1

Performance Analysis of the Vehicular Delay

Tolerant Network
Dusit Niyato∗ , Ping Wang∗ , and Joseph Chee Ming Teo†
∗ School of Computer Engineering, Nanyang Technological University (NTU), Singapore
† Institute for Infocomm Research, Singapore

Abstract—Delay tolerant network (DTN) based on vehicular sources in the DTN under noncooperative environment was
communications (i.e., vehicular delay tolerant network or VDTN) never considered in the literature.
is considered in this paper. In VDTN, there is no direct end-to- In this paper, we consider delay tolerant network based on
end connection between the source and destination (i.e., sink). In
this case, the traffic source transmits data to a mobile router. This the vehicular communications which is referred to as vehicular
mobile router in a vehicle receives and stores data in a buffer. delay tolerant network (VDTN). In VDTN, the traffic sources
The vehicle can move and once it is in the transmission range, the and sinks do not have a direct transmission path among each
data in a buffer is forwarded to the sink. The buffer management other. However, a vehicle is used as a carrier to carry the
and its queueing model are proposed to analyze the performances data from traffic sources to sinks. When a vehicle moves
of a mobile router in this vehicular delay tolerant network.
With this analytical model, various performance measures (e.g., into the transmission range of traffic source, the roadside
throughput and delay) can be obtained. This queueing model unit of traffic source transmits the data to a mobile router
is then used as a tool to study the behavior of the traffic deployed in that vehicle. This vehicle can travels among
source in a competitive environment which is due to the fact places and again once it moves into the transmission range
that the transmission resources of a mobile router are shared of the sink, a mobile router transmits data in its buffer to
among multiple traffic sources. Therefore, the traffic sources
have to noncooperatively optimize their transmission strategies the roadside unit of the sink. Note that this packet delivery
to achieve the highest utility. This queueing model formulation scheme is similar to the core-aided scheme in [7], in which
will be useful to investigate the performance and behavior of the the packet is stored and forwarded by the core node (i.e.,
vehicular delay tolerant network. a mobile router). The performances of a mobile router in a
Keywords-Delay tolerant network, queueing analysis. vehicle is studied by formulating the queueing model. With
this analytical model, various performances of the mobile
router to forward data from the traffic sources to the sink can
I. I NTRODUCTION be obtained. Then, since the data from multiple traffic sources
share the resources (i.e., buffer and transmission time) of a
Delay or disruption tolerant network (DTN) has been in-
mobile router, the transmission parameter (i.e., transmission
troduced for the situation where the connection between the
rate or transmission probability) of one traffic source will
nodes is sparse. As a result, unlike traditional mobile ad hoc
affect the performance and hence the utility of other traffic
network (MANET), the end-to-end path between a source and
sources (i.e., opponents). We study this competitive situation
destination (e.g., gateway or sink) will only be available for
by formulating a noncooperative game model. The solution of
a brief and unpredictable period of time. DTN has gained
this game is obtained as the Nash equilibrium which ensures
research attention due to its variety of applications in military,
that none of the traffic source will change the strategy, given a
disaster discovery, and emergency response systems where
set of strategies of other traffic sources. The queueing model
the communication infrastructure may not exist. In DTN, the
for a mobile router and the noncooperative game model for
data from the source is relayed through mobile nodes to the
the traffic sources will be useful to analyze the performance
destination. However, when the connection to other nodes
and behavior of the independent entity in the vehicular delay
is not available, the data is stored in the buffer. Once the
tolerant network.
mobile node travels into the transmission range of other nodes,
the data in a buffer can be forwarded to the next hop. In
this way, the packet can reach the destination by hopping II. R ELATED W ORK
over the mobile nodes even though the connection between Routing is one of the important issues in delay tolerant
the mobile nodes may not be always available. The major network (DTN), [1], [2], [3]. The major objectives of routing
research issues in DTN are the routing protocol [1], [2], [3], in DTN is to deliver packets from the source to the destination
the investigation of DTN in various applications [4], [5], [6], by means of mobility of the node. Since the end-to-end path
and the performance analysis [7], [8], [9], [10]. However, the may not be available, routing schemes have be optimized the
current literature on DTN research does not consider the buffer data dissemination by utilizing the connectivity information
management mechanism nor queueing model for a mobile and network conditions maintained by each node. For exam-
router. More importantly, the interaction among the traffic ple, in [1], the routing scheme based on the estimates of the
average inter-contact time between the mobile nodes in the
This work was supported by a grant from the Nanyang Institute of
Technology, Singapore, and in part by the Natural Sciences and Engineering network was proposed. This routing scheme was designed to
Research Council (NSERC) of Canada. minimize the packet delivery time. The routing properties in
 2

terms of loop-free forwarding and polynomial convergence III. S YSTEM M ODEL

were studied which ensure the performances of the packet We consider a network which is composed of I locations. In
delivery in DTN. In [2], the packet delivery scheme based this network, there are multiple traffic sources sending data to
on the super node architecture and epidemic routing was the sinks. However, there is no direct connection between the
introduced. With epidemic routing, the packet is forwarded traffic source and sink. In this case, there are total N mobile
to other contacted nodes (i.e., nodes with direct connection). routers, each of which is installed on a vehicle, in this network.
Unlike traditional epidemic routing, the packet is forwarded A mobile router receives data from the traffic sources. This
to the super nodes to improve the performance and reduce data is stored in a buffer. Then, once the vehicle moves into the
overhead. The super nodes are then responsible to carry the transmission range of a sink, the data is retrieved from a buffer
packet to the destination. In [3], the routing in DTN based on and forwarded to the sink. We assume that the vehicle moves
thermodynamics was presented. In this scheme, the concept among locations in a network randomly with random speed.
of temperature is applied to each node to measure the amount The traffic sources and sinks are connected with the roadside
of data that can be delivered to the destination. The control of units communicating with a mobile router (Fig. 1). A mobile
temperature (i.e., packet forwarding) is based on the principles router uses a single queue to buffer data delivering to the
that govern the heat transfer between object. For example, the same sink regardless the traffic source. In particular, the data
amount of heat (i.e., packets) to be transferred between two of multiple traffic sources whose destination is sink i0 (i.e., a
nodes is proportional to the difference in their temperature. roadside unit of sink i0 is at location si0 ) is aggregated into the
With this thermodynamics inspired routing scheme, the rate same queue. The transmission between the roadside unit and
of successful packet delivery can be significantly improved. mobile router on the vehicle is time-slot based. The successful
DTN was applied to many applications (e.g., [4], [5], transmission between a traffic source, mobile router n, and a
[6]). In [4], vehicle-to-vehicle (V2V) communication based (n)
sink at any location si is denoted by µi for i ∈ {1, . . . , I}
on DTN was proposed. The issues of V2V communication and n ∈ {1, . . . , N }.
in high and predictable mobility, the large scale network,
and partitioned network were discussed when applying DTN
Traffic
routing and packet delivery policy. Also, the experimental source
Traffic s3
platform was developed. In [5], the sensor network based
source
on DTN was proposed. This sensor network based on DTN
was applied to monitor lake water quality in rural lake
and noise level logging in urban areas and along highways. s2 s4 s5
Again, the prototype for these applications was developed. Sink
In [6], the application of DTN in business application was
Traffic
reported and a coordination protocol was developed. Such source s1
a protocol can be used for financial application in a mobile
network environment. With DTN architecture, the distributed Fig. 1. System model for the MMR network.
transaction processing and management can be supported even
though the connection among business entities is not always Note that this system model can be applied to the applica-
available. tion of environment monitoring. In this case, a data source
To understand the performance of DTN under node mobil- is a sensor node. The vehicle of the observer visits these
ity, the analytical models were proposed (e.g., [7], [8], [9], sensors occasionally. The data collected from the sensors is
[10]). For examples, in [7], the mobility model for a mobile transported to the data center.
node in DTN was proposed. Given this mobility model, the
packet delivery ratio and delay were analyzed. In addition, IV. Q UEUEING A NALYSIS OF V EHICULAR D ELAY
the queueing model of the traffic source was formulated. T OLERANT N ETWORK ROUTER
In [8], the performance of epidemic routing scheme was
In this section, the queueing model based on discrete-
studied using continuous-time Markov chain. The state of the
time Markov chain is formulated for a particular buffer in a
system is the number of nodes in the network having the
mobile router of a vehicle. We first define the state space. The
packet. The packet reaches the sink which is considered to be
probability transition matrix is derived. The method to obtain
the absorbing states in the model. The delay of the packet
the steady state probability is presented, and the performance
delivery was derived. In [9], the performances of homing
measures are obtained.
pigeon based DTN were analyzed. In such a network, each
node has a dedicated messenger (i.e., pigeon). This messenger
periodically carries a batch of packet from the home node (i.e., A. State Space and Transition Matrix
source node), delivers to the destinations, and returns home. Let S denote a set of the locations in a network. The state
The travel of a messenger has the fixed route and speed. The space of a queue in a mobile router buffering the packet with
analytical model was developed to obtain the waiting time of destination at sink i0 can be expressed as follows:
the individual and bulk packets. However, none of the above-
(n)
mentioned work considered the queue at a mobile router. Ψi0 = {(S, X ); S ∈ S, X ∈ {0, 1, . . .}} (1)
 3

where S is the location of a vehicle and X is the number of B. Transition Matrix

packets in a queue. The state transition matrix of a queue in a mobile router is
At a traffic source i, the packets are transmitted to a denoted by P which is defined as follows:
mobile router n when a vehicle is in the transmission range  
of a roadside unit of a traffic source at location D E
h i si . Let  F1 G1 H1 
(n) (n) (n) (n)
di→i0 = di→i0 ,0 · · · di→i0 ,k · · · di→i0 ,D denote a  
P= F 2 G 2 H 2
. (5)
row matrix of packet transmission probability by a traffic  
.. .. ..
source i to a mobile router n with a destination sink at . . .
location si0 (i.e,. sink i0 in short). D is the maximum batch
(n) where
size where k = {0, . . . , D}. The element of this matrix di→i0  
can be obtained, for example, from the probability of using p̃0,0 ··· p0,D−1
 .. .. .. 
different modulation and coding rate [11]. At a mobile router D =  . . . 
n, the packet arrival probability of a queue for the packet with p̃D−1,0 ··· pD−1,D−1
destination at sink i0 can be obtained from  
p0,D 0 0
XD µ 0 ¶  .. .. 
(n) (n) k (n) (n) 0 E =  . . 0 
ai→i0 ,k = di→i0 ,k0 (µi )k (1 − µi )k −k (2)
0
k pD−1,D · · · pD−1,2D−1
k =k
 
pkD,(k−1)D · · · pkD,kD−1
for k = {0, 1, . . . , D}. The average transmission rate of a  .. 
Fk =  .. 
traffic source i to a mobile router P n with destination sink 0 . .
(n) D (n)
i0 can be obtained from λi→i0 = k=1 kai→i0 ,k and total 0 0 p(k+1)D−1,kD−1
PN (n)  
transmission rate is λi→i0 = n=1 λi→i0 , where N is the
pkD,kD ··· pkD,(k+1)D−1
total number of vehicles in the network. The traffic source  .. .. .. 
Gk =  . . . 
can control this transmission rate to achieve the highest utility.
p(k+1)D−1,kD · · · p(k+1)D−1,(k+1)D−1
In this regard, the equilibrium transmission rates will be  
obtained from a noncooperative game formulation which will pkD,(k+1)D 0 0
 .. .. 
be presented later in this paper. Hk =  . . 0 
Similarly, the departure probability of a queue in a mobile p(k+1)D−1,(k+1)D · · · p(k+1)D−1,(k+2)D−1
router n transmitting to a sink i0 can be obtained from
where each row of this matrix P indicates the number of
D
X µ ¶ packets in queue. The element px,x0 denotes the probability
(n) (n) k0 (n) (n) 0
ri0 ,k = ei0 ,k0 (µi0 )k (1 − µi0 )k −k (3) matrix for the number of packets in queue changes from x in
k
k0 =k the current time slot to x0 in the next time slot. This matrix
(n) can be defined as follows:
where ei0 ,k is the probability of transmitting k packets by a  
mobile router in a time slot. p1,1 (x, x0 )
 .. 
For a queue buffering the packet with destination sink i0 px,x0 = M(n)  .  (6)
of a mobile router n, the state representing the location in a pI,I (x, x0 )
network can be divided into three subsets. These subsets are a
subset of locations of traffic sources, a subset of location of a where each row of this matrix px,x0 indicates the location of
sink, and a subset of locations without traffic source and sink a vehicle. The element of this matrix px,x0 can be obtained
which are respectively denoted by Ssour sink road from
n,i0 , Sn,i0 , and Sn,i0 . For 
(n)
an example of network in Fig. 1, we have s1 , s2 , s3 ∈ Ssour 1,5 , 
 ai→i0 ,x0 −x , i ∈ Ssour 0
n,i0 , x ≥ x
s4 ∈ Sroad
1,5 , and s 5 ∈ S sink
1,5 for mobile router n = 1 and sink 0
pi,i (x, x ) = r
(n)
0, i ∈ Sn,i0 , x ≥ x0
sink (7)
i0 = 5. The mobility of a vehicle with mobile router n among 
 i,x−x
0, otherwise.
the locations in a network can be modeled using transition
matrix M(n) which is defined as follows: For matrix p̃x,0 which captures the case that the number of
 (n) (n)
 packets in queue is smaller than the transmission rate of a
m1,1 · · · m1,I mobile router. The element p̃x,0 of matrix D is denoted by
 . .. 
M(n) =   .. . 
 (4) p̃i,i (x, 0) where
(n) (n) D
mI,1 · · · mI,I X (n)
p̃i,i (x, 0) = ri,k , for i ∈ Ssink
n,i0 . (8)
where I = |S| is the total number of locations in a network, k=x
where |S| is the cardinality of set S (e.g., I = 5 for a network Note that F1 = · · · = Fk = · · · , G1 = · · · = Gk = · · · ,
(n) PI (n)
shown in Fig. 1). mi,î for î=1 mi,î = 1 denotes the and H1 = · · · = Hk = · · · . The stability of a queue can be
probability of a vehicle n staying at location si in the current determined from the condition θHk 1 < θFk 1, where θ =
time slot and moving to location sî in the next time slot. θK, θ1 = 1, K = Fk + Gk + Hk , and 1 is a row matrix of 1
 4

with a proper size. If the stability condition is satisfied, there D. Queueing Performances
is a matrix R which is the minimal non-negative solution of We consider the networks as shown in Fig. 1. The maximum
R = Fk + RGk + R Hk 2
(9) transmission range between the roadside unit and mobile
router is 125 meters. The typical vehicle speed is 45km/h. The
such that successful transmission probability is µ = 0.95. We assume
π i+1 = π i R. (10) that a single rate transmission is used by a traffic source and
a mobile router (i.e., D = 1). In this regard, a traffic source
This matrix R can be obtained iteratively by using (n)
chooses a transmission probability in which di→i0 ,1 = λi
(n)

(n) (n)
R(t + 1) = Ft + R(t)Gk + R2 (t)Hk (11) and di→i0 ,0 = 1 − λi for i0 = 5.
For a network model shown in Fig. 1, the mobility transition
until the condition maxi,j |Ri,j (t + 1) − Ri,j (t)| < ² is matrix can be expressed as follows:
satisfied. Here, Ri,j (t) is the element of matrix R(t) and ²  (n) 
(n)
is a small number (e.g., ² = 10−9 ). Next, the matrices of m1,1 0 0 1 − m1,1 0
 (n) (n) 
steady state probabilities π 0 and π 1 are obtained by solving  0 m2,2 0 1 − m2,2 0 
 
the following equations: M(n) =  (n) (n)
0 
· ¸  0 0 m3,3 1 − m3,3
P .
 (n) (n) (n) (n) (n) 
D E m
 4,1 m 4,2 m 4,3 1 − i6=4 m 4,i m 4,5 
B(R) = (12)
Fk Gk + RHk 0 0 0 1 − m5,5
(n) (n)
m5,5
[π 0 , π 1 ] = [π 0 , π 1 ] B(R) (13) (n)
−1
1 = π 0 1 + π 1 (I − R) 1 (14) With an average vehicle speed vi at location siµ, the prob-
¶
(n)
(n) (n) vi T
ability mi,i can be obtained from mi,i = min 1, 2Ei
where I is an identity matrix with a proper size. The steady
state probability of a queue with x packets when a vehicle is where Ei is the transmission range and T is the length of time
at location si can be obtained from slot. For the numerical results, we assume that a vehicle travels
£ ¤ to locations (i.e., states) s1 , s2 , s3 , and s5 with probabilities
π(i, x) = π bx/Dc xD+i (15) 0.09, 0.12, 0.15, and 0.64, respectively.
Fig. 2 shows an average packet delivery delay under dif-
which is the xD + i element of a row matrix π bx/Dc .
ferent vehicle speeds. As expected, as the speeds of vehicle
increases, the duration of a mobile router without connection
C. Performance Measures to the traffic sources and a sink becomes shorter. Conse-
1) Average Number of Packets in Queue of a Mobile quently, the packet delivery delay decreases. Also, as the
Router: The average number of packets in queue in a mobile transmission probabilities of all traffic sources increase (i.e.,
router n for the packet with destination sink i0 can be obtained 0.015 → 0.020), the average packet delivery delay increases
from since there is more average number of packets in a queue of
à I !
(n)
XX X a mobile router. For brevity of the paper, we omit the plot of
xi0 = x π(i, x) (16) queueing performances whose results are expected (e.g., as the
x=1 i=1 transmission probability increases, the performances degrade,
where and so on).
PX Pthis I
calculation is truncated at X for which
x=1 i=1 π(i, x) < ².
2) Throughput: The throughput of a traffic source i trans- 520
(1)
λi =0.015
mitting packet to a mobile router n to the destination sink i0
Avearge delay (seconds)

500
λ(1)=0.020
can be obtained from 480 i

(n) (n) (n) 460

τi→i0 = φi λi→i0 , for i ∈ Ssour
n,i0 (17)
440
(n)
where φi is the probability that a vehicle n is at location si . 420
(n) (n)
This probability φi can h be obtained by solving φ = φM i
400

(n) (n) (n) 40 50 60 70 80 90 100

and φ1 = 1, and φ = φ1 · · · φi · · · φI . The Speed of vehicle (km/h)

total queue throughput is Fig. 2. Packet delivery delay under different speeds of vehicle.
(n)
X (n) (n)
τi0 = φi λi→i0 . (18)
i∈Ssour
n,i0
V. C OMPETITION AMONG T RAFFIC S OURCE
3) Average Packet Delivery Delay: The average delay can In this section, the competition among traffic sources to
be obtained using the Little’s law as follows: share the resource of a mobile router to deliver their packets
(n)
xi0 to the sinks is formulated as a noncooperative game. The
(n)
w i0 = (n)
. (19) definitions of the utility and the Nash equilibrium solution
τi0 for the strategies of the traffic sources are presented.
 5

A. Noncooperative Game Formulation congested and the increase in cost is at a higher rate than that
The noncooperative game formulation of this network can of throughput. The highest point of the utility is defined as
be described as follows. The players of this game are the the best response of a traffic source 1 given the strategies
(1) (1)
traffic sources. The strategy of each player is a set of the λ2 and λ3 of the opponents. This best response is different
transmission probabilities to the different mobile routers. The for the different transmission duration of a vehicle at a sink.
payoff is defined as the utility which is the function of the In other woard, as a vehicle stops at a sink longer, more
throughput and cost as follows: number of packets can be forwarded from a mobile router to a
sink. Therefore, a traffic source can increases its transmission
N X³
X ³ ´
(n) probability to achieve higher utility.
Ui (λi , λ−i ) = u1 log 1 + u2 τi→i0 (λ)
n=1 i0
´ 17.55
(n)
−u3 C (xi0 (λ)) (20)
17.5

where u1 , u2 , and u3 are the constants. λi is a set of strategies 17.45 Best response of
of a traffic source hi for all mobile routers and all destination
i
traffic source 1

Utility
sinks, i.e., λi = λ(1) (n) (N ) 17.4
1 · · · λ i0 · · · λ I
. λ−i is
a set of strategies of all other opponents, and λ is a set of 17.35

strategies of all traffic sources. C (x) is the cost function of a 17.3 Transmission duration at sink 200.00 seconds
traffic source (e.g., price charging by a mobile router or delay). Transmission duration at sink 166.67 seconds
17.25
In this paper, we assume that this cost is a linear function of 0.04 0.045 0.05 0.055 0.06 0.065 0.07
λ(1)
the average number of packets in queue which is a function 1

(n)
of the strategies of all traffic sources (i.g., xi0 (λ)). Fig. 3. Utility under different strategy (i.e., packet transmission probability).
If all traffic sources are rational to maximize their utilities,
the Nash equilibrium can be considered as the solution for Similarly, Fig. 4 shows the total utility of a traffic source 1
the transmission strategies. The Nash equilibrium of a non- when there are two independent vehicles in a network. Note
cooperative game is a set of strategies λ∗i with the property that the mobility model of the second vehicle is similar to that
that no player can increase his payoff by choosing a different of the first vehicle (i.e., as in (20)). However, the probability of
action, given other players’ sets of strategies λ∗−i . That is, going to the locations is different. In this case, a roadside unit
of a traffic hsource 1 can adjust
i the transmission probabilities
Ui (λ∗i , λ∗−i ) ≥ Ui (λi , λ∗−i ), ∀i. (21) (1) (2)
(i.e., λ1 = λ1 λ1 ) for the mobile routers in both ve-
The Nash equilibrium can be obtained from best response of hicles independently. Again, there is a certain set of strategies
each player. This best response is defined as an optimal set of such that a traffic source 1 can achieve the highest total utility
strategies of a particular player given the strategies of other and this point is the best responseh given a set of strategies from
i
players. The best response of a traffic source i is defined as other opponents (i.e., λ−i=1 = λ(1) 2
(2)
λ2 λ3
(1) (2)
λ3 ).
follows: This best response can be used to obtain the Nash equilibrium
λ∗i = BR i (λ∗−i ) = arg max Ui (λi , λ−i ) (22) for the transmission probabilities of all traffic sources in the
λi network. Note that to simplify the presentation, the following
However, since the utility which is a function of the through- results consider the network with a single vehicle. However,
put and the average number of packets in queue (i.e., τi→i0
(n) the similar results are expected for the network with multiple
(n) vehicles.
and xi0 ) are obtained from the queueing model, numerical
The best responses of the traffic sources 1 and 2 are shown
method is applied to obtain the best response of each traffic
in Fig. 5. As expected, when one traffic source changes the
source. The Nash equilibrium is considered to be the solution
strategy, the best responses of other traffic sources change.
of the following optimization formulation
For example, as a traffic source 1 increases a transmission
I
X probability, the best strategies of other traffic sources are also
¯ ¡£ ¤¢¯
Minimize: ¯λi − BR i ··· λî ··· ¯ (23) to increase the transmission probabilities. In this case, an
i=1 intersection of all best responses is the point where the Nash
for î 6= i. equilibrium is located. Similarly, as the transmission duration
of a vehicle increases, the transmission probabilities of the
traffic sources increases due to the longer transmission time of
B. Numerical Results a mobile router. Consequently, the throughput increases which
We consider the case of a single vehicle and a single sink is shown in Fig. 6. We also observe that the throughput of a
as shown in Fig. 1. Fig. 3 shows the utility of a traffic source traffic source 1 is higher than that of traffic source 2 and both
(1) (1)
1 where λ2 = λ3 = 0.04. As the transmission probability are higher than that of traffic source 3. Since a vehicle travels
increases, the utility of a traffic source 1 first increases due to to location s1 with the lowest probability, a traffic source 1
(1)
the achievable throughput τ1→5 . However, at a certain point, has to transmits more data to a mobile router to achieve the
the utility decreases, since a queue of a mobile router becomes highest utility. The same reason is also applied to a traffic
 6

VI. S UMMARY
We have considered the vehicular delay tolerant network
Best response of traffic source 1 with two vehicles
32.65
(VDTN). A vehicles has the mobile router communicating
32.6 with roadside unit to receive data from the traffic sources and
deliver to the sink. First, the queueing model to investigate
32.55 the performances of a mobile router has been presented. With
Total utility

32.5
this queueing model, various performance measures, e.g.,
average number of packets in a queue of a mobile router,
32.45 throughput, and average delay, can be obtained considering
the mobility model of the vehicle. Then, we have considered
32.4
the noncooperative environment of the traffic sources. In such
32.35
0.08 0.07
0.08 an environment, the utility of a traffic source is defined based
0.07
0.06
0.05 0.05
0.06
on the throughput and the cost of packet delivery which is
0.04 0.04
λ(2)
(1)
λ1 a function of the number of packets in queue of a mobile
1
router. The noncoopertive game has been formulated. The
Fig. 4. Utility of a traffic source 1 with two vehicles in a network. Nash equilibrium for the sets of strategies of the traffic sources
has been considered as the solution of this game. For the future
work, the property of the Nash equilibrium, i.e., existence and
source 2 whose throughput is higher than that of a traffic uniqueness will be analytically investigated.
source 3 (i.e., probability of a vehicle to visit location s2 is
larger than that of location s3 , but smaller than that of location R EFERENCES
s1 ). [1] V. Conan, J. Leguay, and T. Friedman, “Fixed point opportunistic
routing in delay tolerant networks,” IEEE Journal on Selected Areas in
Communications, vol. 26, pp. 773-782, June 2008.
[2] H. Samuel, W. Zhuang, and B. Preiss, “Routing over Interconnected
0.06 Heterogeneous Wireless Networks with Intermittent Connections,” in
BR1 Proceedings of IEEE International Conference on Communications
BR2 (ICC), pp. 2282-2286, May 2008.
0.055
[3] M. Kalantari and R. J. La, “A DTN packet forwarding scheme inspired by
Transmission duration at
a sink = 166.67 seconds
thermodynamics,” in Proceedings of Annual Conference on Information
0.05
Sciences and Systems (CISS), pp. 1216-1221, March 2008.
[4] L. Franck and F. G.-Castineira, “Using Delay Tolerant Networks for
Car2Car communications,” in Proceedings of IEEE International Sym-
posium on Industrial Electronics (ISIE), pp. 2573-2578, June 2007.
(1)

0.045
λ2

[5] P. McDonald, D. Geraghty, I. Humphreys, S. Farrell, and V. Cahill,

“Sensor Network with Delay Tolerance (SeNDT),” in Proceedings of
0.04 International Conference on Computer Communications and Networks
(ICCCN), pp. 1333-1338, August 2007.
Nash Transmission duration at [6] I. Carreras, J. Rana, and L. Telesca, “Coordination protocol for financial
0.035 a sink = 200 seconds
equilibrium application in delay tolerant networks,” in Proceedings of International
Conference on Communication Systems Software and Middleware and
0.03 Workshops (COMSWARE), pp. 283-286, January 2008.
0.045 0.05 0.055 0.06 0.065 0.07 [7] M. Abdulla and R. Simon, “The Impact of the Mobility Model on Delay
λ(1)
1 Tolerant Networking Performance Analysis,” in Proceedings of Annual
Simulation Symposium (ANSS), pp. 177-184, March 2007.
Fig. 5. Best responses in terms of packet transmission probabilities of the [8] Y. K. Ip, W.-C. Lau, and O.-C. Yue, “Performance Modeling of Epidemic
traffic sources 1 and 2. Routing with Heterogeneous Node Types,” in Proceedings of IEEE
International Conference on Communications (ICC), pp. 219-224, May
2008.
[9] H. Guo, J. Li, A. N. Washington, C. Liu, M. Alfred, R. Goel, L. Burge,
P. Keiller, “Performance Analysis of Homing Pigeon based Delay
0.08 Tolerant Networks,” in Proceedings of IEEE Military Communications
(1)
τ1 −> i’ Conference (MILCOM), October 2007.
[10] A. Krifa, C. Baraka, and T. Spyropoulos, “Optimal Buffer Management
Throughput (packets/second)

0.07 τ(1)
2 −> i’ Policies for Delay Tolerant Networks,” in Proceedings of IEEE Communi-
τ(1) cations Society Conference on Sensor, Mesh and Ad Hoc Communications
3 −> i’
0.06
and Networks (SECON), pp. 260-268, June 2008.
[11] Q. Liu, S. Zhou, and G. B. Giannakis, “Queuing with adaptive modu-
0.05
lation and coding over wireless links: cross-Layer analysis and design,”
IEEE Transactions on Wireless Communications, vol. 4, no. 3, pp. 1142-
0.04 1153, May 2005.

0.03
150 200 250
Transmission duration at a sink (seconds)

Fig. 6. Adaptation of the Nash equilibrium.