IEEE
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.