scispace - formally typeset
Search or ask a question
Proceedings ArticleDOI

A Framework for Cross-Layer Measurements in Wireless Networks

TL;DR: A framework for wireless network performance measurements with the scope of being as generic as possible is formulates, which pinpoint the necessity of a cross-layer approach and the importance of carefully selected and positioned equipment for the accuracy of the measurements.
Abstract: This paper formulates a framework for wireless network performance measurements with the scope of being as generic as possible. The methodology utilises a cross-layer approach in order to address the limitations of traditional layered techniques. A lot of work in the research community uses the channel power (Cp) to predict performance metrics in higher layers. There are currently two methods to measure Cp; either by using a spectrum analyser or from WiFi card information (RSSI). The paper discusses the correct configuration of a spectrum analyser (SA), to measure Cp. This paper, also provides a comparison of both SA and RSSI results produced inside an anechoic chamber for three different applications. The behaviour of the RSSI values showed significant discrepancy with both the SA results and what was intuitively expected. The results pinpoint the necessity of a cross-layer approach and the importance of carefully selected and positioned equipment for the accuracy of the measurements.

Summary (1 min read)

1 Introduction

  • In the last decade wireless communications have become omnipresent and the trends indicate that different wireless communication technologies will converge and work in conjunction with a wired network backbone.
  • The underlying issues of the wireless medium and how these influence higher layers is not yet understood in the research community.
  • In addition, the paper provides a discussion of different methods of assessing the physical layer performance of wireless networks and provides a comparison of their results after practical implementation.
  • The paper proposes the use and discusses the correct configuration of a spectrum analyser (SA) for measuring channel power under experiments taking place inside an anechoic chamber.

3.2 Setting the gate sweep parameters

  • In wireless measurements of channel power, the frequency characteristics of frames are very important.
  • This means that the SA will continue sweeping the desired frequency range (in their case Wi-Fi channel 6: 2.428 GHz - 2.448 GHz) after the frame has stopped being transmitted.
  • This is a serious problem when measuring channel power.
  • The gate mode is controlled by two parameters: i. Gate position: For iPerf, UDP packets were 1470 bytes (1534 bytes in total as measured from the Fluke) and for iTunes the packets were 1472 bytes (1536 in total as seen from the Fluke).

4 Practical Results and Discussion

  • A difficulty in performing these measurements is that ideally the receiving host (Host 1 in their case), dipole and monitoring devices should be in exactly the same location to ensure that the same power is measured in each case.
  • Both Fluke and Asus capture more frames when the AP was closer despite the reduced number of retransmissions.
  • Asus’ RSSI values do not indicate the real signal strength conditions as reported by the SA and the Fluke.
  • The value of Triggers for experiment (c) indicates that more packets have been transmitted by Host 2 in comparison to experiment (a).

Did you find this useful? Give us your feedback

Content maybe subject to copyright    Report

This item was submitted to Loughborough’s Institutional Repository
(https://dspace.lboro.ac.uk/) by the author and is made available under the
following Creative Commons Licence conditions.
For the full text of this licence, please go to:
http://creativecommons.org/licenses/by-nc-nd/2.5/

A Framework for Cross-Layer Measurements in Wireless Networks
Konstantinos G. Kyriakopoulos,William G. Whittow, David J. Parish
Department of Electronic and Electrical Engineering
Loughborough University, Loughborough, LE11 3TU, U.K.
e-mail: {elkk, elwgw, d.j.parish}@lboro.ac.uk.
Abstract
This paper formulates a framework for wireless net-
work performance measurements with the scope of being as
generic as possible. The methodology utilises a cross-layer
approach in order to address the limitations of traditional
layered techniques. A lot of work in the research commu-
nity uses the channel power (Cp) to predict performance
metrics in higher layers. There are currently two meth-
ods to measure Cp; either by using a spectrum analyser or
from WiFi card information (RSSI). The paper discusses the
correct configuration of a spectrum analyser (SA), to mea-
sure Cp. This paper, also provides a comparison of both SA
and RSSI results produced inside an anechoic chamber for
three different applications. The behaviour of the RSSI val-
ues showed significant discrepancy with both the SA results
and what was intuitively expected. The results pinpoint the
necessity of a cross-layer approach and the importance of
carefully selected and positioned equipment for the accu-
racy of the measurements.
1 Introduction
In the last decade wireless communications have become
omnipresent and the trends indicate that different wire-
less communication technologies will converge and work
in conjunction with a wired network backbone. There has
been a lot of research on how to improve the already es-
tablished protocols (TCP/UDP) and to optimise them for a
wireless channel medium. However, the underlying issues
of the wireless medium and how these influence higher lay-
ers is not yet understood in the research community.
In order to improve the performance of hybrid wireless
and wired networks the research community needs to exam-
ine the nature of the protocol stack as a whole and not sep-
arate it into layers. Examining just the higher layers does
not reveal much information about the issues of the lower
layers and similarly examining only lower layers and their
parameters does not provide any information on their effect
on the application layer. Thus, to understand the impact of
the wireless channel medium on the upper layers, a cross-
layer approach is required.
This paper formulates a framework for wireless network
performance measurements in a cross-layer manner with the
aim of being as generic as possible, but applied here to a
specific wireless technology. In addition, the paper provides
a discussion of different methods of assessing the physical
layer performance of wireless networks and provides a com-
parison of their results after practical implementation. The
paper proposes the use and discusses the correct configu-
ration of a spectrum analyser (SA) for measuring channel
power under experiments taking place inside an anechoic
chamber.
The rest of the paper is structured as follows: In Section
2, related work on current methods for measuring wireless
network performance is presented along with some discus-
sion on their shortcomings. Section 3 is divided into two
parts: the first part describes the test-bed in which the ex-
periments took place in and the second part explains the
gate mode and how to carefully set the spectrum analy-
sers parameters. Section 4 presents the practical results
of the cross-layer measurements from three applications:
ping, iPerf and iTunes (video streaming). Finally, Section 5
presents the conclusions.
2 Related Work
Initially, the research community tried to measure wire-
less networks from a wired vantage point, i.e from a wired
host having some knowledge of the wireless network and in
some cases through the use of SNMP logs [4]. However,
SNMP logs provide summarised information polled period-
ically and does not expose the instantaneous characteristics
of the wireless channel, which is important for traffic char-
acterisation and network diagnosis (security) purposes.
The research developed by Maryland University [7, 8]
suggests deploying multiple wireless network hosts captur-
ing traffic from the wireless network. The advantages of
this methodology are that sniffers do not interact with the
network, as they are just passive monitoring devices. In
2009 Fifth Advanced International Conference on Telecommunications
978-0-7695-3611-8/09 $25.00 © 2009 IEEE
DOI 10.1109/AICT.2009.47
237
Authorized licensed use limited to: LOUGHBOROUGH UNIVERSITY. Downloaded on June 26, 2009 at 07:42 from IEEE Xplore. Restrictions apply.

addition, they provide detailed, per frame PHY/MAC infor-
mation of the data such as Received Signal Strength Indica-
tor (RSSI), noise and signal strength, throughput and error
rates. The above papers try to address the following chal-
lenges that arise from this methodology: i) Find an ideal
location for monitoring the wireless traffic, ii) address the
limited capabilities for each sniffer, i.e. signal strength, pro-
cessing power, disk space, and iii) combine the captured
traces from various sniffers to provide a better view of the
wireless traffic as some sniffers have a better view of an area
than others.
The above references do not provide a cross-layer ap-
proach to wireless network measurements. The work in [6]
is very close in nature to the above references but follows a
cross-layer approach. In [6], two models for mapping Sig-
nal to Noise (SNR) to throughput are presented and verified
with measurements from three applications: i) iPerf, ii) FTP
and iii) LANFielder. The same models could be extended
to new applications.
However, [6] has the following disadvantage; according
to the IEEE 802.11 standard, the RSSI value is a 1-byte
value (max value 255) that maps the RF energy received by
the chipset of the wireless card. It is intended for use by the
Wi-Fi card internally between the link and physical layers.
It is not intended to be of any particular accuracy for mea-
suring actual signal strength. Furthermore, the 802.11 stan-
dard does not define any particular mechanism nor requires
that all 255 values should be used. Thus, each vendor has
different maximum RSSI values and usually uses its own
(often undisclosed) methodology to map the RSSI to the re-
ceived power (in dBm) [5]. The RSSI values are therefore a
source of confusion and uncertainty [5].
Finally, Naples University [1, 2, 3] follows a different
approach with the aim of correlating the values of major
physical layer quantities (i.e. Channel power, SNR, Sig-
nal to Interference Ratio (SIR)) in the wireless channel to
those characterising the key higher layers parameters. The
goal was to assess the performance of one protocol layer as
a function of that of another or several other layers. The
uniqueness of this approach is that it does not depend on
the RSSI values taken from the firmware of the NIC but on
signal and noise measurements taken from a spectrum anal-
yser, independently in a semi-anechoic chamber.
3 Methodology
3.1 Testbed
It is difficult to measure electromagnetic waves from a
single source as there is interference from many external
objects, such as mobile phones and local area networks etc.
It is also difficult to control the reflections from surfaces
such as tables and walls. An anechoic chamber is a solution
to these problems and consists of a metallic box with its
interior walls covered in radar absorbing material (RAM).
The metal box shields the experiment from external sources,
while the RAM absorbs a very high percentage of the waves
from the source and effectively places the experiment in
an infinitely large space with minimal reflections from the
walls.
The methodology that the authors followed is actually
very close in concept to that of Naples University [1, 2, 3]
with the following differences: The experiments happen
inside a fully Anechoic chamber and a Network Protocol
Analyser (NPA) was utilised to generate the trigger for con-
trolling the gate mode of the SA. In addition, we measure
RSSI values from two wireless network monitoring devices.
Therefore, in this work the Naples and the Maryland meth-
ods are combined to create a new measurement framework.
This helps to conclude how accurate RSSI values are for
measuring channel power at the PHY layer.
The following devices were used in the testbed: 1) A D-
link DWL-900AP+ Access Point (AP). 2) A Toshiba Satel-
lite Pro L300 laptop running Windows as Host 1. 3) A sec-
ond Toshiba Satellite Pro L300 laptop acted as a wireless
monitoring device running Backtrack Linux using an Asus
Wi-Fi card with the Ralink rt73 chipset. This chipset was
selected because it supports raw monitoring mode (rfmon),
which allows wireless monitoring. The Wireshark tool was
used for wireless frame capturing. 4) A Macbook laptop as
Host 2. 5) A Netgear DS 108 hub. 6) A Fluke Optiview Se-
ries III Integrated Network Analyser with a wireless NIC.
7) A Spectrum Analyser (SA) Advantest R3182 connected
to a 2.4 GHz dipole. 8) An Anritsu MD1230A Network
Protocol Analyser (NPA).
Inside the full-anechoic chamber were located the AP,
a laptop (Host 1), the monitoring devices: a laptop run-
ning Wireshark under Linux (with Asus Wi-Fi card) and
the Fluke device, and a directional antenna as shown in
Fig. 1. The Fluke provides Signal Strength and the Asus
monitoring device provides RSSI values. The reason for
using two devices was to compare different manufacturers
tools because, as was discussed before, different manufac-
turers use different techniques to calculate the RSSI/Signal
Strength information. In this case, we compare the Ralink
rt73 chipset (Asus) with the Fluke Wi-Fi card.
A directional antenna (dipole) was attached to the SA in
order to measure channel power. A dipole antenna radiates
equally in all directions (isotropically) in a plane perpen-
dicular to the dipole. However, the dipole has a null at ei-
ther end. These can be thought of as blind spots where the
antenna has difficulty receiving (or sending) signals. The
dipole antenna in our experiments was vertically orientated
to minimise the effect of the laptop (Host 1) on the mea-
sured channel power (see Fig. 2a).
In our experiments (see Section 4), the majority of the in-
238
Authorized licensed use limited to: LOUGHBOROUGH UNIVERSITY. Downloaded on June 26, 2009 at 07:42 from IEEE Xplore. Restrictions apply.

Anechoic chamber
A.P.
Host 1
HUB
S.A.
N.P.A.
Trigger
Monitoring
Laptop
Antenna
Host 2
X
FLUKE
Figure 1. A diagram of the testbed used.
formation was sent from Host 2 (via the AP) to Host 1. Host
1 sent back replies and acknowledgements. As Host 1 was
much closer to the dipole than the AP, its contribution to the
channel power would have been much larger if the dipole
had been orientated horizontally or an omni-directional an-
tenna had been used. A section of RAM was placed be-
tween the dipole and Host 1 (see Fig. 2 (a)). This reduced
the channel power by 12 dBm in a specific test when only
Host 1 was transmitting information.
Host 2 was connected with Ethernet to a hub, which
propagated frames to the AP in the chamber and to the An-
ritsu NPA outside the chamber. The latter was set to monitor
mode in order to monitor the frames received by the access
point and generated triggers when several predefined condi-
tions were met. In our case, the conditions were the specific
IP source address (of Host 2) and destination IP address of
Host 1. The trigger was used by the SA as a sign to start
sweeping. Therefore, the SA only measured the channel
power when a frame is sent from Host 2 to Host 1.
(a) (b)
Figure 2. (a) A photo of the anechoic cham-
ber with the equipment. The AP is behind
the camera. (b) A photo of the oscilloscope
displaying the trigger 0.5 ms before an icmp
echo request packet.
3.2 Setting the gate sweep parameters
In wireless measurements of channel power, the fre-
quency characteristics of frames are very important. How-
ever, usually the frame duration in a wireless network is
very small, in fact smaller than the sweep time of the SA.
This means that the SA will continue sweeping the desired
frequency range (in our case Wi-Fi channel 6: 2.428 GHz
- 2.448 GHz) after the frame has stopped being transmit-
ted. The SA will not measure a signal inbetween frames
and therefore the constructed frequency spectrum will be
damaged as “holes” are introduced in the frequencies swept
at this time (Fig. 3 (a)). This is a serious problem when
measuring channel power. In order to circumvent this phe-
nomenon, the spectrum analyser should be set in gate mode
(Fig. 3 (b)). The gate mode is controlled by two param-
eters: i. Gate position: The time instance the SA should
start sweeping - achieved with an external trigger. This is
the delay between the NPA seeing the packet in the hub and
the packet being actually modulated and transmitted from
the AP. ii. Gate width: The duration of time for which the
SA measures the signal. Gate mode allows one frequency
sweep to be composed of several packets.
Trigger
sweep
sweep when there is no frame
Frame 1 Frame 2
Amplitude
Frequency
(a)
sweep
resume
sweep
at next
frame
Gap because frame stopped
- Waiting for next trigger
Trigger
Trigger
Frequency
Time
(b)
Figure 3. (a) The graph represents the dis-
play of a SA when the sweep captures in-
stances when no signal is present (SA in non-
gate mode). (b) In the gate mode, the SA re-
ceives a trigger and starts sweeping for a pre-
defined time duration (gate width) and then
pauses until another trigger is received, re-
suming the sweep of the subsequent portion
of the frequency spectrum.
Before starting the practical experiments, the above gate
parameters were examined for all three considered applica-
tions (ping, iPerf and iTunes). In order to achieve this, spe-
cial alteration was made to the hardware of the AP. Specifi-
cally, an intermediate frequency (IF) signal at 380 MHz was
taken from the AP to allow connection to an oscilloscope
(DSO) for examination of the signal.
By triggering the DSO with the output from the NPA, we
239
Authorized licensed use limited to: LOUGHBOROUGH UNIVERSITY. Downloaded on June 26, 2009 at 07:42 from IEEE Xplore. Restrictions apply.

Table 1. Parameters used in gate mode for
each application
Ping iPerf iTunes
Packet length in 802.11 (bytes) 120 1534 1536
Gate width (ms) 0.258 1.44 1.44
Gate position (ms) 0.5 1.06 1.06
could measure accurately both the gate position and gate
width (see Fig. 2 (b)). The experiments showed that the
trigger distance and the frame duration depend on the length
(in bytes) of the packet that is modulated. Each of the three
applications that are considered in this paper has constant
packet lengths.
For ping; Host 2 sends echo requests. For iPerf; Host 2
is the client and sends UDP packets and for iTunes, Host 2
streams the video. The channel power from the spectrum
analyser was measured using packets from Host 2 to Host
1 (as the trigger of the SA is generated under these condi-
tions) and for averaging the RSSI/Signal Strength values of
the monitoring devices, only packets from Host 2 to Host 1
were used.
For the ping application, we used the default ping size
of 64 bytes (120 bytes as measured from the Fluke, i.e. 64
bytes + 20 for IP header + 8 for LLC + 28 for 802.11). For
iPerf, UDP packets were 1470 bytes (1534 bytes in total as
measured from the Fluke) and for iTunes the packets were
1472 bytes (1536 in total as seen from the Fluke). The pa-
rameters for each application are presented in Table 1. Note.
without the gate mode, the channel power was reduced by
11 dBm, 7dBm and 1.5 dBm for ping, iPerf and iTunes re-
spectively.
4 Practical Results and Discussion
A difficulty in performing these measurements is that
ideally the receiving host (Host 1 in our case), dipole and
monitoring devices should be in exactly the same location to
ensure that the same power is measured in each case. Unfor-
tunately, in practice this is impossible. Furthermore, if the
objects are placed close together, they will interfere with
each other electromagnetically. This is due to the dipole
measuring unwanted signals emitted from the receiving host
and to the reflections caused by having passive metal objects
in close proximity to the antennas. Note, Host 1, Fluke and
the monitoring laptop all contain antennas and their ability
to receive signals may be reduced by surrounding objects,
their orientation and due to the behaviour of the specific an-
tennas. In the course of our experiments, we discovered
that if the RAM was placed in certain geometries it could
adversely affect the performance of the antennas.
In this section, the cross-layer measurement results for
ping, iPerf and iTunes are presented. The ping program
on Host 2 generated 600 ping requests. iPerf was used as
a UDP packet generator with a bit rate of 1 Mbps and a
duration of 60 seconds. During this period of time, 5104
UDP packets were generated from Host 2. iTunes streamed
two video clips to Host 1. Video 1 had a duration of 4:54
and Video 2 was 6:15 minutes long. Fluke and Asus av-
eraged Signal Strength/RSSI values were compared against
the measured channel power from the SA in order to exam-
ine their accuracy. In addition Fluke and Asus performance
in wireless monitoring was compared.
Table 2 shows the results for ping. As the distance of the
AP from Host 1 decreased, the higher layer results (aver-
age Round Trip Time (RTT) and duplicate packets) were
improved. The channel power (Cp) measured from the
SA, and the Fluke averaged Signal Strength increased as
the distance decreased, which clearly reveals stronger sig-
nal intensity. Fluke also captured less retransmissions on
the wireless link and slightly less frames in total at the
shorter distance. Total frames for the Fluke device include
frames from both directions of communication and retrans-
missions.
The RSSI results from the Asus Wi-Fi card agree with
the Fluke and SA in terms of signal strength. However, the
captured frames are much less in comparison to the number
of captured frames from the Fluke. One reason for this is
due to a possible driver - software - hardware compatibil-
ity problem with the specific instrumentation that did not
capture frames generated from Host 1 and destined to the
AP, thus limiting the possible capture capability to one way
communication (AP to Host 1). Even though the capturing
process was limited to one way, Asus still dropped a lot of
frames that it was expected to capture.
For the iPerf application (Table 3), the SA channel
power, Fluke’s average Signal Strength and jitter follow the
same behaviour as with the ping program, i.e. enhanced
performance when the distance decreased. The Asus Wi-
Fi card does not show increasing performance in capturing
packets when the AP is closer. Surprisingly its averaged
RSSI value remains very close to the value at 4.5 m. This
clearly shows the prohibiting performance of specific Wi-Fi
cards when accurate physical layer measurements are re-
quired due to the RSSI limitations.
Tables 4, 5 show the results for streaming two different
video files in iTunes. The metric named Triggers refers
to the number of packets seen by the NPA. These Trigger
packets meet the specified triggering conditions. The num-
ber of Triggers is ideally the same as the number of packets
that both Fluke and Asus should capture, ignoring packets
from Host 1 and retransmissions at the MAC layer of the
240
Authorized licensed use limited to: LOUGHBOROUGH UNIVERSITY. Downloaded on June 26, 2009 at 07:42 from IEEE Xplore. Restrictions apply.

Citations
More filters
Proceedings ArticleDOI
12 May 2011
TL;DR: A novel approach for monitoring large and complex wireless networks, fast deployed which operate seamlessly and in real time and is highly applicable for unmanaged and partly managed wireless networks such as Ad-hoc, first responders, self deployed and any highly dynamic network.
Abstract: Monitoring and analyzing wireless networks for network structure and behavior is a complex task. Such monitoring often requires creating extra traffic, dedicated hardware and a prior knowledge of the network components and structure. In this paper we present a novel approach for monitoring large and complex wireless networks, fast deployed which operate seamlessly and in real time. The suggested framework uses few passive sniffers in order to sample the WiFi communication in the "air" per packet and have an extended cover range due to overhearing abilities. This monitoring system requires no prior knowledge of the network structure. We have designed, implemented and deployed such a passive monitoring system and used it to monitor the campus WLAN network (Wi-Fi). Experimental results show that the suggested framework is highly applicable for unmanaged and partly managed wireless networks such as Ad-hoc, first responders, self deployed and any highly dynamic network.

9 citations

Journal ArticleDOI
TL;DR: An experimental analysis of the effects of cross traffic on the performance of video streaming over Wi-Fi, based on cross-layer measurements and objective video quality metrics evaluated through a standardized approach.

5 citations


Cites background from "A Framework for Cross-Layer Measure..."

  • ...In the recent past, cross-layer measurements have come out to be a powerful option to assess and predict the performance of wireless and hybrid networks, as well as to troubleshoot them [9-16]....

    [...]

Proceedings ArticleDOI
21 Nov 2017
TL;DR: This work uses the famous Root Cause Analysis (RCA) technique for comparing traces issued from different simulations and real experiments, that includes the study of the root causes of dissimilarities, to detect and analyze a performance anomaly between NS-3 simulation and lab wireless testbed when transmitting data over a WIFI 802.11 link.
Abstract: Network simulators are often used for their simplicity and cost regarding wireless networks. However, their realism is often criticized and their results challenged. The main concern comes from the modeling of the PHY and MAC layers. To assess the performances of these simulators and their models, the results of simulations are often compared with experimental results. However, the comparison methodologies used in these studies may introduce biases. This work focuses on accurately discovering and analyzing the reasons for the calibration problems or implementation bugs in simulators and experimental devices. For this purpose, we leverage the famous Root Cause Analysis (RCA) technique for comparing traces issued from different simulations and real experiments, that includes the study of the root causes of dissimilarities. Throughout the paper, our RCA-based method has been applied to detect and analyze a performance anomaly between NS-3 simulation and our lab wireless testbed when transmitting data over a WIFI 802.11 link. It especially details how low level traffic traces have been generated in both environments for similar scenarios, and how they can accurately be compared and their differences analyzed.

3 citations


Cites background from "A Framework for Cross-Layer Measure..."

  • ...They generally are built in an anechoic room for being able to control the air environment that has to be free of external signal, and in which it is possible to control interferences and noise injection [12], [13]....

    [...]

Proceedings ArticleDOI
03 May 2010
TL;DR: Cross layer measurement is suggested as a solution to the problem of understanding and analysis of such complex communication issues and a framework in which appropriate performance measurements can be made from a WiFi network supporting a video streaming application is proposed.
Abstract: Wireless networks such as WiFi suffer communication performance issues in addition to those seen on wired networks due to the characteristics of the radio communication channel used by their Physical Layers (PHY). Understanding these issues is a complex but necessary task given the importance of wireless networks for the transfer of wide ranging packet steams including video as well as traditional data. Simulators are not accurate enough to allow all the intricacies of such communication to be accurately understood, especially when complex interactions between the protocols of different layers occurs. The paper suggests cross layer measurement as a solution to the problem of understanding and analysis of such complex communication issues and proposes a framework in which appropriate performance measurements can be made from a WiFi network supporting a video streaming application. The framework has been used to collect these measurements at the PHY, MAC, Transport and Application layers. Analysis of the collected measurements has allowed the effects of noise interference at the PHY to be related to the perceived performance at the Application Layer for a video streaming application. This has allowed the effect of the SNR on the download time of a video sequence to be studied.

2 citations


Cites background from "A Framework for Cross-Layer Measure..."

  • ...In the recent past, cross-layer measurements have come out to be a powerful option to assess and predict the performance of wireless and hybrid networks, as well as to troubleshoot them [9-14]....

    [...]

Dissertation
20 Oct 2014
TL;DR: Dans cette optique, nous proposons un modele d'analyse de cause racine (RCA) concu pour detecter les differences de configuration, d'implementation ou de modelisation entre le simulateur and le banc de mesure a partir of donnees issues of traces experimentales.
Abstract: La multiplication des reseaux sans-fil de l'Internet des Objets aura pour consequence l'augmentation des problemes d'acces au medium et une forte pollution spectrale. Pour maintenir la qualite des communications, les noeuds devront devenir plus adaptatifs et exploiter un maximum d'informations sur l'etat du medium et des communications. Dans cette optique, cette these presente 3 contributions ayant respectivement traits a la mesure des reseaux sans-fil, a leur modelisation et a la recherche de solutions d'adaptabilite accessibles pour les noeuds de ces futurs reseaux. Par consequent, la 1re des contributions concerne la conception et la mise en oeuvre d'un banc de mesure experimental compatible avec les besoins de ces etudes (i.e. fournir des possibilites de mesures cross-layer a partir du niveau physique et un controle maximal des perturbations du medium). La solution developpee est celle d'un banc concu a l'interieur d'une chambre anechoique dans laquelle du bruit perturbateur est injecte a l'aide d'une antenne directionnelle. a l'interieur de la chambre, les mesures sont effectuees a l'aide d'equipements WIFI et d'equipements de mesure RF. La 2e contribution vise a prendre en compte les mesures issues de ce banc experimental dans le but de tester et d'ameliorer le realisme du simulateur ns- 3 et de ses modeles. En effet, malgre leur peu de realisme, les simulateurs de reseau comme ns-3 sont utilises pour tester de nouveaux protocoles ou de nouvelles applications sans-fil. Dans cette optique, nous proposons un modele d'analyse de cause racine (RCA) concu pour detecter les differences de configuration, d'implementation ou de modelisation entre le simulateur et le banc de mesure a partir de donnees issues de traces experimentales. L'application de ce modele a conduit a une amelioration importante du realisme des modeles WIFI du simulateur. La 3e et derniere contribution consiste a appliquer les algorithmes d'apprentissage SVR, k-nn et DT en vue de l'estimation predictive du debit IP mesure sur un lien sans-fil. Les estimations se font respectivement a partir des valeurs de SNR, de RSS et de bruit mesurees au niveau du noeud recepteur. Les differents algorithmes sont evalues selon la precision de leurs estimations mais aussi sur leurs caracteristiques fonctionnelles (e.g. taille des modeles, ...). Les resultats indiquent que les algorithmes SVR et DT utilises avec le SNR permettent les estimations les plus precises. De plus, ces 2 algorithmes offrent les meilleures performances respectivement en termes de memoire utilisee par le modele et de temps de calcul.

1 citations


Cites background from "A Framework for Cross-Layer Measure..."

  • ...Le travail détaillé dans ce chapitre suit les recommandations des travaux désignés dans l’état de l’art sous l’appellation bancs de laboratoire [15, 66]....

    [...]

  • ...Dans [66], des mesures effectuées en chambre anéchöıques sont utilisées pour trouver les caractéristiques statistiques des mesures de puissance du signal obtenues sur du matériel standard....

    [...]

  • ...En revanche, le banc décrit dans ce chapitre diffère des travaux actuels par plusieurs aspects qui améliorent les mesures recueillies : • la précision des mesures est ainsi rendue meilleure de part les équipements et les méthodes utilisés : par exemple, les auteurs de [15] utilisent une chambre semi-anéchöıque et [66] utilise plusieurs antennes de réception....

    [...]

  • ...(c) Alternativement à la solution (b), du bruit peut-être injecté dans l’environnement pour augmenter le nombre d’erreurs et la difficulté de réception [66]....

    [...]

  • ...Néanmoins, étant données les qualités de cet environnement pour la mesure RF, son utilisation s’est étendue à la mesure réseau et notamment dans la conception de nombreuses plates-formes de mesure du lien sans-fil comme dans [15, 66]....

    [...]

References
More filters
Proceedings ArticleDOI
01 Jun 2002
TL;DR: The goals of this study are to extend the understanding of wireless user behavior and wireless network performance, and to characterize wireless users in terms of a parameterized model for use with analytic and simulation studies involving wireless LAN traffic.
Abstract: This paper presents and analyzes user behavior and network performance in a public-area wireless network using a workload captured at a well-attended ACM conference. The goals of our study are: (1) to extend our understanding of wireless user behavior and wireless network performance; (2) to characterize wireless users in terms of a parameterized model for use with analytic and simulation studies involving wireless LAN traffic; and (3) to apply our workload analysis results to issues in wireless network deployment, such as capacity planning, and potential network optimizations, such as algorithms for load balancing across multiple access points (APs) in a wireless network.

566 citations


"A Framework for Cross-Layer Measure..." refers background in this paper

  • ...e from a wired host having some knowledge of the wireless network and in some cases through the use of SNMP logs [4]....

    [...]

Proceedings ArticleDOI
01 Oct 2004
TL;DR: The pitfalls that an actual wireless monitoring system for an IEEE 802.11 based wireless network needs to be aware of are identified, and feasible solutions to avoid those pitfalls are provided.
Abstract: Many studies on measurement and characterization of wireless LANs (WLANs) have been performed recently. Most of these measurements have been conducted from the wired portion of the network based on wired monitoring (e.g. sniffer at some wired point) or SNMP statistics. More recently, wireless monitoring, the traffic measurement from a wireless vantage point, is also widely adopted in both wireless research and commercial WLAN management product development. Wireless monitoring technique can provide detailed PHY/MAC information on wireless medium. For the network diagnosis purpose (e.g. anomaly detection and security monitoring) such detailed wireless information is more useful than the information provided by SNMP or wired monitoring. In this paper we have explored various issues in implementing the wireless monitoring system for an IEEE 802.11 based wireless network. We identify the pitfalls that such system needs to be aware of, and then provide feasible solutions to avoid those pitfalls. We implement an actual wireless monitoring system and demonstrate its effectiveness by characterizing a typical computer science department WLAN traffic. Our characterization reveals rich information about the PHY/MAC layers of the IEEE 802.11 protocol such as the typical traffic mix of different frame types, their temporal characteristics and correlation with the user activities. Moreover, we identify various anomalies in protocol and security of the IEEE 802.11 MAC. Regarding the security, we identify malicious usages of WLAN, such as email worm and network scanning. Our results also show excessive retransmissions of some management frame types reducing the useful throughput of the wireless network.

207 citations


"A Framework for Cross-Layer Measure..." refers background in this paper

  • ...The research developed by Maryland University [7, 8] suggests deploying multiple wireless network hosts capturing traffic from the wireless network....

    [...]

DOI
05 Jun 2005
TL;DR: The results indicate that WM enables reliable analysis of the collected traces, and should encourage the wireless research community to use this technique for a wide variety of research areas, such as traffic analysis, user mobility and handoff analysis, and MAC/PHY anomaly detection.
Abstract: Wireless monitoring (WM) is a passive approach for capturing wireless-side traffic with rich MAC/PHY layer information. WM can suffer, however, from low capture performance, i.e., high measurement loss, due to the unreliable wireless medium. In this paper, we experimentally show that WM can perform reliable and accurate measurements on wireless traffic, in actual, non-ideal channel conditions.We demonstrate how to increase capture performance by merging traces from multiple monitoring devices. This merging enables WM to capture over 99% of the IP layer traffic and over 97% of the MAC/PHY frames in a controlled experiment. Our results indicate that WM enables reliable analysis of the collected traces, and should encourage the wireless research community to use this technique for a wide variety of research areas, such as traffic analysis, user mobility and handoff analysis, and MAC/PHY anomaly detection.

65 citations


"A Framework for Cross-Layer Measure..." refers background in this paper

  • ...The research developed by Maryland University [7, 8] suggests deploying multiple wireless network hosts capturing traffic from the wireless network....

    [...]

Journal ArticleDOI
TL;DR: A cross-layer approach is presented, which provides for several measurements to be concurrently carried out at different layers through a proper automatic station to correlate the values of the major physical-layer quantities exhibited by those characterizing the key higher layers' parameters in the presence of interference.
Abstract: Assessing the overall performance of wireless communication networks is of key importance for optimal management and planning. With special regard to wireless networks operating in an unlicensed band, evaluating overall performance mainly implies facing the coexistence issues, which are associated with the contemporaneous presence of true and interfering signals at the physical layer. This task is difficult to fulfill only on the basis of single-layer measurements, if not prohibitive; a partial perspective of network behavior would, in fact, be gained. With this concern, a cross-layer approach is presented hereinafter. It provides for several measurements to be concurrently carried out at different layers through a proper automatic station. It aims to correlate the values of the major physical-layer quantities (e.g., channel power and signal-to-interference ratio) exhibited by those characterizing the key higher layers' parameters (e.g., packet-loss ratio and one-way delay) in the presence of interference. A first step toward a full characterization of how the effects of a problem, which is experienced at the physical layer, propagates along the whole protocol stack, can thus be taken.

43 citations

Journal ArticleDOI
TL;DR: Results of several measurements aimed at establishing the sensitivity of IEEE 802.11b carrier sensing mechanisms to continuous interfering signals and evaluating the effects of triggered interference on packet transmission are assessed.
Abstract: Researches and development efforts in wireless networking and systems are progressing at an incredible rate. Among them, measurement and analysis of performance achieved at network layer and perceived by end users is an important task. In particular, recent advances concerning IEEE 802.11b-based networks seem to be focused on the measurement of key parameters at different protocol levels in a cross-layered fashion, because of their inherent vulnerability to in-channel interference. By adopting a cross-layer approach on a real network set-up operating in a suitable experimental testbed, packet loss against signal-to-interference ratio in IEEE 802.11b-based networks is hereinafter assessed. Results of several measurements aimed at establishing the sensitivity of IEEE 802.11b carrier sensing mechanisms to continuous interfering signals and evaluating the effects of triggered interference on packet transmission.

23 citations

Frequently Asked Questions (2)
Q1. What contributions have the authors mentioned in the paper "A framework for cross-layer measurements in wireless networks" ?

This paper formulates a framework for wireless network performance measurements with the scope of being as generic as possible. The paper discusses the correct configuration of a spectrum analyser ( SA ), to measure Cp. This paper, also provides a comparison of both SA and RSSI results produced inside an anechoic chamber for three different applications. 

As for future work, the trigger pulses that control the SA should ideally be generated from the AP that transmits and receives the frames on the wireless channel rather than from an external monitoring device ( NPA in their case ). A DSO captures the examined signal in the time domain and allows further processing of the signal information off-line.