Cz35565572

P.P.Nagaraja Rao et al. Int. Journal of Engineering Research and Application www.ijera.com
Vol. 3, Issue 5, Sep-Oct 2013, pp.565-572
www.ijera.com 565 | P a g e
A Novel Approach for Controlling Mobile Data Offloading
Between 3GPP and Non-3GPP Access Networks.
P. P. Nagarajarao1
, G. Sunil2
1
Associate Professor , Department of ECE,SVPCET,Puttur
2
Associate Professor , Department of ECE,SVPCET,Puttur
ABSTRACT
Mobile network traffic is growing rapidly, and service providers must manage their networks efficiently to meet
consumer demand. Mobile data offloading is the use of complementary network technologies for delivering data
originally targeted for cellular networks. Access network discovery and selection function (ANDSF) is the most
complete 3GPP approach to date for controlling offloading between 3GPP (such as HSPA or LTE) and non-
3GPP access networks (such as Wi-Fi or WIMAX). The purpose of the ANDSF is to assist user devices to
discover access networks in their vicinity and to provide rules (policies) to prioritize and manage connections to
all networks. In this paper, we propose a new ANDSF- assisted Wi-Fi regulator method based on end user’s
high-level motion states, such as driving ,walking, and recognizing whether the user is motionless or not, to
avoid unnecessary Wi-Fi scanning and connections. The client-based Wi-Fi connection manager, which
performs cyclic Wi-Fi scanning, is compared with the proposed method. According to the performance results,
the proposed method shows significant performance enhancement in terms of efficient Wi-Fi control and
connectivity with 3G and Wi-Fi.
Keywords-component; ANDSF, Mobile Data offloading, LTE, Handover, Wi-Fi,WIMAX
I. INTRODUCTION
According to the latest Cisco Visual
Networking Index report [1], monthly global mobile
data traffic will exceed 10 exabytes by 2016; Asia, in
particular Japan, South Korea, Indonesia, and China,
will account for 40% of that amount. Furthermore,
the continuous advent of mobile services and the high
speeds of mobile connections are the major factors in
the growth of mobile data traffic.
Today, mobile users frequently use many
smartphone/tablet applications, such as social
networking applications, location based service
applications, and mobile messengers, to communicate
with application servers and peers. It will be unlikely
to change mobile user trends for the use of such
applications, and the traffic volume will be increase
even more. Thus, growing mobile data traffic brings
the requirements of mobile infrastructure
improvement and a new wireless technology, such as
Long Term Evolution (LTE) [2] to fulfill mobile user
demand. Mobile network operators (MNOs) might
have several approaches to deal with the mobile data
traffic explosion. The approaches are additional
spectrum acquisition, existing 3G network upgrades
such as a multichannel assignment, LTE deployment,
and alternative networks, such as Wi-Fi. Since all the
approaches, except the fourth one, are dependent on
MNO policy, thus we limit the discussion to an
alternative network as a solution for mobile data
offloading between 3GPP and non-3GPP access
networks in this paper.
The solutions for mobile data offloading can
be categorized into two solutions, such as a client-
based solution and a server-based solution. Various
Wi-Fi connection managers that are the applications
for connecting the Wi-Fi more easily and quickly are
available on user equipment (UE), and it can be a
client-based solution for mobile data offloading.
However, with the client-based solution, it is
impossible for the MNO to gain visibility and control
over Wi-Fi traffic and the user experience for better
mobile traffic management. In addition, most Wi-Fi
connection managers always require turning on a Wi-
Fi interface on the UE, and thus it can cause
unnecessary Wi-Fi scanning. Therefore, a number of
unnecessary Wi-Fi connections might be established.
Recently, ANDSF has been specified in
3GPP standards 23.402 [3] and 24.312 [4] as a
server-based solution. The ANDSF is an entity within
an Evolved Packet Core (EPC) to assist user
equipment (UE) in discovering non-3GPP access
networks and provide UE with rules and operator
policies to connect to the non-3GPP access networks.
However, the ANDSF server needs to acquire the
current geographical location of the UE to provide
discovery information, which is a list of available
Wi-Fi networks near by the UE’s current location.
Thus, it can also cause an unnecessary energy drain
on the UE owing to its power intensive location
sensing and the synchronizing operations between the
ANDSF server and the UE.
To address the problems of unnecessary Wi-
Fi scanning and improve the energy efficiency of the
ANDSF operation, we propose a novel ANDSF-
RESEARCH ARTICLE OPEN ACCESS

assisted Wi-Fi control method by considering the
user’s motion states. Since the limitation of Wi-Fi
coverage, the most efficient use of Wi-Fi is in a
stationary state. Thus, we propose a new method of
detecting motion states, such as walking and driving,
and of recognizing whether the user is stationary or
not in a short period time. The aim of the proposed
method is to avoid unnecessary Wi-Fi scanning and
unnecessary Wi-Fi connections while users are
moving. The performance of the ANDSF system that
employs the proposed method is compared to a
client-based Wi-Fi offloading solution that performs
periodic Wi-Fi scanning without considering the
motion states.
The remainder of the paper is organized as
follows. Section II discusses the limitations of the
client-based Wi-Fi offloading approach. Section III
describes the proposed efficient ANDSF-assisted Wi-
Fi control method. Section IV evaluates the
performance of the proposed method. Finally,
Section V concludes the paper.
II. LIMITATIONS OF CLIENT-BASED
WI-FI OFFLOADING APPROACH
This section analyzes a client-based solution
for mobile data offloading. In particular, we highlight
the problems of unnecessary Wi-Fi scanning and
unnecessary Wi-Fi connections with the client-based
solution for mobile data offloading.
A. Client-based Wi-Fi Connection Manager
Recently, various Wi-Fi connection
managers have become available in smart phone
application markets, such as Google Play. Most Wi-
Fi connection managers perform Wi-Fi scanning at
regular intervals between 1 second and 60 seconds.
However, this approach can cause unnecessary Wi-Fi
scanning, and thus a number of unnecessary Wi-Fi
connections might be established. To confirm this
inefficient Wi-Fi scanning and to find the reason why
users turn off the Wi-Fi interface, we did some
simple user experience tests. As the first test, we
turned on the Android Wi-Fi connection manager and
listened to Internet radio (NHK Radio World) for 5
minutes while walking around in Tokyo with passing
several Wi-Fi hotspot zones. During the 5 minutes,
the device switched six times between 3G and Wi-Fi,
and 49.6% (Total: 1,986,627 bytes) of the traffic was
downloaded through the Wi-Fi access network as
shown in Fig. 1.
Figure 1. Received traffic with Android Wi-Fi
connection manager
Although, almost 50% of the traffic is
offloaded through the Wi-Fi access network, it leads
to a bad user experience since the Internet radio
service was interrupted and delayed many times due
to frequent switching between 3G and Wi-Fi. Fig. 2
shows the received Internet radio traffic without the
Android Wi-Fi connection manager. In this test, we
turned off the Android Wi-Fi connection manager
and confirmed that there was no interruption to the
Internet radio reception. Therefore, it is clear
evidence as to why users turn off the Wi-Fi interface.
Figure 2. Received traffic without Android Wi-Fi
connection manager
B. ANDSF-assisted Wi-Fi Control
To allow UE to know where and how to
choose a non-3GPP access network, such as Wi-Fi,
the Technical Specification 23.402[3] defines the
ANDSF. In addition, to assist UE in discovering non-
3GPP access networks, the ANDSF server provides
ANDSF information, which is represented by the
ANDSF Management Object described in Technical
Specification 24.312[4]. The ANDSF information
includes discovery information, which is a list of Wi-
Fi access networks available to the UE based on the
UE current location and selection rules, which is a
list of prioritized rules that control which Wi-Fi
access networks should be used based on the UE
current location. The exchange of ANDSF
information can be triggered by the client pull mode
or the server push mode. The initial client pull mode
occurs after the UE establishes a 3G connection. At
this point, the UE should provide location
information (geographical location or base station
ID) and request ANDSF information from the
ANDSF server. Then, the ANDSF server processes
this request by sending a prioritized list of Wi-Fi
access networks for the UE current location. A client
pull mode also occurs when the UE moves to another
location. Thus, according to the ANDSF system,
when the UE moves to another location, the UE
needs to retrieve ANDSF information by using the
pull method from the ANDSF server every time.
However, in the case of the user motion state of
walking and driving, remaining within the 3G
coverage area is much better than Wi-Fi access due to
the limitations of Wi-Fi coverage. As a result, with
the current ANDSF system, intensive ANDSF

information retrievals will be accrued. To confirm
this inefficient ANDSF information retrieval, we
tested the ANDSF information retrieval operation by
using the ANDSF server and client developed
according to the TS 24.402 and the TS 24.312. Fig. 3
shows the cumulative ANDSF information retrieval
traffic and the time for policy retrieval attempts for
44 minutes.
We measured the ANDSF information
retrieval traffic while riding on a train and walking.
In the test result, the ANDSF information was
retrieved 33 times between the ANDSF server and
the client, and the cumulative ANDSF information
retrieval traffic increased as shown in Fig. 3. As we
pointed out before, a ANDSF information retrieval
operation while the user is moving is not necessary
since it can cause unnecessary energy drain on the
UE owing to its power intensive location sensing and
the ANDSF information retrieval operations between
the ANDSF server and the client.
Figure 3. Cumulative ANDSF Information Retrieval
Traffic
III. ANDSF-ASSISTED WI-FI
CONTROL FOR MOBILEDATA
OFFLOADING
In this section, the proposed ANDSF-
assisted Wi-Fi control method is described.
A. Discussion on ANDSF Limitations and
Alternatives
Figure 4. The ideal case of Wi-Fi Control
The ideal case of Wi-Fi control is to turn on
the Wi-Fi interface automatically when the user is
stationary, which provides good quality Wi-Fi as
shown in Fig. 4. In Fig. 4, the three different types of
Wi-Fi are installed in each place and only the orange
color coded Wi-Fi is acceptable to the user. This Wi-
Fi information can be delivered by the ANDSF
server, which is one of the benefits of the server-
based mobile data offloading solution compared to
the client-based solution. To achieve this, the
following requirements should be fulfilled. At first,
the UE needs the latest ANDSF information,
including good quality Wi-Fi information, when the
UE is ready to access the Wi-Fi network. Thus, the
most appropriate time for the policy retrieval is not
when the UE moves to another location but when the
UE is stationary. Second, the UE needs a new
function to detect the user’s high-level motion state
of walking or driving and recognizing whether the
user stationary or not in a short period. Current
positioning technologies can be used for detecting the
user’s high-level motion states but GPS [5] is
extremely power hungry due to its power intensive
location sensing, and the Wi-Fi-based positioning
method [6] always needs to keep turning on the Wi-
Fi interface, which can also cause an unnecessary
energy drain on the UE, as well as unnecessary Wi-Fi
scanning. To address these problems, we propose a
new user high-level motion detection and tracking
method based on 3G/4G(CDMA, WCDMA, LTE)
RSSI (received signal strength indicator) variation.
With the user’s high-level motion detection, periodic
3G/4G RSSI measurement by the UE is a natural and
essential behavior; thus, the energy drain by periodic
RSSI measuring is unavoidable among the total
energy consumption. We measured the 3G RSSI and
BSID variations while moving and remaining in one
location, respectively, using three different types of
smartphones (HTC ISW13HT, Sony Xperiaacro HD,
and Kyosera Urbano). Fig. 5 and Fig. 6 show the 3G
RSSI/BSID variations in the moving state and the
stationary state, respectively.

Figure 5. 3G RSSI/BSID Variation of Moving State
Figure 6. 3G RSSI/BSID Variation of Stationary
State
In Fig. 5, the 3G RSSI/BSID changes irregularly
during the measurement time period. In contrast,
there are relatively low variations in 3G RSSI values
as shown in Fig. 6. With this analysis, we confirm
that the RSSI/BSID variations can be used to decide
whether a user is stationary or not.
B. Proposed Automatic Wi-Fi Control
Mechanism
To improve Wi-Fi control in terms of
avoiding unnecessary ANDSF information retrieval
and unnecessary Wi-Fi scanning, we propose a
method for automatic Wi-Fi control that consists of
the following steps: 3G RSSI/BSID measuring during
a given time interval; decision on the user’s motion
state using 3G RSSI/BSID variations; prevention of
duplicated Wi-Fi scans in the same place; ANDSF
policy retrieval; and selective Wi-Fi connections
based on Wi-Fi RSSI. The first step is the decision on
the user’s motion state using 3G RSSI/BSID
variations. To turn on the Wi-Fi interface
automatically when a user is stationary only, the UE
needs to observe the variations in 3G RSSI/BSID
during given time intervals as shown in Fig. 7.
Figure 7. An example of decision on user’s motion
state
For each 3G RSSI/BSID checkpoint time,
the UE compares current 3G RSSI/BSID values with
the previous 3G RSSI/BSID values. If both 3G
RSSI/BSID variations are within the acceptable 3G
RSSI variation threshold, the UE detects the current
motion state as stationary (Line 2 in Fig. 8). The
second step is prevention of duplicated Wi-Fi scans
in the same location (Line 4 in Fig. 8). Fig. 7 shows
an example of the Wi-Fi scanning rule. Although the
UE is stationary after check time interval 3, it does
not perform a Wi-Fi scanning. The reason is that in
comparison to the previous result (Check Time
Interval 2), the 3G BSID is unchanged, and the 3G
RSSI variations of the previous 3G RSSI value and
the current 3G RSSI value are within the range of the
acceptable 3G RSSI variation (ex. ±2 dBM) so that
the UE detects that the user is stationary. Here, since
there were no available Wi-Fi APs in the previous
Wi-Fi scan results (Check Time Interval 2), the UE
should not perform a Wi-Fi scan after Check Time
Interval 3. The third step is the ANDSF policy
retrieval (Line 15 in Fig. 8). If the UE detects that it
moves to another location and is then stationary, the
UE requests the ANDSF policy for the current
location, which is at least a different ANDSF policy
retrieval method compared to the current ANDSF
system. The final step is the Wi-Fi scanning and
connection considering the Wi-Fi RSSI. If the Wi-Fi
RSSI is under the predefined Wi-Fi RSSI threshold,
the UE terminates the Wi-Fi connection according to
the proposed algorithm.
RC: Current Measured 3G RSSI
RP: Previous Measured 3G RSSI
RT: Acceptable 3G RSSI Variation Threshold
between RCand RP BC: Current Measured 3G BSID
BP: Previous Measured 3G BSID
1: Measuring 3G RSSI/BSID variation during given
time interval;
2: IFRC‐RPwithinRTandBC=BP// Decision on User s

motion state
3: User motion state: stationary detected;
4:
IF Rp‐RP-1withinRTandBp=BP-1// Prevention of
duplicated WiFi scan
5: Check Previous WiFi scan result;
6: IF Available WiFi exists
7: Turn ON WiFi Interface;
8:Perform WiFi Connection considering WiFi RSSI;
9: ELSE
10:RP= RP-1
11:RC= NULL
12:Bp= BP-1
13:END
14:ELSE
15:
Request ANDSF Information; // Start
ANDSF policy retrevial
16:IF Available WiFi exists in vicinity of UE
17:Turn ON WiFi Interface;
18:
Perform WiFi Connection considering WiFi
RSSI;
19:ELSE
20:RP= RP-1
21 RC= NULL
22:Bp= BP-1
23:END
24: ELSE
25:RP= RP-1
26:RC= NULL
27:Bp= BP-1
28: END
Figure 8. Algorithm for automatic Wi-Fi control
method
IV. PERFORMANCE EVALUATION
Figure 9. ANDSF Information and User’s motion
detection function
To evaluate the performance of the proposed
automatic Wi-Fi control method, we developed a
prototype of the ANDSF server and client running
the Android OS. The performance of the proposed
method was compared to the Android Wi-Fi
connection manager, which is a client-based Wi-Fi
offloading solution that performs periodic Wi-Fi
scanning.
Fig. 9 shows the ANDSF policy delivered by the
ANDSF server and the user’s motion detection
function running on the Android OS. An ANDSF
policy consists of the policy ID, including relevant
parameters, such as a prioritized access network list
and the validity period for the policy and scenario ID,
including a list of available Wi-Fi information in the
vicinity of the UE.
To validate the proposed automatic Wi-Fi control
method and find the most appropriate values (3G
RSSI check interval, and acceptable 3G RSSI range)
for user motion detection, the proposed method was
evaluated in terms of the number of Wi-Fi scans and
the time cost for the Wi-Fi connection in a real
environment with a fixed period of 10, 20, 30, 40, 50,
and 60 seconds respectively.

Figure 10. Route for Wi-Fi scan and connection test
(Ikebukuro in Tokyo)
For the number of Wi-Fi scan tests, we moved along
the selected route as shown in Fig. 10 in order to
confirm how many times Wi-Fi scans were
performed while we were moving. In addition, the
time cost for the Wi-Fi connection was also tested in
a location that provided a good quality of Wi-Fi in
the Ikebukuro area.
Figure 11. Number of Wi-Fi scan test results (3G
RSSI check interval from 10 to 30 seconds)
Table 1. Time cost for Wi-Fi connections (3G RSSI
check interval from 10 to 30 seconds)
Figure 12. Number of Wi-Fi scan test results (3G
RSSI check interval from 40 to 60 seconds)
Table 2. Time cost for Wi-Fi connections (3G RSSI
check interval from 10 to 30 seconds)
Interval
3G RSSI_40 3G RSSI_50
3G
RSSI_60
Time
RSSI
Range 1 2 3 1 2 3 1 2 3
1 89 49 89 410 59 159 189 70 310
2nd 89 50 130 260 59 109 130 144 129
3 d
49 90 90 320 170 59 130 129 70
4 h 89 49 129 - - - - - -
5th 209 89 49 - - - - - -
6 h
50 50 49 - - - - - -
Interval
3G RSSI_10 3G RSSI_20 3G RSSI_30
Time
RSSI
1 2 3 1 2 3 1 2 3
Range
1st 28 19 20 59 49 49 349 69 40
2nd 43 30 19 89 39 49 99 40 40
3 d 59 20 29 69 50 44 159 43 39
4 h 49 29 42 49 49 29 279 40 39
5th 29 50 49 30 29 30 39 73 39
6 h
34 39 18 49 69 63 279 69 99
7 h
29 19 40 29 49 29 69 39 69
8th 31 25 53 89 30 29 69 39 40
9 h 49 39 20 70 29 79 39 39 39
10 h
19 30 20 109 49 30 40 69 39
Average 37 30 31 64.2 44.2 43.1 142 52 48.3

7 h 210 89 49 - - - - - -
8th 59 60 49 - - - - - -
9th 169 49 49 - - - - - -
10 h
49 89 49 - - - - - -
Fig. 11 and Fig. 12 show the results of the number of
Wi-Fi scan tests. With this result, we confirmed that
the number of Wi-Fi scans decreased and the time
cost for the Wi-Fi connection increased by increasing
the 3G RSSI check interval time as shown in Table 1
and Table 2. Considering the above results, we
decided that the 3G RSSI check interval time was 30
seconds and the acceptable 3G RSSI
variation was ±2 dBM for the configuration values of
the proposed automatic Wi-Fi control method since
the combination of these two values had good
performance compared to the others.
Comulative ANDSF Information Retrival Traffic (Existing ANDSF)
Comulative ANDSF Information Retrival Traffic (Proposed ANDSF)
Figure 13. Cumulative Policy Retrieval Traffic
To verify the efficient ANDSF information retrieval
operation of the proposed ANDSF-assisted system,
the cumulative ANDSF information traffic was
measured while riding on a train and walking for 44
minutes as shown in Fig. 13. Since the ANDSF
information retrieval, only starts when the UE detects
the motion state as stationary, a number of
unnecessary ANDSF policy retrievals are not ever
going to happen while users are moving. Thus, the
intensive ANDSF information requests will be
decreased dramatically compared to the existing
ANDSF system.
Table 3. Average results of Wi-Fi scans and
connection in Fig. 14
Number Of Wi-Fi Scan Number of Wi-Fi Connection
Android CM Proposed CM Android CM Proposed CM
8.1 2.5 8.6 0.3
To verify the efficient Wi-Fi scanning and connection
of the proposed automatic Wi-Fi control method, the
proposed method was compared to the Android Wi-
Fi connection manager, which performed periodic
network scanning. The 10 tests were performed in the
Ikebukuro area as shown in Fig. 10. According to the
performance evaluation results, the proposed method,
the number of Wi-Fi scans and connections was
fewer than the number from the Android Wi-Fi
connection manager. Table 3 shows the average
results of Wi-Fi scans and connections.
Figure 14. Comparison of a number of Wi-Fi scans
and connections
Table 4 shows the average results of Wi-Fi scans and
connections on the train. We also evaluated the
proposed method with the Android Wi-Fi connection
manager while riding on the train for around one
hour.
Table 4. Average results of Wi-Fi scanning and
connection on the train
Number Of Wi-Fi Scan Number of Wi-Fi Connection
Android CM Proposed CM Android CM Proposed CM
180 17 17 3
This is a result of the fact that the proposed automatic
Wi-Fi control method can reduce the number of
unnecessary Wi-Fi scans and connections while the
user is moving so that it can prevent an unnecessary
energy drain on the UE owing to its frequent Wi-Fi
scanning.
I. CONCLUSION
In this paper, we proposed a novel automatic
Wi-Fi control method in order to decide when a Wi-
Fi scan and the ANDSF policy retrieval should be
performed by the UE. Performance results confirmed
that the number of unnecessary Wi-Fi scans and
ANDSF information retrievals were dramatically
reduced. Therefore, the proposed automatic Wi-Fi
control method can lead to user experience
enhancement, and thus mobile data offloading
through Wi-Fi can be realized.
Average 109 66 73 330 96 109 149 114 169
50000
45000
40000
35000
Byte
30000
25000
20000
15000
10000
5000
0
1 201 401 601 801 1001 1201 1401 1601 1801 2001 2201 2401 2601

REFERENCES
[1] Cisco, Visual Networking Index: Global
Mobile Data Traffic Forecast Update, 2011–
2016, February 2012.
[2] Muntean, V.H., Marius Otesteanu,
"WiMAX versus LTE - An overview of
technical aspects for Next Generation
Networks technologies," ISETC 2010,
November 2010.
[3] 3GPP TS 23.402, “Architecture
Enhancement for Non-3GPP Accesses
(Release 11),” December 2011.
[4] 3GPP TS 24.312, "Access Network
Discovery and Selection Function (ANDSF)
Management Object (MO) (Release 11),"
December 2011.
[5] JeongyeupPaek, Joongheon Kim, Ramesh
Govindan, "Energy-efficient rate-adaptive
GPS-based positioning for smartphones,"
MobiSys 2010, June 2010.
[6] Place lab: A privacy-observant location
system. http://www.placelab.org/.

Cz35565572

More Related Content

What's hot (19)

Viewers also liked (20)

Similar to Cz35565572 (20)

Recently uploaded (20)

Cz35565572