[ieee 2012 ieee student conference on research and development (scored) - pulau pinang, malaysia...

An Experimental Study of 802.11 Access Point

Network Behavior

#N.A. Abd Ghafar

1, W. Hashim

1, A.F. Ismail

2 S. Dzulkifly

1 and K. Abdullah

2

1Wireless Communication Group, MIMOS Berhad, Kuala Lumpur, Malaysia

2Dept. of ECE, Faculty of Engineering, International Islamic University Malaysia, Kuala Lumpur, Malaysia

[email protected],

[email protected],

[email protected],

[email protected], and

[email protected]

Abstract—The paper presents a summary of related

literatures and the findings from preliminary network

behavior analyses of an IEEE 802.11 Wireless Local Area

Networks public access point. The objective of the study is to

investigate and identify the wireless network’s characteristics.

Received Signal Strength from a public access point was

examined and the internet connection speed through the said

access point, from end user’s perspective was scrutinized.

Correlation of internet connection performance based on

received signal strength from the access point was put under

evaluation. Such notion has been evaluated and likely to be

refuted based on the experimental findings. Having a high

received signal strength alone does not warrant a good internet

connection speed since there are likely to be other influencing

factors and parameters involved. Since classifying the behavior

of an access point is not always straightforward, this study

confines the monitoring of IEEE 802.11 Wireless Local Area

Network behaviors from the perspective of access point users.

Index Terms—IEEE 802.11, network behavior, wireless

LAN, quality of service (QoS), signal strength

I. INTRODUCTION

In today's fast-paced, technologically-driven world, there has been a widespread growth in the usage of wireless local area network (WLAN). The WLAN IEEE 802.11 or more commonly known as WiFi has been becoming the connection mode of choice. This is partly due to substantial price drop of the peripherals required namely the wireless access points (APs) and the wireless network cards. The decreasing cost of fixed broadband services, including digital subscriber line (DSL) and fiber-to-home (FTH); as well as the increases in WLAN data rates have also been the reasons for the enormous deployment of WLAN over the last few years. Most universities, schools and public places offer WiFi connectivity and internet users are becoming more dependent on laptops, PDAs, iPads, smart phones and other technological gadgets to get connected to the internet.

In the default setting of most common set-up, a WLAN users or WLAN receiver stations will execute scanning either “active” or “passive” to gain the information of available AP or APs. In passive scanning mode, beacon frames are periodically sent out by available APs containing information about the AP including service set identifier (SSID) and supported data rates. The user equipment (UE) WLAN transceiver then computes the received signal

strength (RSS) by measuring the power present the receiver against the originally transmitted peak power. Active scanning on the other hand allows UE to receive immediate response from AP without having to wait for the periodic beacon frames transmission. The UE basically broadcasts probe frames and any APs within range will respond with a reply.

To simply presume the behavior of a WiFi network based solely on signal strength might prove invalid. This is because there are other important parameters that can affect the UE, and furthermore, degrades the network performance as well as the link quality. In addition, the so-called strongest signal strength (SSS) selection policy can cause high user concentrations in the said AP with the highest RSS. While on the other hand, there might be unpopulated nearby APs that were not selected due to slightly lower RSS i.e. farther distance between UE and the APs. It is of critical note that the throughput of each UE decreases proportional as the number of UEs connected to the same AP increases [1]. Therefore a highly concentrated AP can cause severe ripple effects to the rest of the network components [2, 3].

Theoretically, the network behavior of an AP is described by its critical performance elements and they are summarized as follows:

A. Maximum Theoretical Data Rate

The maximum theoretical data rate represents the upper limit data rates allowable for each of IEEE 802.11 standards, measured in Mbps. This value relies on the hardware of both UE and AP. For example the raw data rates of an IEEE 802.11a/g is up to 54Mbps where as for the IEEE 802.11b, the maximum allowed transmission rates of data frames is only up to 11Mbps.

B. Actual Average Data Rate

This data rate represents the actual data rate on average when a UE is connected to an AP. The parameter is dependent on the network performance, where a poorer network performance leads to a lower data rate of UE. A low actual average data rate can severely affects the channel occupation which will be explained below.

C. Time Interval of Channel Occupation

This parameter indicates the time interval a UE is associated to an AP. Transmitting a low transmission rate frame will take a longer time compared to when transmitting

This study is part of the research funded by Ministry of Science, Technology and Innovation Malaysia (MOSTI)

978-1-4673-5160-7/12/$26.00 ©2012 IEEE

S6-1 2012 IEEE Student Conference on Research and Development

272

the same size frame at a higher transmission rate. Consequently, a lower actual average data rate will increase the time interval of the channel occupation by a UE. This phenomenon can severely affect other higher transmission rate UE if they are all connected to the same AP.

D. AP Traffic Load

This parameter is determined by the concentration level of UEs to specific AP. Imbalanced AP traffic load might occur if the AP accommodates a large number of UEs hence resulting in having a highly congested traffic load. The load balancing in AP has been a crucial research issue since the imbalance load can ultimately degrades the overall network throughput.

E. Channel Interference

Channel interference implies the interference level while a UE communicates with an AP. Sources of interference can vary from due to close proximity of numerous APs as well as interference from other external transmitting sources. In monitoring the network behavior, this parameter should be carefully tackled to avoid severe network degradation.

II. FOCUSES AND WAY-FORWARDS

In this study, initial focuses are directed on the

monitoring of RSS values as well as measuring the actual

internet connection speeds. The information collected at the

UE was acquired by executing bash script in LINUX

environment. The network behavior of this AP, emulating a

public access point was further investigated by scrutinizing

these two critical parameters. The RSS indicates the total

signal power received in decibels referenced to 1 milliwatt

(dBm) [4]. As can be observed in Eq. 1 below, the RSS of

an AP is highly dependent on the distance between the AP

and the UE, path loss and other losses caused by the signal

propagation effects such as reflection, diffraction and

scattering as elaborated in [5]

RSS = r - 10αlog10 (d) - otherLosses (1)

r in the above equation is the initial received signal

strength, d is the distance from AP to UE, α is the path loss

exponent, and otherLosses is the sum of losses introduced

by the signal propagation along the distance d. In this

experimental study, the network behavior of an AP was

demonstrated in such a way where even by having a high

RSS at UE, this still does not guarantee a good internet

connection speed.

The rest of the paper is structured as follows: Section III

provides a summary of the related works. Section IV

explains a brief description of the system model used in data

monitoring and section V highlights the experimental results

gained. The conclusions drawn from this pilot study are

summarized in section V.

III. RELATED WORK

It is critical to monitor the attributes that can potentially

affect the end user performance since each of the

performance metrics can affect the entire network resources

as a whole. Various AP network parameters are highlighted

below by focusing on different parameters of interest.

In [6], the network behavior of an access point is based

on the AP lowest attenuation and the highest Signal-to-

Noise ratio (SNR) while in [7], the researchers draw

attention on the upstream and downstream bandwidth of an

AP. Such scenario is also exercised in [8], whereby the AP

network performance is not based solely on the RSS but the

focus is given on the minimum bandwidth of that particular

AP. Lastly, as investigated in [9], the network behavior is

evaluated based on the traffic of an AP corresponding to the

achievable hypothetical bandwidth. From all these literatures

studied, the network performance parameter lacks on the

internet connection speed and AP network behavior

knowledge which of crucial factors since connectivity and

network reliability based on the historical AP behavior

performance have become very critical to the internet users.

IV. SYSTEM MODEL

In this section, the configuration of the network monitoring setup is described, detailing how the data are being acquired throughout the experimental work. In the subsequent subsection, the methodology used in data monitoring is explained and the network performance parameters that are being considered to be analyzed are listed out.

A. Time Frame

Data measurement reported is acquired over the course of four consecutive days from Monday, May 14 2012 till Thursday, May 17 2012. The time duration of interest starts from 8:30 am to 5:30pm since it reflects a typical working hours in an office.

B. Infrastructure

The data monitoring is carried out using a desktop equipped with a PCI adapter for wireless LAN. The network configurations are presented in Fig. 1. To capture the RSS and the internet connection speed, LINUX operating system (ubuntu 11.04) is installed. The terminal extracts out the beacon frame from the monitored AP of interest. The AP used in this study is an unencrypted public AP. The AP is installed in an office with approximately 15 to 20 users connected to it. Table 1 below lists out the detailed information of this AP.

TABLE I. ACCESS POINT INFORMATION

Attributes Specifications

SSID Guest

Frequency 2.462 GHz

MAC address 00:1b:2a:95:8a:91

Transmit bitrate 54.0 Mbps

C. Data Trace and Analysis To measure the RSS parameter, an iw command line is

ran from the terminal. The beacon frame of the AP is automatically generated and stored in a file that contains the MAC address, SSID, carrier frequency, transmit packet bytes, receive packet bytes, RSS and transmit bit-rate. The parameter of interest i.e. the RSS is then consolidated in Excel software to reveal the signal behavior over time.


273

For the internet connection speed monitoring, a bash script is executed from the terminal. It provides the ping (latency), average download speed and the average upload speed. This bash script functions entirely over HTTP to measure the maximum compatibility of the deployed AP. The ping value (latency) indicates the time it takes to get a response from the selected server when the HTTP request is being sent. In this study, the same server in Kuala Lumpur region is used to indicate the uniformity of the connection speed collected. The download speed is regarded as the downloading of a binary file from the web server to the client, and the download speed is measured to estimate the connection speed. Based on this result, a specific amount of data can be selected for download in a real test. The main purpose is to determine the appropriate amount of data that can be downloaded in 10 seconds, ensuring enough data gained for an accurate result but do not involve long period of time. The upload speed on the other hand is determined when a small amount of random data is generated in the client and sent to the web server to estimate the connection speed. Based on this result, an appropriately sized chunk of randomly generated data is selected for uploading purposes. The upload test data are then performed in chunks of uniform size data to the server-side. They are sorted by speed, and the fastest half is averaged to eliminate anomalies.

Figure 1. Wireless network configuration ©2012 MIMOS Berhad. All

rights reserved

V. EXPERIMENTAL RESULTS

In this section, the plots of RSS and the download speed are presented. The monitored RSS is presented in Fig. 2 while Fig. 3 represents the graph of download speed for Monday, May 14, 2012. From Fig. 2 and Fig. 3, the peak download speed is 17.29 Mbps at 10:00 am with corresponding RSS value of -60 dBm. The minimum achieved download speed is at 12:00pm with only 0.49 Mbps yields at corresponding RSS value of -61 dBm. This indicates that although download speed value is at the lowest level within the time frame of the observation period, the RSS somehow does not give indication of a direct proportionality.

Fig. 4 and Fig. 5 shows the data plot of RSS and

download speed for the following day of data monitoring.

Similar behaviors as previously discussed is observed;

whereby the maximum download speed detected is 21.29

Mbps at RSS value equals to -62 dBm. The minimum

download speed on the other hand is 0.81Mbps at

corresponding RSS value of -63dBm. Thus, the same

conclusion does imply to this pattern of behavior. By visual

inspection, at the peak RSS value of the day; -48 dBm, does

offer download speed value of equals to 14.66 Mbps while

the least RSS value of -65 dBm which in turn provides

download speed of 2.79 Mbps. To analyze this behavior, one

might infer that RSS values are directly proportional to the

download speed. But, this is not always the case when later

in the following figures such deduction are proven incorrect.

Figure 2. RSS over time-of day on Monday, May 14, 2012

Figure 3. Download speed over time-of day on Monday, May 14, 2012

Figure 4. RSS over time-of day on Tuesday, May 15, 2012

0

10

20

30

40

8:3

0

9:0

0

9:3

0

10

:00

10

:30

11

:00

11

:30

12

:00

12

:30

13

:00

13

:30

14

:00

14

:30

15

:00

15

:30

16

:00

16

:30

17

:00

17

:30

Do

wn

load

Sp

ee

d (

Mb

ps)

Time-of-day

Download Speed - Monday

-80 -70

-60 -50

-40 -30

-20 -10

0

8:3

0

9:0

0

9:3

0

10

:00

10

:30

11

:00

11

:30

12

:00

12

:30

13

:00

13

:30

14

:00

14

:30

15

:00

15

:30

16

:00

16

:30

17

:00

17

:30

RSS

(d

Bm

)

Time-of-day

RSS - Tuesday


274

Figure 5. Download speed over time-of day on Tuesday, May 15, 2012

Referring to Fig. 6 and Fig. 7, the peak RSS is at -

59dBm with the download speed of equal to 12.39 Mbps,

while the minimum RSS is -70 dBm with corresponding

value of download speed equals 12.82 Mbps. By having

such behavior, it can be concluded that the previous

assumption of direct proportionality between RSS and

download speed does not imply in most cases.

Figure 6. RSS over time-of day on Wednesday, May 16, 2012

Figure 7. Download speed over time-of day on Wednesday, May 16, 2012

The monitored data of RSS and download speed on the

final day, which was recorded on Thursday, May 17, 2012

are presented in Fig. 8 and Fig. 9. To further reiterate the

newly noticed behavior, the minimum RSS value obtained is

-69 dBm and the corresponding download speed was

recorded at two different time-of-days which were at 2.00

pm and at 5.30 pm at the connection speed value of 13.13

Mbps and 18.24 Mbps respectively. The difference of 5.11

Mbps between these connection speeds while being at the

same RSS level indicates that RSS value does not represent

the same corresponding values of download speed.

Figure 8. RSS over time-of day on Monday, May 17, 2012

Figure 9. Download speed over time-of day on Thursday, May 17, 2012

VI. CONCLUSIONS

In this paper, the preliminary network behavior of a

public access point in a fairly small office environment has

been analyzed and presented. The data monitoring were

carry out on four consecutive days during time between 8.30

am until 5.30 pm. The objectives of this study are to extend

the understanding of the wireless behavior of IEEE 802.11

WLAN from the perspective of Wi-Fi users. It is intended

that the new findings can be applied to manage issues in

wireless network capacity planning. The outcomes can be

exploited and elaborated further as inputs in designing smart

algorithms for an AP load balancer between multiple APs in

a wireless network. The potential gain of such benefits can

likely outperform the classic method that is based only on

the RSS.

There are many important parameters that need to be

observed when analyzing the AP network behavior from the

user’s perspective. In this study, focuses are directed on the

RSS value as well as the internet connection speed which is

monitored using a LINUX terminal. The correlation between

these two parameters are interpreted and further analyzed.

From the graphs, it is evident that representing the AP

behavior solely based on the RSS does not give a correct

indication on the real performance of the AP.

For future work, the network behavior of multiple APs

can be analyzed in order to carry out a comparison study

between different APs to see the diversity in the

performance.

0

10

20

30

40 8

:30

9:0

0

9:3

0

10

:00

10

:30

11

:00

11

:30

12

:00

12

:30

13

:00

13

:30

14

:00

14

:30

15

:00

15

:30

16

:00

16

:30

17

:00

17

:30

Do

wn

load

Sp

ee

d (

Mb

ps)

Time-of-day

Download Speed -Tuesday

-80

-60

-40

-20

0

8:3

0

9:0

0

9:3

0

10

:00

10

:30

11

:00

11

:30

12

:00

12

:30

13

:00

13

:30

14

:00

14

:30

15

:00

15

:30

16

:00

16

:30

17

:00

17

:30

RSS

(d

Bm

)

Time-of-day

RSS - Wednesday

0

10

20

30

40

8:3

0

9:0

0

9:3

0

10

:00

10

:30

11

:00

11

:30

12

:00

12

:30

13

:00

13

:30

14

:00

14

:30

15

:00

15

:30

16

:00

16

:30

17

:00

17

:30

Do

wn

load

Sp

ee

d (

Mb

ps)

Time-of-day

Download Speed - Wednesday

-80

-60

-40

-20

0

8:30 9:30 10:30 11:30 12:30 13:30 14:30 15:30 16:30 17:30

RSS

(d

Bm

)

Time-of-day

RSS - Thursday

0

10

20

30

40

8:3

0

9:0

0

9:3

0

10

:00

10

:30

11

:00

11

:30

12

:00

12

:30

13

:00

13

:30

14

:00

14

:30

15

:00

15

:30

16

:00

16

:30

17

:00

17

:30

Do

wn

load

Sp

ee

d (

Mb

ps)

Time-of-day

Download Speed - Thursday


275

ACKNOWLEDGMENT

The studies are part of deliverables to Ministry of Science, Technology and Innovation of Malaysia. The reported research findings are part of the main research funded under Science fund. It is a collaboration project under supervision of MIMOS Berhad incorporating researchers from International Islamic University (IIUM).

REFERENCES

[1] Y. Fukuda, T. Abe, and Y. Oie, “Decentralized Access Point for Wireless LANs”, In Proceedings of Wireless Telecommunications Symposium, 2004.

[2] A. Balachandran, P. Bahl, and G. Voelker, “Hot-spot Congestion Relief and Service Guarantees in Public-Area Wireless Networks”, In Proceedings of Workshop on Mobile Computing Systems and Applications, 2002.

[3] Y. Bejerano, S.-J. Han, and L. Li, “Fairness and Load Balancing in Wireless LANs Using Association Control”, In Proceedings of ACM Mobicom, 2004, pp. 315–329.

[4] M. Sauter From GSM to LTE: An Introduction to Mobile Networks and Mobile Broadband. John Wiley and Sons, 2010.

[5] E. C. L. Chan, G. Baciu, Introduction to Wireless Localization:With Iphone SDK Examples. John Wiley and Sons, 2012.

[6] M. Abusubaih, J. Gross, S. Wiethoelter and A.Wolisz, “On Access Point Selection in IEEE802.11 Wireless Local Area Networks”, In Proc. Of 6th IEEE International Workshop on Wireless Local Networks, WLN'06, Tampa, FL, USA, November, 2006.

[7] S. Vasudevan, K. Papagiannaki, C. Diot, J. Kurose, and D. Towsley, “Facilitating Access Point Selection in IEEE 802.11 Wireless Networks”, In ACM Sigcomm IMC, Berkeley, October 2005.

[8] R. Akl and S. Park, “Optimal Access Point Selection and Traffic Allocation in IEEE 802.11 Networks,” Proceedings of 9th World Multiconference on Systemics, Cybernetics and Informatics (WMSCI 2005): Communication and Network Systems, Technologies and Applications, paper no. S464ID, July 2005.

[9] V. Ghini, S. Ferretti and F. Panzieri, “A Strategy for Best Access Point Selection”, In Proc. Of 3rd IEEE/IFIP Wireless Days (WD) 2010 Conference, October 2010.


276

[ieee 2012 ieee student conference on research and development (scored) - pulau pinang, malaysia...

Documents