Coalitional game theoretic approach for ieee 802

International INTERNATIONAL Journal of Computer JOURNAL Engineering OF and COMPUTER Technology (IJCET), ENGINEERING ISSN 0976-6367(Print),
&
ISSN 0976 - 6375(Online), Volume 5, Issue 9, September (2014), pp. 50-60 © IAEME
TECHNOLOGY (IJCET)
ISSN 0976 – 6367(Print)
ISSN 0976 – 6375(Online)
Volume 5, Issue 9, September (2014), pp. 50-60
© IAEME: www.iaeme.com/IJCET.asp
Journal Impact Factor (2014): 8.5328 (Calculated by GISI)
www.jifactor.com
IJCET
© I A E M E
COALITIONAL GAME THEORETIC APPROACH FOR IEEE 802.11E
STANDARD BASED ON INCENTIVE AND CREDIT
Dr. Sanjay Tejbahadur Singh
Department of Computer Engineering,
P K Technical Campus, Chakan, Pune
50
ABSTRACT
Vehicular ad hoc network (IEEE 802.11e) is dynamic network where independent identity
node helps each other to forward the message. But some time node may not be interested in
forwarding the message. Thus it is essential to give node some incentive to participate in message
forwarding game on the basis of credit based system. Our approach based on credit based system as
well as coalitional game theory proposes an hybrid approach to solve this issue. Existing message
forwarding technique like proportional, receipt counting and Frame have not addressed load
balancing issue. We have further extended our system by applying load balancing on credits since we
don’t want credit to be concentrated on limited number of nodes. Our proposed approach
outperforms all above existing approaches by increasing delivery ratio with 16 to 20% and reducing
Virtual Currency Center (Vcc) credit time computation by 46 to 48%.
Index terms: Coalitional Game Theory, Credit Base System, IEEE 802.11e Standard, Incentive
Scheme, Load Balancing.
I. INTRODUCTION
Vehicular ad hoc networks (VANETs) consist of smart vehicles and roadside equipment to
support communication between vehicles. The assumption is that in VANETs, each vehicle has OBU
(On Board Unit) for optimized message delivery among vehicles. To solve the problem of
intermittent connectivity between vehicles under certain circumstances like night time or rural area
one solution point of view is store-carry and- forward message switching in VANETs. Thus depend
on type how message get delivered different routing protocols (e.g., [1]–[5]) have been proposed to

International Journal of Computer Engineering and Technology (IJCET), ISSN 0976-6367(Print),
increase the success rate of message delivery. Availability of good protocol doesn't guarantee the
utilization of protocol in the application.
The question related to node behavior for cooperating in VANET arises due to two reasons.
First an individual user can be selfish for its VANET node. Therefore that individual may not be in
scène where it will help other node to forward their message due to its limited resources like its own
storage space. Second reason is that depend on protocol conformability probability of selecting
special kind of vehicle is more as compared to normal vehicle. In [6] proposed system routing
protocols, nodes from same community always get preference as compared to normal vehicle to
forward the message. Another situation is that where many nodes are interested in forwarding the
message but most of their resources like wireless bandwidth or storage space get consumed during
communication. Therefore to save their resources they have to deviate from predefined protocol
unwillingly. Hence it is highly important to involve more number of nodes to cooperate in
forwarding messages. Today the automotive industry controls the vehicle manufacture. However,
after the vehicles are sold, they are under the full control of the users. So if user wants, he can consult
with some expert hackers to change the VANET protocols running in the vehicles and to make
themselves the “free riders” in the network without contributing anything. Hence, we believe that
rather than making forceful interaction in VANETs which is not easy to achieve, try to design system
will take advantage of user selfish nature.
Currently there are two types of existing incentive mechanisms for stimulating cooperation in
51
wireless networks:
Reputation based approaches (e.g., [7],[8])
In Reputation-based approaches behavior of nodes gets observed by neighbor nodes and to
stimulate cooperation among uncooperative nature nodes punishment is given. But in VANETs to
observe whether node has deviating behavior of a selfish node is very difficult. Since the guarantee of
node neighbor will remain same is very less.
Today’s work on incentive-aware routing in delay-tolerant networks [10] tit-for-tat
mechanism has been used where neighbor punish or give reward to node based on history they have
observed. However, in VANETs the probability of two nodes will be meet is very low if compare it
with history data that can be store within node memory. But this scenario will be different for
centralized controller authority.
Credit-based approaches
Credit-based mechanism motivates nodes to cooperate among themselves to get proper
reward. In existing works like traditional multi hop networks many question related to credit based
system had got asked and based on that solution was suggested. The question like how many credit
node should receive, how the end to end connection will set the credit based mechanisms. In this
paper, we have use coalitional game theory to solve the forwarding cooperation problem in VANETs
with its vulnerability problem of load balancing. We propose an incentive and credit based scheme
for VANETs and analysis of it for coalitional games theory. In addition, we extend our approach to
take the limited number of credits of each node for load balancing. In particular, we say that in
coalitional game theory VANET nodes act as players and they cooperate in forwarding message if
they run the routing protocol. In game theory when the players in a subset decide to cooperate within
the subset, the subset is called a coalition. In particular, we are creating scenario of grand coalition
where all nodes will receive incentives to develop a grand coalition.

52
II. RELATED WORK
A number of DTN routing protocols have been proposed, out of which most of can roughly
be classified into two categories depend on their strategies:
• Flooding (replication)-based protocols (e.g., [16] and [17]) and
• Forwarding-based protocols (e.g., [18], and [19])
Flooding-based protocols give guarantee of message to reach destination by making replica of
message and thus messages will reach to destinations with higher probability. For example, in
epidemic routing [9], node with copy will give copy to all the nodes to whom he meet.
In [6] to forward message evolutionary game has proposed where community of vehicles (like
bus community, taxi community) are created from sample space of vehicles. But in this approach as
number of community increases success probability decreases. Again in [11] Broadcast storm
problem has been solved using the volunteer’s dilemma Game but as amount of Vehicle increases
Percentage of performance stability decreases. On other hand in [12] reputation-based cooperative
game has been played but then in that also player participate in the game by the force of neighbor to
deduct its credit if it does not forward the message rather interest of its own.
Another secure incentive framework related to commercial ad dissemination has been
suggested in [13]. In this protocol mapping nature of node message sending is one to many only. But
in case of single or selective number of destination this protocol is not applicable. Again in [14]
routing scheme proposed where protocol ensures that node follows the protocol for routing decision.
While our objective is motivating nodes participate in message forwarding game and at same time
make routing decision to balance number of credits on the nodes.
III. PROBLEM STATEMENT
A. Need of Load balancing in Credit Based System
In credit based system there are some node with large no of credit point and some node with
less number of credit point since there is no constraint on participation of node in the game. There
would be situation occur where node want to send data may not succeed due to lack of credit. Hence
probability of participation is inversely proportional to number of credit in node. Therefore with
improved approach in coalitional game theory for message forwarding there is need of way to
preserve same number of credit on most of node.
B. Coalitional Game and the Core
Formally, a coalitional game formation processes same as defined in [20] as follows:
Definition 1:
In this coalitional game we are considering ordered pair (N, v), where N is the set of players
and subset of N will be called as coalition, and v is a characteristic function which will assign a real
number to each coalition for which ranger varies from 2N to R such that v() = 0. Whenever all
subset of N coalition will come together they will form grand coalition and v will payoff of that
coalition. For example coalition S, v(S) is the amount of overall benefit when all nodes within
coalition S will cooperate with each other.

Definition 2:
In this a coalitional game the core C(v) is the set of payoff allocation vectors x RN, s.t.
53
(1)
where xi is the payoff allocation to player i. Each player will get benefitted in same way in this game
if they work as part of coalition game or do same work individually.
C. Coalitional Game Formation in VANET Message Forwarding
We consider a VANET with a set of mobile nodes where message will get forwarded if two
nodes come within the transmission range of each other. Therefore at basic level we are considering
general routing protocol R. In this model, we are redefining routing protocol and assuming that every
node will follow only this routing protocol. In the VANET, directly messages are delivered to the
destination if destination node is within range of source node else forwarded by intermediate nodes
before reaching to their destinations. When we say message has been forwarded at that time node
may or may not keep copy of message before forwarding. In VANET the forwarding coalitional
game (N, v) starts when the message is generated by source (src) and ends when message (or its
copy) successfully received by the destination (dest) or discarded by an intermediate nodes during
communication. We can say two nodes are in coalition if communication between these two nodes
occurs the way defined in R protocol.
IV. SYSTEM ARCHITECTURE
Fig 1: System Architecture

The VANET proposed system architecture consist of smart vehicle equipped with OBU and
virtual currency center (Vcc) which act as central authority. Vcc can be nearby some infrastructure
like petrol pump or gas station. When vehicle pass nearby Vcc then communication between Vcc and
node occur. During that meeting period information about credit gets exchanged. In many other
incentive scheme Vcc has been proposed like (e.g., [13]). Some system proposed Vcc consists of
nodes virtual account which keeps the information about node virtual currency in form of credits.
Credits are unit of virtual currency and it gets distributed among node account by Vcc. It is not
compulsory for node to connect to Vcc all the time.
Initially Vcc will put equal amount of currency in each node account. If a node wants to send
message then that node become source node and where that message wants to be get deliver that
node become destination node. All other nodes which help source node to forward message to
destination node are called as intermediate nodes. If any node helped another node in forwarding the
message then for that work credit will be get putted into node and whenever node come in contact
with Vcc its account status get updated. When source node want to send message to destination, it
will predict number of intermediate node will take part in forwarding the message and it will put that
much credit in message packet. Buttyan and Hubaux in [15] proposed stimulation approach where
nuglets act as virtual currency. They also proposed two payment models which are the Packet Purse
Model and the Packet Trade Model. In the Packet Purse Model, the sender send message packet with
nuglet in it. As message get forwarded node to node nuglets share by intermediate node. But in this
method if packet runs out of nuglet then packet drops occur. In case of Packet Trade Model, packet
get buy from source node to intermediate node and get sold to next node. Such process get repeated
until packet received by destination node that means indirectly destination node pays for message
forwarding.
But in our case we are trying to avoid issue of both models. In our case when message packet
run out of credit then instead of dropping that packet it will get store in intermediate node until Vcc
arrive. In Vcc message packet get refill from source node account and continue with its journey. If
number of intermediate require is less as compared to predicted intermediate node then message
packet arrives at destination side with remain credit. At that time destination will add those extra
credits into source account whenever it comes in contact with Vcc.
Once destination node receive message next thing is acknowledgement (ACK). Destination
will send ACK to source node same way as that of message but only difference is number of credits
get share. The number of credit get node from message is always greater than ACK credit. To
implement this system smart vehicle equipped with OBU as well as a tamper-proof hardware.
Tamper-proof hardware help node in correcting the amount of message packet for addition or
deduction of credit.
Credit distribution during message forwarding also divided into 3 types. First is normal
message which will get forwarded by intermediate node and appropriate credit will be get putted into
node. Second is ACK message for which process will be same but amount of credit earn by node by
forwarding ACK is less as compared to normal message. Third is emergency message where no
credit will involve. Emergency message are higher priority message which will be get used in case of
emergency situation like accident, bridge collapse etc. Normal message and ACK will be forwarded
in end to end terminal way while emergency message will be get forwarded in flooding based way.
The work of VCC has been divided into 3 categories as follow
54
• update credit status of node:
Vcc will check node current status and update that status within its virtual account whenever
node passes nearby it.

55
• refill message packet:
If during message forwarding message packet credit becomes zero and when such message
packet node pass Vcc area at that time message packet get refill from source node account.
• Sell credit to node:
It is also possible to buy credit from Vcc using money.
A. Allocation of Payoff
We are trying to design Grand Coalition game where payoff allocation will be different for
different player node and message type. Like in case of emergency message situation priority will be
given to message rather than credit allocation. In this game all node will get different credit
allocation according to their work.
Therefore, we consider payoff should be given differently for different node as follow:
• Payoff Allocation to the Source Node:
Credit for message forwarding will get deducted from source node since its source node who
wants to send the message.
• Payoff Allocation to Intermediate Nodes:
Intermediate nodes are nodes who actually doing the work of forwarding the message.
Therefore credit distribution will actually happen in between intermediate nodes and message
packets only.
• Payoff Allocation to Destination Nodes:
Destination node will not get any credit since it is doing any work like forwarding the
message and again it is not node who initiated the message forwarding. It is just accepting
message which addressed by source node.
B. Mathematical model
At start all node will be credited by same amount by Vcc.
If (SaccMinbal) then
Obtain iddestination and set source node as idcurrent
While message not reach to destination
{ Check If (Msgcredit 0) then
{ If iddestination within range of idcurrent then send message directly to it
Update the Sacc for remaining credit
Else check Qlist for nodes with credit within range.
{ If Qlist nodes credit are same then
send message from idcurrent to idperimeter ,drop credit share to node and set
idperimeter idcurrent
Else send message from idcurrent to idthreshold drop credit share to node and
set idthreshold idcurrent
} } }
Fig 2: Protocol to balance load of Credits

In this model Sacc is Credit of source account. Minbal is minimum balance node should have to
forward message. In some cases this balance could be minimum credit required to reach message to
destination by considering predicted intermediate node. Destination address is denoted by iddestination
and current node is denoted by idcurrent . How much credit has been left in message packet is denoted
by Msgcredit.
Current node will maintain Qlist which will contain list of all nodes present within range it
with their credit information. Qlist will also contain two node information which are idperimeter and
idthreshold. idperimeter is node reside on perimeter range line of current node and towards direction of
destination node and it get selected if all nodes within range of current node having same credits.
idthreshold is node with minimum number of credit in Qlist of current node. .
In [20] credits given to node on the basis of number of receipt (meeting records) they are
having. Those receipts get submitted to Vcc and at that time node gets credits. That means node
credits information for itself cannot be updated unless and until it come in contact with Vcc. During
two Vcc connection period, node travel with its old information of credits obtained by previous Vcc.
Thus if we consider this approach for our improved LB protocol then node selection will be based on
its old credits information. But in our game each time message get forwarded by intermediate node
credits get putted into node at that time only. Therefore every time message get forwarded node itself
get updated on that time only even if does not come in contact with Vcc. Hence next time when
another node makes the decision to select node for forwarding the message, it always gets freshly
updated credit information.
Another point in [20] is that every intermediate node will get credits iff message successfully
delivered to destination. That means if message does not reach to destination all efforts made by
intermediate node to forward it get wasted. But in our system if node utilizing their resources to
forward message they will get credit for that. Therefore every node knows each time they will
forward the message, they will definitely get credits rather than indefiniteness of successful message
delivery credits allocation. Hence it leads more number of nodes to participate in game.
56
V. ANALYSIS
In analysis section we compare result of existing game theory system with our proposed
system. We use the ns-2 simulator to evaluate the performance of our algorithm in terms of the
message loss ratio, delivery rate, etc. We simulated a traffic scenario with a high vehicle density. The
VCC is located at roadside and vehicles can associate with the VCC when they come in contact with
it. The NS-2 simulation parameters are the following:
15 nodes placed at random in an area of 700m x 500m. We utilize the VANET mobility
model [20], together with real data from the Malaga, Spain database to generate the vehicles’
mobility traces. The default transmission range is 50m. The inter-vehicular distance varied from 5 m
to 50 m to simulate the scenarios with different traffic densities. We assume each vehicle contain
interface queue capable of storing maximum 60 packets in it and packets will be deleted when the
buffer is full. In all of the simulations, packets are 100 KB in size, and each signature is 128 bits.
Inter-vehicle Messages are sent by vehicle 100kbps CBR data rate at each vehicle. IEEE 802.11a is
used to simulate the medium access control layer transmission protocol as was done by [5]. The
bandwidth of the channel is 6 Mb/s.
A. Credit Load Balancing
The first aim our experiment to verify whether our incentive scheme with help of Improved
LB can balance Credits on more number of nodes or not. In this experiment we formed coalition and
assume that every node is going to follow Improved LB protocol only. We vary size of coalition
according to node behavior and to testify different scenario.

Fig. 3: Average Credits on Each node Fig. 4: Credits on Each Node at Time 42.811324
In one of the scenario we include 15 active nodes to form five coalitions, out of which three
coalitions consist of 3 nodes while other two consisting of four and two nodes respectively. We have
recorded the average number of credits on each node after 180 minutes and shown in Fig. 3. We
observe that credit has been distributed in such way that credits balance would not be concentrated to
limited number of nodes
Credit balancing among node does not occur after some periodic time. Instead every time
before message forwarding dynamic node selection occur which based on node credits condition. To
verify our system we took result of different scenario with same number of active nodes after
42.811324 minutes. Fig. 4 shows credits available on each node. From Fig. 4 it is observed that
number of credits balance has been achieved. Therefore from result we can say we are trying to
eliminate situation where interested node cannot send message due to lack of credits.
57
B. Packet Loss ratio
Fig. 5: Packet Loss Ratio vs Node per Km Fig. 6: Average Reward Vs Contribution
In application of message forwarding packet loss on the network could lead to serious
problem. In case of weighted Graph 60% message loss observe for light traffic while in other hand

for blind flooding message loss is 55%. For forwarding dilemma game message loss percentage is
low as compared other two which is 45% as shown in fig, 5.
58
C. Nodes Contribution Reward
In fig. 6 we have compare different scheme for rewarding the node according to its
contribution. Each node gets credit according to its contribution in the game. Most of scheme gets
less average credits due their scheme rule. The rules such as node will get credits if and only if
message successfully delivered to destination. But if some intermediate node helps another node to
forward message and after some time that message gets loss. As its not intermediate node fault since
intermediate node has done its work therefore there is no any reason to waste its efforts by not giving
credits to it. But in our system as contribution of node increases average rewards also increases since
we are giving credits during message forward process only.
D. Success Probability
Fig. 7: Success Probability VS Rewards Fig. 8: Delivery Ratio for different scheme
Fig. 7 shows success probability of Evolutionary Game [6] and Improved LB. The Success
probability for single species in evolutionary game keeps increasing when rewards increased and
when reward (G) become 2.4 its success probability reach to 1. That means as rewards increases,
number of nodes quantity to acquire credits also increases. On other hand success probability for
improved LB from very start almost 1 because in Improved LB protocol we are selecting needy node
(node with less number of credits) out of interested node. Therefore to get credits, chances of needy
node to forward that message become very high.
E. Delivery Ratio
Now for next experiment of delivery rate percentage we changed message forwarding
methods such as Frame, receipt counting and proportional and compare those result with Improved
LB. From fig. 8 the delivery ratio is higher in Improved LB as compared to other three methods. This
proves the applicability of the Improved LB scheme.

Fig. 9: Delivery Ratio for different protocol Fig. 10: Time to compute credit on Vcc
In Fig. 9 we have compared delivery ratio of Improved LB protocol with existing DSR and
AODV protocols. By observing Fig .11 we can say that Due to proper selection for message
forwarding in Improved LB its performance increase dramatically.
59
F. Credit Computation on VCC
Fig. 10 shows time required to compute credits on VCC. In [20] receipt get submitted to VCC
and after calculating number of receipts node gets credits. Therefore more time get wasted on
calculating receipt and then distribute credits to each cooperative node. On other hand in our system
VCC does not have overhead of receipt calculation or credits distribution. Since most of Credits
distribution happens during communication only. Thus time required to process credits information
on VCC in our Improved LB coalition Game is very low as compared to normal coalition.
VI. CONCLUSION
In our proposed algorithm we are overcoming drawback of receipt depend credit based
system where node get credit iff message successfully deliver to destination. If due to some reason
message gets drop then all effort done by intermediate node just go to waste. Again receipt credit
based system node store receipt and when Vcc come in contact then account gets updated. But in
load balancing forwarding message to node with less number of credits, this decision has done at
dynamic time. Therefore this decision cannot wait Vcc availability and its updation.
In our proposed system we are motivating more nodes to participate in the game by their own
will to gain credits. But at same time we are restricting more selfish natured nodes from getting more
credits to achieve credit load balancing. Again our system performance does not reduce even if
quantity of number of nodes increases.
VII. REFERENCES
[1] E. Jones, L. Li, and P. Ward, “Practical routing in delay-tolerant networks,” in Proc. ACM
Chants Workshop, Aug. 2005, pp. 237–243.
[2] S. Jain, K. Fall, and R. Patra, “Routing in a delay tolerant network,” in Proc. SIGCOMM,
2004, pp. 145–158.
[3] A. Vahdat and D. Becker, “Epidemic routing for partially connected ad hoc networks,” Duke
Univ., Durham, NC, Tech. Rep. CS-200006, Apr. 2000.

[4] B. Burns, O. Brock, and B. N. Levine, “MV routing and capacity building in disruption
tolerant networks,” in Proc. INFOCOM, Mar. 2005, pp. 398–408.
[5] T. Spyropoulos, K. Psounis, and C. S. Raghavendra, “Spray and wait: An efficient routing
scheme for intermittently connected mobile networks,” in Proc. WDTN, Aug. 2005,
pp. 252–259.
[6] Di Wu, Jing Cao, Yan Ling, Jiang chuan Liu, Limin Sun. ‘Routing algorithm based on multi
community evolutionary game for VANET’ in JOURNAL OF NETWORKS, VOL. 7, NO. 7,
JULY 2012
[7] S. Marti, T. Giuli, K. Lai, and M. Baker, “Mitigating routing misbehavior in mobile ad hoc
networks,” in Proc. MOBICOM, Boston, MA, Aug. 2000, pp. 255–265.
[8] P. Michiardi and R. Molva, “CORE: A collaborative reputation mechanism to enforce node
cooperation in mobile ad hoc networks,” in Proc. CMS, Portoroz, Slovenia, Sep. 2002,
pp. 107–121.
[10] U. Shevade, H. H. Song, L. Qiu, and Y. Zhang, “Incentive-aware routing in DTNs,” in Proc.
60
ICNP, Oct. 2008, pp. 238–247.
[11] Filipe Maciel Roberto, Joaquim Celestino J´unior, Henning Schulzrinne “Using a symmetric
game based in volunteer’s dilemma to improve Vanets multihop broadcast communication”.
2011 IEEE 22nd International Symposium on Personal, Indoor and Mobile Radio
Communications
[12] J. J. Jaramillo and R. Srikant, “DARWIN: Distributed and adaptive reputation mechanism for
wireless ad-hoc networks,” in Proc. MOBICOM, Montreal, QC, Canada, Sep. 2008,
pp. 87–98P.
[13] S. Lee, G. Pan, J. Park, M. Gerla, and S. Lu, “Secure incentives for commercial ad
dissemination in vehicular networks,” in Proc. MOBIHOC, Sep. 2007, pp. 150–159.
[14] F. Wu, T. Chen, S. Zhong, L. Li, and Y. R. Yang, “Incentive-compatible opportunistic
routing for wireless networks,” in Proc. MOBICOM, San Francisco, CA, Sep. 2008,
pp. 303–314.
[15] L. Buttyan and J. P. Hubaux, “Enforcing service availability in mobile ad-hoc WANs,”
IEEE/ACM Workshop on Mobile Ad Hoc Networking and Computing (MobiHOC), Boston,
MA, August 2000. [Online]. Available:
http://icawww.epfl.ch/Publications/Buttyan/ButtyanH00.ps
[16] J. Burgess, B. Gallagher, D. Jensen, and B. N. Levine, “MaxProp: Routing for vehicle-based
disruption-tolerant networks,” in Proc. INFOCOM, Apr. 2006, pp. 1–11.
[18] S. Jain, K. Fall, and R. Patra, “Routing in a delay tolerant network,” in Proc. SIGCOMM,
2004, pp. 145–158.
[19] E. Jones, L. Li, and P. Ward, “Practical routing in delay-tolerant networks,” in Proc. ACM
Chants Workshop, Aug. 2005, pp. 237–243.
[20] Tingting Chen, Liehuang Zhu, Fan Wu,Sheng Zhong, “Stimulating Cooperation in Vehicular
Ad Hoc Networks: A Coalitional Game Theoretic Approach”, IEEE Transactions on
Vehicular Technology, Vol. 60, No. 2, February 2011.
[21] Priti Bhardwaj and Rahul Johari, “Routing in Delay Tolerant Network using Genetic
Algorithm”, International Journal of Computer Engineering Technology (IJCET),
Volume 4, Issue 2, 2013, pp. 590 - 597, ISSN Print: 0976 – 6367, ISSN Online: 0976 – 6375.

Coalitional game theoretic approach for ieee 802

More Related Content

Viewers also liked (9)

Similar to Coalitional game theoretic approach for ieee 802 (20)

More from IAEME Publication (20)

Recently uploaded (20)

Coalitional game theoretic approach for ieee 802