IRJET- An Incentive Framework for Cellular Traffic Offloading

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 09 | Sep 2018 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 6842
AN INCENTIVE FRAMEWORK FOR CELLULAR TRAFFIC OFFLOADING
Sindhu Preethini R1, Varnika R S 2, Nandhini R3
123Computer Science and Technology, Bannari Amman Institute Of Technology, Sathyamangalam,Erode
ABSTRACT - Cellular networks (e.g., 4G) are currently facing severe traffic overload problems
caused by excessive traffic demands. Offloading part of the cellular traffic through other forms of
networks, such as Delay Tolerant Networks (DTNs) and WiFi hotspots, is a promising solution. However,
since these networks can only provide intermittent connectivity to mobile users, utilizing them for
cellular traffic offloading may result in a non-negligible delay. As the delay increases, the users’
satisfaction decreases. In this paper, we investigate the tradeoff between the amount of traffic being
offloaded and the users’ satisfaction. We provide a novel incentive framework to motivate users to
leverage their delay tolerance for cellular traffic offloading. To minimize the incentive cost given an
offloading target, users with high delay tolerance and large offloading potential should be prioritized for
traffic offloading. To effectively capture the dynamic characteristics of users’ delay tolerance, our
incentive framework is based on reverse auction to let users proactively express their delay tolerance by
submitting bids. We further illustrate how to predict the offloading potential of the users by using
stochastic analysis for both DTN and WiFi cases. Extensive trace-driven simulations verify the efficiency
of our incentive framework for cellular traffic offloading.
1. INTRODUCTION
The recent popularization of cellular networks
(e.g., 4G) provide mobile users with ubiquitous
Internet access. However, the explosive growth
of user population and their demands for
bandwidth-eager multimedia content raise big
challenges to the cellular networks. A huge
amount of cellular data traffic has been
generated by mobile users, which exceeds the
capacity of cellular network and hence
deteriorates the network quality. To address
such challenges, the most straightforward
solution is to increase the capacity of cellular
networks, which however is expensive and
inefficient. Some researchers studied on how to
select a small part of key locations to realize
capacity upgrade, and shift traffic to them by
exploiting user delay tolerance. Remaining the
capacity of cellular networks unchanged,
offloading part of cellular traffic to other
coexisting networks would be another desirable
and promising approach to solve the overload
problem. Some recent research efforts have been
focusing on offloading cellular traffic to other
forms of networks, such as DTNs and WiFi
hotspots, and they generally focus on
maximizing the amount of cellular traffic that can
be offloaded. In most cases, due to user mobility,
these networks available for cellular traffic
offloading only provide intermittent and
opportunistic network connectivity to the users,
and the traffic offloading hence results in non-
negligible data downloading delay. In general,
more offloading opportunities may appear by
requesting the mobile users to wait for a longer
time before actually downloading the data from
the cellular networks, but this will also make the

users become more impatient and hence reduce
their satisfaction.
In this paper, we focus on investigating the
tradeoff between the amount of traffic being
offloaded and the users’ satisfaction, and
propose a novel incentive framework to
motivate users to leverage their delay tolerance
for traffic offloading. Users are provided with
incentives; i.e., receiving discount for their
service charge if they are willing to wait longer
for data downloading. During the delay, part of
the cellular data traffic may be opportunistically
offloaded to other networks mentioned above,
and the user is assured to receive the remaining
part of the data via cellular network when the
delay period ends.
The major challenge of designing such an
incentive framework is to minimize the incentive
cost of cellular network operator which includes
the total discount provided to the mobile users,
subject to an expected amount of traffic being
offloaded. To achieve this goal, two important
factors should be taken into account; i.e., the
delay tolerance and offloading potential of the
users. The users with high delay tolerance and
large offloading potential should be prioritized in
cellular traffic offloading.
First, with the same period of delay, the
users with higher delay tolerance require less
discount to compensate their satisfaction loss. To
effectively capture the dynamic characteristics of
the users’ delay tolerance, we propose an
incentive mechanism based on reverse auction
which is proved to conduct a justified pricing. In
our mechanism, the users act as sellers to send
bids, which include the delay that they are
willing to experience and the discount that they
want to obtain for this delay. Such discount
requested by users is called “coupon” in the rest
of the paper. The network operator then acts as
the buyer to buy the delay tolerance from the
users.
Second, with the same period of delay,
users with larger offloading potential are able to
offload more data traffic. For example, the
offloading potential of a user who requests
popular data is large, because it can easily
retrieve the data pieces from other contacted
peer users during the delay period. Also, if a user
has high probability to pass by some WiFi
hotspots, its offloading potential is large. To
effectively capture the offloading potential of the
users, we propose two accurate prediction
models for DTN and WiFi case respectively.
The optimal auction outcome is
determined by considering both the delay
tolerance and offloading potential of the users to
achieve the minimum incentive cost, given an
offloading target. The auction winners set up
contracts with the network operator for the
delay they wait and the coupon they earn, and
other users directly download data via cellular
network at the original price. More specifically,
the contribution of the paper is three-fold:
We propose a novel incentive framework,
Win-Coupon, based on reverse auction, to
motivate users leveraging their delay tolerance
for cellular traffic offloading, which have three
desirable properties: 1) truthfulness, 2)
individual rationality, 3) low computational
complexity.We provide an accurate model using
stochastic analysis to predict users’ offloading
potential based on their data access and mobility
patterns in the DTN case. We provide an
accurate Semi Markov based prediction model to
predict users’ offloading potential based on their
mobility patterns and the geographical
distribution of WiFi hotspots in the WiFi case.
The rest of the paper is organized as follows. In
Section 2 we briefly review the existing work.
Section 3 provides an overview of our approach
and the related background. Section 4 describes
the details of our incentive framework, and
proves its desirable properties. Section 5
evaluates the performance of Win Coupon
through trace-driven simulations and Section 6
discusses further research issues. Section 7
concludes the paper.

2. RELATED WORK
To deal with the problem of cellular traffic
overload, some studies propose to utilize DTNs
to conduct offloading. Ristanovic et al. propose a
simple algorithm, Mix-Zones, to let the operator
notify users to switch their interfaces for data
fetching from other peers when the
opportunistic DTN connections occur. Whitbeck
et al design a framework, called Push-and-Track,
which includes multiple strategies to determine
how many copies should be injected by cellular
network and to whom, and then leverages DTNs
to offload the traffic. provide three simple
algorithms to exploit DTNs to facilitate data
dissemination among mobile users, in order to
reduce the overall cellular traffic. Many research
efforts have focused on how to improve the
performance of data access in DTNs. In the
authors provide theoretical analysis to the
stationary and transient regimes of data
dissemination. Some later works disseminate
data among mobile users by exploiting their
social relations. Being orthogonal with how to
improve the performance of data access in DTNs,
in this paper, we propose an accurate model to
capture the expected traffic that can be offloaded
to DTNs to facilitate our framework design.
Public WiFi can also be utilized for cellular
traffic offloading. In the authors design HotZones
to enable users turning on WiFi interfaces when
a WiFi connection is expected to occur based on
the user mobility profile and location
information of hot zones covered by WiFi. In the
authors measure the availability and the of-
floading performance of public WiFi based on
vehicular traces. Lee et al. consider a more
general mobile scenario, and present a
quantitative study on delayed and on-the-spot
offloading by using WiFi. The prediction of future
WiFi availability is important to the offloading
scheme design, and has been studied i. In the
authors propose to enable mobile users to
schedule their data transfers when higher WiFi
transmission rate can be achieved based on the
prediction. In a Lyapunov framework based
algorithm, called SALSA, is proposed to optimize
the energy-delay tradeoff of the mobile devices
with both cellular network and WiFi interfaces.
Different from the existing work, in this paper,
we propose an accurate model to predict how
much traffic that can be offloaded via WiFi
hotspots if a mobile user is willing to wait for
certain delay time.
All the existing offloading studies have not
considered the satisfaction loss of the users
when a longer delay is caused by traffic
offloading. To motivate users to leverage their
delay tolerance for cellular traffic offloading, we
propose an auction based incentive framework.
Auction has been widely used in network design.
Applying auc-tion in the spectrum leasing is one
of the most practi-cal applications. Federal
Communications Commission (FCC) has already
auctioned the unused spectrum in the past
decade, and there are a large amount of works
on wireless spectrum auctions. Moreover,
auction has also been applied for designing
incentive mechanism to motivate selfish nodes to
forward data for others. However, none of them
has applied auction techniques to cellular traffic
offloading.

This paper substantially extends the
preliminary version of our results appeared in
we mainly focused on how to stimulate users to
offload cellular traffic via DTNs. In this paper, we
propose a more general framework which
considers both DTNs and WiFi case. We provide
an accurate model to predict users’ offloading
potential in the WiFi case and perform trace-
driven simulations to evaluate its performance.
In addition, we change the data query model in
[18] to more realistic Zipf-like distribution to
evaluate our framework.
Fig. 1. The main idea of Win-Coupon.
3. OVERWIEW
3.1 The Big Picture
In this section, we give an overview of the Win-
Coupo
n
frame
work.
By
consi
derin
g the
users’
delay
tolera
nce and offloading potential, Win-Coupon uses a
reverse auction based incentive mechanism to
motivate users to help cellular traffic offloading.
Figure 1 illustrates the main idea. The network
operator acts as the buyer, who offers coupons
to users in exchange for them to wait for some
time and opportunistically offload the traffic.
When users request data, they are motivated to
send bids along with their request messages to
the network operator. Each bid includes the
information of how long the user is willing to
wait and how much coupon he wants to obtain
as a return for the extra delay. Then, the network
operator infers users’ delay tolerance. In
addition, users’ offloading potential should also
be considered when deciding the auction
outcome. Based on the historical system
parameters collected, such as users’ data access
and mobility patterns, their future value can be
predicted by conducting network modeling, and
then based on the information, users’ offloading
potential can be predicted. e.g., by contacting
other peers which cache the data or moves into
the wireless range of APs. Once delay t passes,
the cellular network pushes the remaining data
pieces to u1 to assure the promised delay. The
losing bidders (e.g. user u3 shown in Figure 1)
immediately download data via cellular network
at the original price.
3.2 User Delay Tolerance
With the increase of downloading delay, the
user’s satisfaction decreases accordingly, the
rate of which reflects the user’s delay tolerance.
To flexibly model users’ delay tolerance, we
introduce a satisfaction function S(t), which is a
monotonically decreasing function of delay t, and
represents the price that the user is willing to
pay for the data service with the delay. The
satisfaction function is determined by the user
himself, his requested data, and various
environmental factors. We assume that each user
has an upper bound of delay tolerance for each
data. Once the delay reaches the bound, the
user’s satisfaction becomes zero, indicating that
the user is not willing to pay for the data service.
Figure 2 shows an example of the satisfaction
function S(t) of a specific user for a specific
data, where tbound is the upper bound of the
user’s delay tolerance. p is the original charge for
the data service, and the satisfaction curve
represents the user’s expected price for the data
as the delay increases. For example, with delay t1
the user is only willing to pay p1 instead of p. p −
p1 is the satisfaction loss caused by delay t1.

Fig. 2. Satisfaction function.
3.3 Auctions
In economics, auction is a typical method to
determine the value of a commodity that has an
undetermined and variable price. It has been
widely applied to many fields. Most auctions are
forward auction which involves a single seller
and multiple buyers, and the buyers send bids to
compete for obtaining the commodities sold by
the seller. In this paper, we use reverse auction
[19] which involves a single buyer and multiple
sellers, and the buyer decides its purchase based
on the bids sent by the sellers. To begin with, we
introduce some notations. Bid (bi): It is
submitted by bidder i to express i’s valuation on
the resource for sale, which is not necessarily
true. Private value (xi): It is the true valuation
made by bidder I for the resources; i.e., the true
price that i wants to obtain for selling the
resource. This value is only known by i. Market-
clearing price (pi): It is the price actually paid by
the buyer to bidder i. This price cannot be less
than the bids submitted by i.
Utility (ui): It is the residual worth of the sold
resource for bidder i, namely the difference
between i’s marketclearing price pi and private
value xi. The bidders in the auction are assumed
to be rational and risk neutral. A common
requirement for auction design is the so-called
individual rationality. Definition 1: An auction is
with individual rationality if all bidders are
guaranteed to obtain non-negative utility. The
rational bidders decide their bidding strategy to
maximize their utility. Let N denote the set of all
bidders. The concept of weakly dominant
strategy is defined as:
Definition 2: bi = βi is a weakly dominant
strategy for user i if and only if: ui(βi, β−i) ≥ ui(β_i,
β−i), ∀β_i_= βi.
Here β−i = {β1, β2, ・・・ , βi−1, βi+1, ・・・ ,
β|N|} denotes the set of strategies of all other
bidders except for bidder i. We can see a weakly
dominant strategy maximizes i’s utility
regardless of the strategies chosen by all other
bidders. If for every bidder, truthfully setting its
bid to its private value is a weakly dominant
strategy, the auction is truthful (strategyproof).
Definition 3: An auction is truthful if each bidder,
say i, has a weakly dominant strategy, in which bi
= xi. The truthfulness eliminates the expensive
overhead for bidders to strategize against other
bidders and prevents the market manipulation.
Also, it assures the efficient allocation by
encouraging bidders to reveal their true private
values. Vickrey-Clarke-Groves (VCG) is the most
well-studied auction format, due to its truthful
property. However, VCG only ensures
truthfulness when the optimal allocation can be
found, and it usually cannot assure the
truthfulness when applied to the approximation
algorithms. Unfortunately, the allocation
problem in Win-Coupon is NP-hard. It is known
that an allocation algorithm leads to be truthful if
and only if it is monotone. In order to maintain
the truthfulness property, we design an
approximation algorithm and make it monotone
in a deterministic sense. Therefore, our incentive
mechanism possesses three important
properties: 1) truthfulness, 2) individual
rationality, and 3) low computational
complexity.
4. MAIN APPROACH OF WIN-CUPON
In this section, we illustrate the details of Win-
Coupon. In the reverse auction based Win-
Coupon, the buyer is the network operator who
pays coupon in exchange for longer delay of the
users. The sellers are the cellular users who sell

their delay tolerance to win coupon. The right
side of Figure 1 shows the flow chart of Win-
Coupon. At first, the network operator collects
the bids to derive the delay tolerance of the
bidders, and predicts their offloading potential.
Then, based on the derived information, a
reverse auction is conducted, which includes two
main steps: allocation and pricing. Finally, the
network operator returns the auction outcome
to the bidders. In the rest of this section, we first
introduce the bidding. Then, we present auction
mechanism and prove its properties. Finally, we
illustrate how to predict bidders’ offloading
potential for both DTN and WiFi cases.
4.1 Bidding
To obtain coupon, the users attach bids with
their data requests to reveal their delay
tolerance. For each user, the upper bound tbound
of its delay tolerance can be viewed as the
resources that it wants to sell. The user can
divide tbound into multiple time units, and
submit multiple bids b = {b1, b2, ……… , bl} to
indicate the value of coupon it wants to obtain
for each additional time unit ofdelay, where l
equals tbound , and e is the length of one time
unit. By receiving these bids, the network
operator knows that the user wants to obtain
coupon with value no less than _ki k=1 bk by
waiting for ki time units. The length of time unit
e can be flexibly determined by the network
operator. Shorter time unit results in larger bids
with more information, which increases the
performance of the auction, but it also induces
more communication overhead and higher
computational complexity. To simplify the
presentation, in the rest of the paper delay t is
normalized by time unit e. As shown in Figure 2,
p − S(t) is the
Fig. 3. Private value.
satisfaction loss of the user due to delay t. Then,
p − S(t) represents the private value of the user
to the delay, namely the user wants to obtain the
coupon with value no less than p−S(t) for delay t.
Thus, the private value of the user to each
additional time unit of delay is x = {x1, x2, ・・
・ , xl}, where xk (k ∀ {1, ・・・ , l}), equals S(k
− 1) − S(k). For example, as shown in Figure 3,
the user wants to obtain the coupon with value
no less than x1 if it waits for one time unit, x1 +
x2 for two time units, and x1 + x2 + x3 for three
time units. Generally, the user can set its bids
with any value at will, however we will prove
that the auction in Win-Coupon is truthful, which
guarantees that the users would bid their private
value; that is, bk = xk, for all k.
4.2 Auction Algorithms
Win-Coupon is run periodically in each auction
round. Usually, the auction would result in an
extra delay for the bidders to wait for the auction
outcome. However, different from other long-
term auctions, such as the FCC-style spectrum
leasing, the auction round in our scenario is very
short, since hundreds of users may request
cellular data service at the same time. Also,
because the bidders who are willing to submit
bids are supposed to have a certain degree of
delay tolerance, the extra delay caused by
auction can be neglected. Next, we describe two
main steps of the auction: allocation and pricing.

4.2.1 Allocation
In traditional reverse auction, the allocation
solution is purely decided by the bids; i.e., the
bidders who bid the lowest price win the game.
However, in our scenario, besides the bids which
express the bidders’ delay tolerance, the
offloading potential of the bidders should also be
considered. Let {t1, t2,…….,t|N|} represent the
allocation solution, where ti denotes the length
of delay that network operator wants to buy
from bidder i.
Note that since each bidder is asked to wait for
integer multiples of time unit, ti is an integer. If ti
equals zero, bidder i loses the game. The
allocation problem in Win- Coupon can be
formulated as follows: Definition 4: The
allocation problem is to determine the optimal
solution {t1, t2, ・・・ , t|N|} which minimizes
the total incentive cost, subject to a given
offloading target.
In Eq.(1), denotes the value of the
coupon that the network operator needs to pay
bidder i in exchange for its delay ti. V di (t) in
Eq.(2) denotes the expected traffic that can be
offloaded, if bidder i downloads data d and is
willing to wait for delay t. We will provide the
details on how to predict V di (t) in Section 4.3
and 4.4 for both DTN and WiFi cases
respectively. We assume that within a short
auction round, each bidder only requests one
data item, so that each i is mapped to a single d.
Thus, this constraint ensures that the total
expected offloaded traffic is no less than the
offloading target v0. Eq.(3) ensures that the
delay that each bidder i waits does not exceed li,
the maximum number of time units that i is
willing to wait. It is easy to prove that our
allocation problem can be reduced to the 0-1
knapsack problem, under the assumption that li
= 1, for all i. The 0-1 knapsack problem is proved
to be NP-hard, and thus our problem is also NP-
hard. Next, we transform the original problem,
and derive the optimal solution of the new
problem by dynamic programming (DP).
4.2.1 Auction algorithm
To develop an intuitive understanding of the
auction algorithm, it is helpful to introduce an
economic equilibrium problem that turns out to
be equivalent to the assignment problem. Let us
consider the possibility of matching then objects
with then persons through a market mechanism,
viewing each person as an economic agent acting
in his own best interest. Suppose that object j has
a price p j and that the person who receives the
object must pay the price pj. Then, the (net)
value of object j for person I is aij− pj and each
person I would logically want to be assigned to
an object ji with maximal value, that is, with
aiji−pji = maxj=1 ,...,n { aij –pj }

We will say that a person I is happy if this
condition holds and we will say that an
assignment and a set of prices are at equilibrium
when all persons are happy. Equilibrium
assignments and prices are naturally of great
interest to economists, but there is also a
fundamental relation with the assignment
problem; it turns out that an equilibrium
assignment offers maximum total benefit (and
thus solves the assignment problem), while the
corresponding set of prices solves an associated
dual optimization problem. This is a
consequence of the celebrated duality theorem
of linear programming. Let us consider now a
natural process for finding an equilibrium
assignment. I will call this process the naive
auction algorithm because it has a serious flaw,
as will be seen shortly. Nonetheless, this flaw
will help motivate a more sophisticated and
correct algorithm.
The naive auction algorithm proceeds in
“rounds” (or “iterations”) starting with any
assignment and any set of prices. There is an
assignment and a set of prices at the beginning of
each round, and if all persons are happy with
these, the process terminates. Otherwise some
person who is not happy is selected. This person,
call him i, finds an object ji which offers maximal
value, that is,
ji∈ arg maxj =1 ,...,n { aij – pj }
and then:
(a) Exchanges objects with the person assigned
to ji at the beginning of the round,
(b) Sets the price of the best object jI to the level
at which he is indifferent between ji and the
second best object, that is, he sets pji to
Pji to pji + γi ,
Where γi = vi− wi
Vi is the best object value, vi= maxj { aij−pj } and
wi is the second best object value
wi= maxj6=ji{ aij−pj }
that is, the best value over objects other than ji
(Note that γi is the largest increment by which
the best object price pji can be increased, with ji
still being the best object for person I )In
traditional reverse auction, the allocation
solution is purely decided by the bids; i.e., the
bidders who bid the lowest price win the game in
our scenario, besides the bids that express the
bidders’ delay tolerance, the offloading potential
of the bidders should also be considered. Let ft1;
t2; …..; tjN jg represent the allocation solution,
where ti denotes the length of delay that
network operator wants to buy from bidder i.
Note that because each bidder is asked to wait
for integer multiples of time unit, ti is an integer.
If ti equals zero, bidder [8] loses the game. The
allocation problem in Win-Coupon can be
formulated
4.2.2. Pricing
The VCG-style pricing is generally used in
forward auction, which involves single seller
with limited resources [8] for sale, and multiple
buyers. The bidders who have the highest bid
win the game, and each winning bidder pays the
“opportunity cost” that its presence introduces
to [5] others it is proved that this pricing
algorithm provides bidders with the incentives
to set their bids truthfully. Based on the basic
idea, in our pricing algorithm [9], the network
operator also pays bidder i the coupon with
value equal to the “opportunity cost” exerted to
all the other bidders due to presence.
5. DISCUSSIONS
In this paper, we mainly focused on the
downloading scenario since the majority of
cellular traffic is on the downlink [31]. We also
separate WiFi and DTN when discussing Win-
Coupon design. Actually, our framework is very
general, and can be extended to fit many other
scenarios. Win-Coupon consists of two parts:
auction based incentive mechanism and
prediction. As long as the volume of offloaded
traffic V di (t) can be predicted, the incentive
mechanism can be adopted for coupon allocation
and pricing under various scenarios such as
uploading, downloading, DTN only, WiFi only, or

hybrid of DTN and WiFi. The only difference
under various scenarios is in the prediction part.
6. CONCLUSION
In this paper, we proposed a novel incentive
framework for cellular traffic offloading. The
basic idea is to motivate the mobile users with
high delay tolerance and large offloading
potential to offload their traffic to other
intermittently connected networks such as DTN
or WiFi hotspots. To capture the dynamic
characteristics of users’ delay tolerance, we
design an incentive mechanism based on reverse
auction. Our mechanism has been proved to
guarantee truthfulness, individual rationality,
and low computational complexity. Moreover,
we design two accurate models to predict the
offloading potential of the users for both DTN
and WiFi cases. Extensive tracedriven
simulations validate the efficiency and practical
use of our incentive framework.
7. REFERENCES
[1] M. Reardon, ”Cisco Predicts Wireless-Data
Explosion,” [Online] Available:
http://news.cnet.com/8301-306863-10449758
266.html.
[2] I. Trestian, S. Ranjan, A. Kuzmanovic, and A.
Nucci, ”Taming the Mobile Data Deluge with
Drop Zones,” IEEE/ACM Trans. On Networking,
2011.
[3] B. Han, P. Hui, V. Kumar, M. Marathe, J. Shao,
and A. Srinivasan, ”Mobile Data Offloading
through Opportunistic Communications and
Social Participation,” IEEE Trans. on Mobile
Computing, 2011.
[4] K. Lee, I. Rhee, J. Lee, S. Chong, and Y. Yi,
”Mobile Data Offloading: How Much Can WiFi
Deliver?” Proc. of ACM CoNEXT, 2010.
[5] A. Balasubramanian, R. Mahajan, and A.
Venkataramani, ”Augmenting Mobile 3G Using
WiFi,” Proc. of ACM MOBISYS, 2010.
[6] N. Ristanovic, J.-Y. L. Boudec, A. Chaintreau,
and V. Erramilli, ”Energy Efficient Offloading of
3G Networks,” Proc. of IEEE MASS, 2011.

IRJET- An Incentive Framework for Cellular Traffic Offloading

More Related Content

What's hot (18)

Similar to IRJET- An Incentive Framework for Cellular Traffic Offloading (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- An Incentive Framework for Cellular Traffic Offloading