AN ADVANCED QOS ANALYSIS AND EVALUATION METHOD FOR MOBILE INTERNET ACCESS

International Journal of Wireless & Mobile Networks (IJWMN) Vol. 8, No. 3, June 2016
DOI: 10.5121/ijwmn.2016.8304 55
AN ADVANCED QOS ANALYSIS AND EVALUATION
METHOD FOR MOBILE INTERNET ACCESS
F. Beritelli, C. Rametta, A. Raspanti*, M. Russo°, F. Scaglione° and G. Spallina*
Dipartimento di Ingegneria Elettrica, Elettronica e Informatica,University of Catania,
V.le A. Doria, 6 – 95125 Catania, Italy
* HELILAB S.R.L. Via IV Novembre, 130 F – 95022 Aci Catena (CT) - Italy
° VICOSYSTEMS S.R.L. V. le O. Da Pordenone, 33 -95128 Catania - Italy
ABSTRACT
The paper proposes a new method for the analysis and evaluation of the Quality of Service (QoS) in a
mobile Internet access scenario. In particular, the paper proposes a throughput evaluation method based
on PathChirp algorithm. The end-to-end bandwidth was estimated by means of the Self Loading of Periodic
Streams (SloPS) technique. The obtained measurements were then analyzed by estimating the degree of
correlation with other parameters that characterize the data transmission such as power, round trip time,
etc. Finally, in order to have greater spatial resolution performance guaranteed by an Internet service
provider, a 3D reconstruction method based on using drones is proposed and some preliminary results are
discussed.
KEYWORDS
Radio Access Evaluation; 3G-4G Networks, Throughput Evaluation Algorithms; RAN KPIs Correlation.
1. INTRODUCTION
Mobile Internet is growing rapidly due to low setup costs and decrease in price. We live in a
wireless world, where the demand for the Internet on our laptops, tablets and smart phones is a
necessity anywhere we go. However, not everyone has access to the data speed they require. The
main reason for slow Internet speed is weak signal strength due to inadequate radio coverage in
critical areas such as those in the suburbs.
The paper proposes a new method for the analysis and evaluation of the Quality of Service (QoS)
in a mobile Internet access scenario. In particular, the paper proposes a throughput evaluation
method based on PathChirp algorithm [1]. The end-to-end bandwidth was estimated by means of
the Self Loading of Periodic Streams (SLoPS) technique [2][3].
Part of the work done in this paper was focused on the identification of an alternative technique to
simple FTP, but still able to provide an accurate estimate of the throughput. This parameter is
often measured through an FTP session, by sending a file and analyzing the time it takes to
transfer it, calculating thus the file transfer speed, expressed in bit/s. The selection of too small
size files could speed the measurement time, but it could lead to an inaccurate estimate of
throughput, conversely the selection of large files would provide a more accurate estimate of
throughput, but it would require more time to complete the transfer and, obtaining the result, still
rank higher in low bandwidth availability.

56
To this end, we have analyzed the PathLoad [2] and PathChirp techniques, both techniques based
on a client-server paradigm, and able to provide an estimate of the available bandwidth end-to-
end.
By comparing these techniques with the FTP technique, it was found that the data rate of the
measured values with PathChirp are very close to the measured values with the FTP technique,
but with the advantage of significantly faster convergence time.
Once the technique to derive the throughput estimates was identified, measurements on certain
KPIs (Key Performance Indicators) were taken and presented in Section III. Following these
measurements, the correlation values were also calculated between the received signal strength
and other parameters measured, in order to assess how the latter were influenced by the signal
strength level received.
In order to estimate the QoS in a Mobile Internet context, nowadays many measurements are
carried out in motion on board a motor vehicle by identifying various test points along the route
in which the measurements are taken. Obviously there are limitations: the measurements are
limited only to areas accessible by a motor vehicle and they provide estimates of the QoS offered
to a preset altitude from ground level, thus not providing any suggestions as to the possible
variations in height.
The final idea is to apply this measuring mechanism aboard a drone, with the considerable
advantage of being able to take measurements in areas not easily accessible and above all provide
an overview of QoS in a 3D scene, with drone being able to take measurements even in height.
2. STATE OF ART
On the one hand, recently we have been assisting to a relevant increase in the drone use in many
different scenarios: inspection of large infrastructure for remote damage assessment and
monitoring purposes [4], support for emergency operations, such as finding missing persons [5],
support in rapid and effective organization of disaster recovery operations [6][7][8], applications
for smart city contexts [9][10], and many other scenarios.
On the other hand, mobile networks are gradually becoming the predominant source of Internet
access, so developers, researchers, network operators, and regulatory communities have taken an
increased interest in measuring end-to-end mobile network performance [11][12].
The first approach to the use of drones in mobile network performance evaluation was already
taken by employing drones carrying smartphones with network testing applications to analyze the
KPIs and estimate the offered QoS [13].
In this paper, however, we propose a new method of network mobile performance evaluation,
which is not based on a simple smartphone, but rather on a dedicated and optimized prototype
installed on a drone.
3. DRONE-INTERNET TECHNOLOGY
Drones for civilian applications (less than 10 kg in weight) have been one of the topics of great
interest in recent years (Fig. 1). The application contexts are numerous and applications range from
video footage, 3D reconstructions of buildings and architectural structures, to playful contexts,

57
etc.As is well known, while our increasingly hyperconnected world brings many benefits, it also
exacerbates inequalities for the share of the global population that has limited or no Internet access.
In some scenarios it is also essential to ensure Internet access as a backup to traditional solutions,
consider, for special events such as earthquakes, the ground war actions, etc. where it is vital to
ensure immediate coverage of Internet access.
Another equally important application area covers the drone as a multi-sensor mobile system that
allows measurement of a parameter set. The benefit is a completely innovative spatial sampling
method that becomes fundamental in many applications in which a greater spatial resolution of the
measured data may be necessary.
Figure 1. Drone for civilian applications
Recently, much has been said of the drone-Internet combination as an innovative technology to
expand Internet access in areas not reached by the traditional radio coverage for the delivery of
services via Internet. Think of Google's top-secret Project SkyBender testing the viability of
millimeter wave technology beamed from solar-powered Titan drones, in order to deliver 5G
internet and other internet-over-the-air projects.
Mountain View has reportedly been experimenting with millimeter wave-based Internet.
Millimeter waves are believed to be capable of transmitting data 40 times faster than LTE and
could become the technology behind 5G internet.
Lately, though, two initiatives — Google Project Loon and Mark Zuckerberg’s internet.org —
have drawn widespread attention to the challenge of using alternative delivery platforms to bring
low cost broadband access to unserved or underserved parts of the world.
Project Loon plans to bring Internet access to remote locations via the network of high-altitude
balloons. As Google describes it, “People can connect to the balloon network using a special
Internet antenna attached to their building.”
Internet.org offers a similar approach, except instead of balloons, it envisions using drones as the
delivery platform.
The benefits of all-wireless network in the sky are clear. Such a network would be far less
expensive, far less disruptive and it would take far less time to build than implementing a wired /
land based infrastructure over very large swaths of the earth where no communications
infrastructure currently exists.

58
4. DESCRIPTION OF THE PROTOTYPE
To perform our tests we developed a prototype by using a small form factor system boards
optimized for wireless routing and network applications, an ALIX2D2 board [14], equipped with
a 500 MHz AMD Geode LX800 CPU and 256 MB DDR DRAM on board; a USB Internet key
has been used to provide cellular connectivity [15][16]; ZeroShell[17] has been used as the
operative system whereas proprietary scripts have been developed by using bash and python
language [18] to evaluate the available end-to-end bandwidth between the prototype and a remote
server acting as the end point of the communication, RSSI, Round Trip Time (RTT) and standard
deviation of RTT. A picture of the hardware employed to realize our prototype is shown in Figure
2.
Two algorithms have been implemented on board the prototype: PathLoad and PathChirp. Both
algorithms evaluate the end-to-end bandwidth by means of the Self Loading of Periodic Streams
(SloPS) technique and they are based on a client-server paradigm: the sender process runs on the
mobile equipment whereas the receiver process runs on a remote host.
Figure 2. The prototype employed for our test bed.
The working principle of the Self Loading of Periodic Streams is very simple: it is based on the
periodic transmission of packet streams from the source to the destination and on the consequent
measurement of the One Way Delay (OWD), which is the time interval between the transmission
of the data packet and its reception at the receiver side.
The number of packets transmitted increases until the growing one way delay; when this happens
it means that the packet transmission rate is greater than the available bandwidth provided by the
radio access network.
Figure 3 shows the above mentioned procedure: when the transmission rate I is compatible with
the available bandwidth (A), i.e. R < A, one way delay shows an almost constant trend; however,
when the packet rate from the source to the destination is greater than the available bandwidth,
i.e. R > A, one way delay increases because of the TCP congestion window mechanism.

59
Figure 3. One way delay values of the ith
packet train in the SloPS technique [3].
In PathLoad the measurement is obtained by iterative sending of a series of K packets of L bit
each of which is transmitted during an interval of T seconds. In such a way the transmission rate
is equal to R = L/T [bit/s]. Each packet of the series has a timestamp to indicate when the data
packet has been created and sent to the receiver node. Once the destination node receives the
packets stream it compares the arrival time (Ai) and the sending time (Si) of the packets in order
to calculate the one way delay of the ith
stream, i.e. Di = Ai – Si.
The sender and the receiver communicate according to the client-server paradigm in order to
establish the available bandwidth. When one way delay at the destination side increases, in fact,
the receiver process notifies the sender that behaves as follows:
a. If R(i) < A, the source process will send the following packets stream at R(i+1) > R(i)
b. If R(i) > A, the source process will send the following packets stream at R(i+1) < R(i)
Furthermore, the rate of the stream (i+1) is established as follows [19]:
1. Two start parameters, RminedRmax, are initialized equal to zero and equal to the ideal
maximum throughput provided by the connection Rmax;
o If R(i) < A then Rmin = R(i)
o If R(i) > A then Rmax = R(i)
2. R(i+1) = (Rmax – Rmin)/2
The iterative process ends when Rmax – Rmin< w, where w indicates the value depending on the
precision of the bandwidth evaluation procedure (obviously, the greater the w the less accurate
the evaluation but, at the same time, the faster the convergence period required by the algorithm).
PathChirp is based on the same principle but, in this case, the algorithm does not need to send
multiple trains of equally spaced data packets increasing at each step the transmission rate.
PathChirp is based on the sending of trains of packets, called chirps, characterized by the fact that
they are not equally spaced but, vice versa, the interval time between a packet and the next one
gradually decreases according to the exponential distribution.
5. PERFORMANCE EVALUATION
A.Description of the Testbed
Tests were carried out in mobility, along predetermined paths and identifying for each path 10
test points where the measurements were taken.

60
Tests have been classified into three scenarios which are considered to be of particular interest,
especially for the study of how Throughput, Round Trip Time (RTT) and Standard deviation of
RTT are related to the RSSI value: the first results refer to the condition of a high quality of the
received signal in a high density of mobile users area; the second scenario, instead, refers to an
urban context where the areas of particular concentration of users have not been crossed; finally,
the last scenario proposes the measurements carried out in a extra-urban environment, where the
received signal levels were very poor.
B.Metrics
The metrics taken into consideration to compare the duplication system and the commonly used
single streaming transmissions can be summarized as follows.
To give a first evaluation of the QoS offered by the mobile radio access network, in our analysis,
we focused on the measurement of the following parameters:
RSSI (Received Signal Strength Indicator)
Throughput
Throughputevaluationlatency
Round Trip Time
Mdev (standard deviation of RTT)
C.RSSI
In telecommunications, received signal strength indicator (RSSI) is a measurement of the power
present in the received radiosignal. RSSI is usually invisible to a user of the receiving device.
However, because signal strength can vary greatly, it has an impact on the functionality in
wireless networking.
A mobile phone signal is the signal strength received by a mobile phone from a cellular network
(on the downlink). This parameter can depend on several factors, such as the distance from the
Base Station, any obstacle such buildings and trees, etc. So the signal strength can vary greatly.
Typically RSSI is measured in dBm. Another way to express RSSI is the Arbitrary Strength Unit
(ASU).Arbitrary Strength Unit (ASU) is an integer value proportional to the received signal
strength measured by the mobile phone.It is possible to map the real signal strength measured in
dBm in ASU by means of mathematical relationship depending on the radio access
technologyError! Reference source not found..
D.Throughput:
This parameter represents the average end-to-end bandwidth between source and destination of
data packet communication; it is possible to evaluate the average and/or the instantaneous value
of the parameter that will have more or less impact according to the type of user application.
Throughput is the key performance indicator in case of data transfer and one of the most
important indicators in case of multimedia content delivery applications such as real-time audio
and video delivery services, i.e. video surveillance, VoIP, private communications system and so
on.

61
E.Throughput evaluation latency
In scenarios characterized by high user’s mobility or in other cases where the radio access
network performance evaluation is carried out by using high mobility vehicles such as drones, the
time needed to evaluate the throughput plays a key role in order to implement an effective and
efficient measurement campaign; with this scope two well known throughput evaluation methods
have been proposed under this perspective, comparing the time needed to converge with the
estimated value and the latter with the value obtained by using the traditional file transfer protocol
procedure.
F.RTT
The measurements of the Round Trip Time were carried out using the ICMP protocol. To get a
fairly accurate measurement of RTT, each value was obtained by sending 30 packets with an
interval of 0.2 sec from each other, for each of them RTT was calculated and finally the average
value was estimated.
G.MDEV
It is the standard deviation, essentially the average measurement of how far each ping RTT is
situated from the mean RTT. The higher mdev is, the more variable the RTT is (over time).This
parameter, calculated using the same set of ICMP packets sent to the calculation of the average
RTT, is an indicator of the homogeneity of the measured delays: the lower the value, the higher
the homogeneity level of the delays.
A.Results
In this section the results obtained by the prototype are illustrated and the conclusions about the
behavior of the measurement system are drawn.
The first test campaign is concerned with the assessment of the bandwidth estimation algorithms;
as previously mentioned PathLoad and PathChirp were installed on board the prototype and their
performances were compared. More specifically, 20 tests were carried out at different points in
the urban area of Catania; the first 10 points characterized by a high signal to noise ratio and the
second 10 points characterized by low signal-to-noise ratio. For each point, both tools were run
for 25 seconds and the detected bandwidth measurements were compared with the results
obtained using one proprietary script based on the detection of the effective end-to-end bandwidth
by means of the common file transfer protocol. The two algorithms have been compared within
the context of traditional wired IP networks, showing that PathChirp offers better performance if
compared to PathLoad; the goal of this test was to verify if such behavior was the same also in the
context of mobile access network. The results found confirm that PathChirp bandwidth evaluation
is more consistent with the value experienced by the end user than the one performed by means of
PathLoad. In particular, Table 1 shows the bandwidth values measured by means of the two
algorithms and by means of a common file transfer protocol measurement mechanism.PathChirp
outperforms PathLoad in all tests: for the twenty performed measurements, each of duration equal
to 25 seconds, the average percentage error equaled 21% for PathChirp and 56% while running
PathLoad.

62
Table 1. PathLoad and PathChirp comparison.
Test PathChirp
[Mb/s]
PathLoad
[Mb/s]
FTP
[Mb/s]
P.Chirp
vs FTP
[%]
P.Load
vs FTP
[%]
1 1.13 0.45 2.1 46,2 78,6
2 1.17 0.48 1.4 16,4 65,7
3 0.93 0.92 1.02 8,8 9,8
4 0.92 0.5 0.73 26,0 31,5
5 0.87 0.5 1.44 39,6 65,3
6 0.82 0.57 1.09 24,8 47,7
7 0.67 0.47 0.92 27,2 48,9
8 0.96 0.29 1.01 5,0 71,3
9 0.99 0.4 1.21 18,2 66,9
10 0.78 0.67 1.46 46,6 54,1
11 1.02 0.39 1.04 1,9 62,5
12 0.8 0.36 1.01 20,8 64,4
13 1.03 0.39 1.16 11,2 66,4
14 1.04 0.42 0.85 22,4 50,6
15 0.99 0.42 0.85 16,5 50,6
16 1.02 0.5 1.05 2,9 52,4
17 1.04 0.4 1.4 25,7 71,4
18 1.02 0.5 1.05 2,9 52,4
19 1.1 0.42 0.85 29,4 50,6
20 0.8 0.36 1.01 20,8 64,4
Once found that the “Pathchirp” technique best approximates the throughput measured with an
FTP session and converges quickly with this value; thus, the following results have been
proposed in connection to the three previously mentioned scenarios.
Table 2 shows the results obtained in a scenario with a high level of received signal and high
concentration of mobile users. Table 3 shows the results obtained in an urban scenario.
Finally, Table 4 shows the results obtained in an extra-urban scenario where the received signal
level was very weak.
Figure 4 (a)Bandwidth evaluation with PathLoad and PathChirp compared with the effective value
measured by means of a FTP session for the first 10 test points
0.0
0.5
1.0
1.5
2.0
2.5
1 2 3 4 5 6 7 8 9 10
Evaluatedbandwidth[Mb/s]
Number of test
Pathchirp Pathload FTP

63
Figure 4 (b) Bandwidth evaluation with PathLoad and PathChirp compared with the effective value
measured by means of a FTP session for the last 10 test points
Figure 5 Measurement time for FTP technique
0
0.5
1
1.5
2
1 2 3 4 5 6 7 8 9 10
Evaluatedbandwidth[Mb/s]
Number of test
Pathchirp Pathload FTP
0
30
60
90
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Timeofmeasure[s]
Number of test
FTP

64
Table 2. High users density urban area.
Table 3. Urban area.
TEST
POINT
RSSI
(ASU)
Average RTT
(ms)
MDEV
(ms)
Bandwidth
(Mbps)
1 lat: 37,526513 22 135,526 7,6 2,172
long: 15,072656
2 lat: 37,5267981 16 133,983 13,815 2,172
long: 15,0750943
3 lat: 37,5265764 27 170,375 28,162 1,5456
long: 15,0785617
4 lat: 37,5251086 18 142,315 14,863 1,8864
long: 15,0747177
5 lat: 37,5268042 26 132,088 0,955 2,1264
long: 15,0696735
6 lat: 37,5284868 26 133,116 10,2 2,1368
long: 15,0689631
7 lat: 37,5333923 23 132,658 11,299 1,9496
long: 15,0677091
8 lat: 37,5366734 20 143,137 20,115 2,5368
long: 15,0674017
9 lat: 37,5382284 17 185,965 14,726 2,0584
long: 15,0706793
10 lat: 37,5293026 28 134,266 15,891 2,1368
long: 15,076781
GPS Coordinates
TEST
POINT
RSSI
(ASU)
Average RTT
(ms)
MDEV
(ms)
Bandwidth
(Mbps)
1 lat.: 37,527555 19 175.827 18.077 2,096
long.: 15,098745
2 lat.: 37,527585 16 142 17.397 2,1024
long.: 15,0859046
3 lat.: 37,5238828 31 140.202 12.047 2,168
long.: 15,1158588
4 lat.: 34,546725 11 132.032 5.859 1,164
long.: 15,1405733
5 lat.: 37,5487245 14 175.673 52.335 1,7672
long.: 15,1429347
6 lat.: 37,553786 24 133.783 6.978 2,1024
long.: 15,14591
7 lat.: 37,553274 11 131.601 4.636 1,8432
long.: 15,144885
8 lat.: 37,54374 17 134.326 10.867 2,1008
long.: 15,141795
9 lat.: 37,525927 20 132.553 4.743 2,096
long.: 15,078637
10 lat.: 37,526034 17 143.901 19.931 2,0984
long.: 15,078421
GPS Coordinates

International Journal of Wireless & Mobile Networks (
To facilitate the understanding of how the RSSI values influence the other Key Performance
Indicators (throughput, RTT and MDEV) we have
with the RSSI.
The level of correlation was calculated by means of the Pearson correlation coefficient.
Given two statistical variables, X and Y, the Pearson correlation index is defined as their
covariance divided by the product of the standard deviations of the two variables:
where σxy is the covariance between
The Pearson correlation coefficient always takes values between
Generally, we can distinguish three types of correlation.
• if ρxy> 0, variables X and Y
• if ρxy= 0, variables X and Y
• if ρxy< 0, variables X and Y
In addition, for direct correlation (and similarly for inverse correlation) we can distinguish:
• if 0 <ρxy< 0.3 we have a weak correlation;
• if 0.3 <ρxy< 0.7 we have a moderate correlation;
• if ρxy> 0.7 we have a strong correlation.
TEST
POINT
1 lat
long
2 lat
long
3 lat 37, 5692652
long
4 lat
long
5 lat
long
6 lat
long
7 lat
long
8 lat
long
9 lat
long
10 lat
long
GPS Coordinates
Journal of Wireless & Mobile Networks (IJWMN) Vol. 8, No. 3, June 2016
Table 4. Suburban area.
Indicators (throughput, RTT and MDEV) we have calculated as these parameters were correlated
divided by the product of the standard deviations of the two variables:
ρxy =
between X and Y, and σx and σyare the two standard deviations.
The Pearson correlation coefficient always takes values between -1 and 1
-1 ≤ ρxy ≤ 1
Generally, we can distinguish three types of correlation.
are called directly correlated;
and Y are uncorrelated;
are called inversely correlated.
we have a weak correlation;
we have a moderate correlation;
we have a strong correlation.
RSSI
(ASU)
Average RTT
(ms)
MDEV
(ms)
Bandwidth
(Mbps)
37,5677718 5 135,753 14,922 0,4208
15,0046058
37,5684959 3 462,014 621,078 0
15,0039841
37, 5692652 3 367,022 521,051 0
15,0031255
37,5674959 1 Unreachable Unreachable 0
15,0054827 host net host net
37,5644837 5 141,27 26,645 0,2752
15,0015622
37,5675927 5 152,362 28,733 0,2896
15,0026377
37,5695428 5 162,457 27,232 0,3088
15,0039383
37,5645331 5 159,122 32,452 0,4256
15,0037947
GPS Coordinates
IJWMN) Vol. 8, No. 3, June 2016
65
calculated as these parameters were correlated
deviations.

Figure 6: Correlation coefficient between RSSI and Throughput, RTT and MDEV
From the results obtained we can evince how, in the case of the first test (Figure
what was expected, there seems to be no correlation between the values of power and
bandwidth.As already mentioned, these measurements were carried out in correspondence of high
RSSI values and, especially, in an area with a high concentrat
of a high level of commitment of the Mobile Network resources. Consequently, in these cases, we
should consider some additional variables not foreseeable in advance that considerably influence
the measured parameters.
: Correlation coefficient between RSSI and Throughput, RTT and MDEV for scenario 1
From the results obtained we can evince how, in the case of the first test (Figure 6), contrary to
RSSI values and, especially, in an area with a high concentration of mobile users, in a condition
66
for scenario 1
6), contrary to
ion of mobile users, in a condition
for scenario 2

These results certainly deserve further study, but seem to be
it is not the signal strength received to limit the available bandwidth, and neither to influence the
Round Trip Time values and their deviation standard.
Instead, by analyzing the results obtained in relation to the second series of measurements (Figure
7), we can see that, in this case, a higher correlation value was reached between RSSI and
Available Bandwidth. In this case, the tests were carried out in
crossing particularly "crowded" or "isolated" areas. Along the route, in addition, we could
experience a range of variability of the much larger RSSI. As we can see, when the RSSI values
decline to relatively low values (te
considerably. It is evident that when the RSSI drops below certain values it especially affects the
value of available bandwidth.
As regards the values of RTT and standard deviation, they are stil
results proposed are related to a suburban area with very poor cell coverage
one can evince from the RSSI values, the received signal levels are very scarce.
Due to the poor conditions of coverage,
measurement took too long. By convention, it was decided to put the value of available
bandwidth to zero when the measurement time exceeded 300 sec. It is also quite clear that very
low RSSI values (<= 3 ASU) lead to a significant instability of Internet connectivity.
In this scenario, by observing the correlation values we can see that there is a strong correlation
between RSSI and Available Bandwidth, RTT and MDEV (direct correlation between RSSI and
available bandwidth, inverse correlation between RSSI and RTT and between RSSI and MDEV).
From the results obtained from the three series of proposed measurements we can conclude that
there is a correlation between the RSSI and the measured parameters
when the RSSI values drop to gradually lower values.
In good received signal level conditions, the variables are decidedly uncorrelated since, in this
case, it was determined that it is certainly not the signal power to influence th
parameters measured but other parameters (e.g. traffic conditions, momentary network
congestion, number of mobile users in the cell, etc.).
These results certainly deserve further study, but seem to be enough to indicate that, in this case,
Round Trip Time values and their deviation standard.
), we can see that, in this case, a higher correlation value was reached between RSSI and
Available Bandwidth. In this case, the tests were carried out in a normal urban scenario, without
decline to relatively low values (test points 4, 5, and 7), the measured bandwidth value drops
As regards the values of RTT and standard deviation, they are still clearly uncorrelated.The latest
results proposed are related to a suburban area with very poor cell coverage (Figure 8)
one can evince from the RSSI values, the received signal levels are very scarce.
Due to the poor conditions of coverage, measurements often failed, and at times the bandwidth
3 ASU) lead to a significant instability of Internet connectivity.
there is a correlation between the RSSI and the measured parameters which becomes stronger
when the RSSI values drop to gradually lower values.
case, it was determined that it is certainly not the signal power to influence th
congestion, number of mobile users in the cell, etc.).
67
for scenario 3
enough to indicate that, in this case,
), we can see that, in this case, a higher correlation value was reached between RSSI and
a normal urban scenario, without
st points 4, 5, and 7), the measured bandwidth value drops
l clearly uncorrelated.The latest
(Figure 8). In fact, as
measurements often failed, and at times the bandwidth
3 ASU) lead to a significant instability of Internet connectivity.
which becomes stronger
case, it was determined that it is certainly not the signal power to influence the network

68
6. CONCLUSIONS AND FUTURE WORKS
The present paper provides an extended performance evaluation method of Quality of Service in
mobile Internet access. In particular, the paper proposes an algorithm for the throughput estimate
based on PathChirp technique, which we have seen to be preferable to PathLoad technique as it is
able to better approximate measurements using the FTP technique and converges with the desired
values in a significantly more rapid way. The convergence speed is a decisive factor especially if
applied to a context in which the measurements are taken by the use of drones. The use of drones
definitely has several advantages:
• possibility of moving in 3 dimensions and consequently greater spatial resolution of the
measurements;
• ability to record measurements in not easily accessible areas;
• carrying out measurements near the radio base station.
In contrast, a problem arises from the autonomy of the drone. Measurements must therefore be
carried out respecting certain timing constraints precisely due to the autonomy of the drone in
flight.During a single flight session it is possible to obtain several measurement points and
therefore have a more precise estimate of the QoS and increased spatial resolution.
ACKNOWLEDGMENT
We would like to thank HELI-LAB S.r.l., drone aerial filming company, and VICOSYSTEMS
S.r.l., company of telecommunications equipment design that has provided the modem-router
multi-sensor on board the drone.
REFERENCES
[1] Vinay J. Ribeiro, Rudolf H. Riedi, Richard G. Baraniuk Jiri Navratil, Les Cottrell, “pathChirp:
Efficient Available Bandwidth Estimation for Network Paths”, in Passive and Active Measurement
Workshop, April2003
[2] Manish Jain, ConstatninosDovrolis, “Pathload: a measurement tool for end-to-end availabla
bandwidth”, in Proc. Of Passive and Active Measurements (PAM) Workshop, 2002.
[3] R.S. Prasad, M.Murray, C.Dovrolis, K.Claffy, “Bandwidth Estimation: metrics, measurement
techniques and tools”, IEEE Network, Nov 2003
[4] C. Eschmann, C.-M. Kuo, C.-H. Kuo, C. Boller,” High-resolution multisensory infrastructure
inspection with unmanned aircraft systems”, International Archives of the Photogrammetry, Remote
Sensing and Spatial Information Sciences, Volume XL-1/W2, 2013 UAV-g2013, 4 – 6 September
2013, Rostock, Germany
[5] Md. Arafatur Rahman, “Enabling drone communications with WiMAX Technology”, Information,
Intelligence, Systems and Applications, IISA 2014, The 5th International Conference on
[6] LudovicApvrille, Yves Roudier, Tullio Joseph Tanzi, “Autonomous drones for disasters management:
Safety and security verifications”,Radio Science Conference (URSI AT-RASC), 2015 1st URSI
Atlantic
[7] Daniel Câmara, “Cavalry to the rescue: Drones fleet to help rescuers operations over disasters
scenarios”, Antenna Measurements & Applications (CAMA), 2014 IEEE Conference on
[8] LudovicApvrille, TullioTanzi ; Jean-Luc Dugelay, “Autonomous drones for assisting rescue services
within the context of natural disasters”, General Assembly and Scientific Symposium (URSI GASS),
2014 XXXIth URSI
[9] R.Viguier,C.,C.Lin ; K.Swaminathan, A.Vega, A.Buyuktosunoglu; S.Pankanti, P.Bose,
H.Akbarpour, F.Bunyak ; K.Palaniappan, G. Seetharaman, “Resilient mobile cognition:Algorithms,
innovations, and architectures - - Computer Design (ICCD), 2015 33rd IEEE International Conference
on

69
[10] AislanGomideFoina Raja Sengupta ; Patrick Lerchi ; Zhilong Liu ; Clemens Krainer, “Drones in
smart cities: Overcoming barriers through air traffic control research”, 2015 Workshop on Research,
Education and Development of Unmanned Aerial Systems (RED-UAS)
[11] UtkarshGoel, Mike P. Wittie, Kimberly C. Claffy, and Andrew Le , “Survey of End-to-End Mobile
Network Measurement Testbeds, Tools, and Services”, IEEE Communications Surveys & Tutorials
(Volume:18 , Issue: 1 ), Ott. 2015
[12] AshkanNikravesh, David R. Choffnes, Ethan Katz-Bassett, Z. Morley Mao, and Matt Welsh, “Mobile
Network Performance from User Devices: A Longitudinal, Multidimensional Analysis”, 15th
International Conference, PAM 2014, Los Angeles, CA, USA, March 10-11, 2014, Proceedings
[13] http://company.nokia.com/en/news/press-releases/2015/07/07/nokia-networks-du-first-in-uae-to-use-
telco-drones-for-better-network-planning-faster-optimization
[14] http://www.pcengines.ch/alix.htm
[15] F. Beritelli, A. La Corte, C. Rametta and F. Scaglione “A Cellular Bonding and Adaptive Load
Balancing based Multi-SIM Gateway for Mobile ad Hoc and Sensor Network” International Journal
on Ad Hoc Networking Systems (IJANS), Vol. 5, No. 3, July 2015
[16] F. Beritelli, A. La Corte, G. Lo Sciuto, C. Rametta, F. Scaglione, “Adaptive VPN BondingTechnique
for Enhancing Dual-SIM Mobile Internet Access”. Proc. of the International Symposium for Young
Scientists in Technology, Engineering and Mathematics (SYSTEM) 2015, Catania, Italy, September
27-29, 2015.
[17] http://www.zeroshell.net
[18] http://www.python.org/
[19] Manish Jain, ConstantinosDovrolis, “End-to-end Available Bandwidth: Measurements, Methodology,
Dynamic and Relation with TCP Throughput”, IEEE/ACM Transactions on Networking, Ago 2003.
[20] 3GPP TS 27.007 version 8.5.0 Release 8
AUTHORS
Francesco Beritelli received the Laurea degree in electronic engineering and the
Ph.D. degree in electronics, computer science, and telecommunications engineering
from the University of Catania, Catania, Italy, in 1993 and in 1997, respectively.
From 1997 to 2000 in collaboration with CSELT (now Telecom Italia LAB) took an
active part in international ITU-T standardization meetings. From 1998 he is
founding member of the “Multimedia Technologies Institute - MTI”, applied
research laboratory operating in DSP (Digital Signal Processor) and CTI (Computer
Telephony Integration) applications. Currently, from 2002, he is Assistant Professor
in the Department of Electric, Electronic and Computer Science Engineering at the
University of Catania. His main research activities are in the area of robust audio and speech signal
classification and recognition, variable bit-rate speech coding, and adaptive-rate voice and dual stream
transmission for mobile IP telephony applications, QoS in mobile Internet access and drone
communications. His interests also include the field of biometric identification and cardiac signal
processing. He is member of the Institute of Electrical and Electronic Engineers (IEEE) and the “IEEE
Signal Processing Society” since 1996 and has 120 scientific publications, mainly in international
journals, books and conference proceedings.
CorradoRametta received the Laurea degree in electronic engineering and the
Ph.D. degree in computer science and telecommunications engineering from the
University of Catania, Italy, in 2008 and in 2012, respectively. From 2009 to 2012,
he was a Research Engineer in the field of wireless communications with the CNIT
(ConsorzioNazionaleInteruniversitario per le Telecomunicazioni). From 2012 he
works as post-doc researcher at the University of Catania. His research interests
include wireless mesh networks, ad-hoc and sensor networks, modeling and
simulation of communications protocols, software defined networking, network
functions virtualization and embedded systems for multimedia applications.

70
Antonio Raspanti founder of Heli-Lab,drone aerial filming company. His research
interests include drone communications and performance evaluation techniques for
mobile Internet access.
Michele Russo received the Laurea degree in Computer Science Engineering from
the University of Catania, Italy, in 2010. His research interests include IoT systems,
VoIP Gateway (SIP compliant) and ICT technologies for environmental control and
monitoring.
Francesco Scaglionereceived the Laurea degree in Telecommunications
Engineeringfrom the University of Catania in 2011. His research interests include
embedded systems,VoIP communications and ICT technologies for environmental
control and monitoring.
GiuseppeSpallina founder of Heli-Lab,drone aerial filming company. His research
interests include drone communications and performance evaluation techniques for
mobile Internet access.

AN ADVANCED QOS ANALYSIS AND EVALUATION METHOD FOR MOBILE INTERNET ACCESS

More Related Content

What's hot

Viewers also liked

Similar to AN ADVANCED QOS ANALYSIS AND EVALUATION METHOD FOR MOBILE INTERNET ACCESS

Recently uploaded

AN ADVANCED QOS ANALYSIS AND EVALUATION METHOD FOR MOBILE INTERNET ACCESS