HRD Programme for Exchange of ICT Researchers/Engineers Through Collaborative Research

FINAL TECHNICAL REPORT

On

Implementing an Experimental Cognitive Radio System for IEEE 802.11 b/g

Submitted To Asia-Pacific Telecommunity Submitted By Mr. Imran Khan Mr. Shujaat Ali Khan Dr. Nandana Rajatheva Dr. Teerawat Issariyakul (Project Coordinator)

September 2008

Acknowledgement We wish to express our deep appreciation and sincere gratitude to the Asia-Pacific Telecommunity (APT) for showing their moral and financial support for the realization of this project.

We would also like to thank the Asian Institute of Technology (AIT), for facilitating us with the exclusive research environment towards the fulfillment of our objectives.

Project Members Mr. Imran Khan Mr. Shujaat Ali Khan Dr. Nandana Rajatheva(Advisor) Dr. Teerawat Issariyakul (Project Coordinator)

ii

Abstract

This report gives an overview of the cognitive radio technique. We are using the cognitive concept to enhance the spectrum efficiency of telecommunications network, deployed in very busy and/or dense areas e.g., airports, hotels, railway stations, student campuses etc. The main idea is to opportunistically identify the free spectrum resources and use these spectrum holes to provide services to the cognitive and/or secondary users.

This report gives the insight into the work progress so far. It also mentions some of the techniques and methodologies to be used for the implementation of APT project. Our model is based on actual measurements in the 2.4GHz ISM band using a signal analyzer to collect complex baseband data. We look at the energy-detection sensing strategy to identify the user behavior by sensing the spectrum. This project aims at attain realistic parameters under various scenarios such as universities, offices, airports, etc.

iii

Table of Contents

Chapter Title Page

Title Page i

Acknowledgement ii

Abstract iii

Table of Contents iv

List of Figures v

Chapter 1 Introduction 1

1.1 Overview 1

1.2 Objective of the study 3

Chapter 2 Literature Review 4

2.1 Types of Cognitive Radio 4

2.2 Functions of Cognitive Radio 5

2.3 Sensing Strategies 6

2.3.1 Energy-based detection 6

2.3.2 Feature-based detection 7

2.4 Cognitive Radio and IEEE 802.11b/g service PHY standard 8

2.5 Dynamic Spectrum Access in WLAN Channels: 9

2.6 A Real Time Cognitive Radio Testbed for Physical and link layer Experiments: 9

Chapter 3 Methodology 11

3.1 Introduction 11

3.1.1 Energy-based detection 11

3.2 Experimental setup 12

3.3 Schedule 13

Chapter 4 Experimental Results 14

4.1 Location 14

4.2 RF Signal Quality 14

4.3 User Behavior 15

APPENDEX A 17

APPENDEX B 18

References 19

iv

List of Figures

Figure Title Page

1.1 Current FCC spectrum allocation 1

1.2 Spectrum use of 0 2GHz frequencies over 10 minutes 2

2.1 Spectrum Utilization Measurement at BWRC 4

2.2 Basic Cognitive Cycle 5

2.3 Receiver chain for the feature-based detection scheme. 7

2.4 Detection of a continuous-phase 4-FSK using energy detection and cyclostationary feature detection. 7

2.5 Improvement in SINR between CR and standard technique. 8

2.6 Conceptual diagram of experiment. 9

2.7 Baseband signal of a WLAN supporting a VoIP conference session. 9

3.1 PSD of the received signal, showing threshold power level 11

3.2 Topology of the AirMagnet Enterprise system 13

3.3 Proposed timeline of the APT project 13 .

v

Chapter 1

Introduction

1.1 Overview There is a rapid advancement in wireless communication technology providing the

network services anywhere and anytime. The extensive success of cellular systems has

put forward a challenge of developing new wireless systems and standards to maximize

the coverage and capacity of the networks. Hence, new systems and standards are being

implemented to provide high speed data and voice communication services by effectively

utilizing the available limited radio resources. As radio spectrum is the most important

resource that need to be efficiently utilized. The cognitive radio technology has been

proposed to fulfill this requirement of efficient radio spectrum utilization.

The Federal Communication Commissions (FCC) frequency allocation chart indicates

multiple allocations over all of the frequency bands in figure 1 [4]. Thus, within the

current regulatory framework, spectrum is a scarce resource, at least at the frequencies

below 3 GHz, which are particularly valuable due to their favorable propagation

characteristics. While the actual measurements that is shown in figure 2 has been taken at

0-2.5GHz frequencies over a period of 10 minutes, showing the low utilization of radio

spectrum [5].

Fig. 1.1 Current FCC spectrum allocation

1

Fig. 1.2 Spectrum use of 0 2GHz frequencies over 10 minutes

The Cognitive radio technology intends fulfill this requirement of efficient radio spectrum utilization. The idea of cognitive radio gives us the radio that senses and reacts to its operating environment. Cognitive radio is very helpful in improving the use of of spectrum resources and also to reduce the engineering and planning time. Frequency of operation, transmitter power and modulation are just a few of the operating parameters that can automatically be adjusted in a cognitive radio system.

The cognitive radio terminology was invented by Mitola [7] and refers to a smart radio which has the ability to sense the external environment, learn from the history and make intelligent decisions to adjust its transmission parameters according to the current state of the environment. Cognitive radios have recently received much attention for two reasons: flexibility and potential gains in spectral efficiency. They can rapidly upgrade, change their transmission protocols and schemes, listen to the spectrum as well as quickly adapt to different spectrum policies.

Cognitive Radios are able to sense the spectrum to see whether it is being used by the primary user. However this sensing operation may be made difficult due to a degraded wireless channel, which has prompted concerns from primary users of the spectrum.

In cognitive networks, there is a need for better interaction between the different layers of the protocol stack to attain the end-to-end goals and performance in terms of resource management, security, QoS or other network goals. Cross-layer design [5] refers to protocol design done by actively exploiting the dependence between the protocol layers to obtain performance gains. Cognitive networks will employ crosslayer design and

2

optimization techniques in order to adapt, simply because a great level of coordination is needed between the traditional protocol layers.

IEEE 802.11b installations are the most widely used in Europe. Also, IEEE 802.11a systems are in operation. If IEEE.11a is to be implemented into an SDR, while the modulation mode should be OFDM. There are major efforts towards the development of joint UMTS/WLAN systems which use the SDR approach.

The Cognitive Radios may harmfully interfere due to the following reasons:

The Cognitive Radio may not be able to reliably detect a Primary User and as a result start sending even though Primary User is using the frequency.

A Cognitive Radio that is using a frequency that was supposed to be free by sensing process may not be able to reliably detect that a primary user has reappeared and therefore may not leave the frequency.

Even if a Cognitive Radio has detected the Primary User, it may fail to relinquish the frequency quickly enough and therefore continue to send creating harmful interference to the Primary User transmission.

1.2 Objective of the study To investigate the behavior of IEEE 802.11 b/g users in various environment and

by using the results taken from these field based experiments.

To formulate a generalized traffic model based on the collected statistics. To investigate the reliability of sensing the radio spectrum by using energy

detection technique and from these measurements, set the operating parameters of

the radio.

To investigate the spectrum efficiency 802.11b/g WLAN system to show the benefits of cognitive implementation.

3

Chapter 2

Literature Review

The basic idea behind cognitive radio is the utilization of unused frequency bands of a primary or a licensed user by a secondary or an unlicensed user without interfering the primary users communication and providing the required QoS for the secondary users (SUs). If we scan the portions of radio spectrum we would find that some frequency bands are largely unoccupied most of the time while some other frequency bands are only partially occupied and the remaining frequency bands are heavily used. The measurements taken at Berkeley show this uneven usage of spectrum between 0 to 6 GHz as illustrated in figure 2.1. From figure 2.1 we can see that the spectrum utilization is heavy below 3 GHz. Actual measurements taken shows that the utilization in 3-4 GHz frequency band is 0.5% while it drops to 0.3% in 4-5 GHz band [13]. Thus, we can say that spectrum is actually not scarce but it is not utilized efficiently. Cognitive radios are proposed to sense this uneven usage of the spectrum and identify a vacant frequency band and automatically decide to access this vacant spectrum.

Fig. 2.1 Spectrum Utilization Measurement at BWRC

2.1 Types of Cognitive Radio

There are two types of cognitive radio depending upon the bandwidth utilization which are as follows [14]

Static bandwidth cognitive radio

4

Dynamic bandwidth cognitive radio

In static bandwidth cognitive radio, all devices use fixed spectral bandwidth to transmit their data using the spectrum holes by opportunistically hopping on those holes when there is no primary user. Examples of such systems are IEEE 802.11 WLANs, IEEE 802.153 bluetooth and IEEE 802.16e WiMAX. In dynamic bandwidth cognitive radio system the system is allowed to expand or contract its spectral bandwidth at any time instant and changes its transmission waveform accordingly. For example this can be achieved by switching the carriers ON and OFF according to the availability of spectrum in Orthogonal Frequency Division Multiplexing (OFDM) modulation or in Multi-Carrier Code Division Multiple Access (MC-CDMA).

2.2 Functions of Cognitive Radio

A cognitive cycle comprising of the major tasks performed by a cognitive radio is as depicted in figure 2.3. The main functions of cognitive radio as shown in the figure are described as follows

Fig. 2.2 Basic Cognitive Cycle

The major function of cognitive radio is spectrum sensing which scans and detects unused spectrum for their use without having any harmful interference with other user. Cognitive radio is designed to be aware of and sensitive to the changes in the surrounding. Now the practical design problem is to reliably detect the primary user on processing the wide bandwidth (multi-gigahertz). In order to make the sensing function more effective, three digital signal processing techniques has been inspected [2]:

Matched filter detection

Energy detection

5

Cyclostationary feature detection

2.3 Sensing Strategies

There are two different strategies that can be used

2.3.1 Energy-based detection

If the transmission standard of the primary user is unknown, then the decision is made on the basis of energy transmitted. The detection can then be formulated by using binary hypothesis test with [11]

(2.1) 01

: , 1,..., : , 1,..., .

i i

i i i

H Y V i NH Y S V i N

= = = + =

where

20(0, )iV C N , complex Gaussian distribution with zero mean and variance 20 .

21(0, )iS C N is complex Gaussian distribution with zero-mean but variance 21 .

The optimal Neyman-Pearson detector is given by

2

1( )

N

ii

T y Y =

>=

2.3.2 Feature-based detection

There are some applications to assume that we know that there is a sharing of the spectrum with a WLAN. In this case, we can develop this knowledge to detect packets more reliably and to extract information on the packet length that is provided in the packets header. The decoding chain used to decode the WLAN packets is shown in Fig. 2.3 [11].

Fig. 2.3 Receiver chain for the feature-based detection scheme.

The analysis shown in [2] is that cyclostationary feature detection is more effective, as it can distinguish modulated signals, interference and noise in low SNR. In cyclostationary feature detection spectral correlation function is used, which is based on two dimensional transform. The Fig. 2.4 shows the difference between energy detection and cyclostationary detection [2].

Fig. 2.4 Detection of a continuous-phase 4-FSK using energy detection and

cyclostationary feature detection.

7

2.4 Cognitive Radio and IEEE 802.11b/g service PHY standard

An experiment through software simulation shows a 20 dB SINR improvement using cognitive techniques in an interference environment over that provided by current IEEE 802.11b/g service PHY standard [9] as shown in Fig. 2.5 [8].

Fig. 2.5 Improvement in SINR between CR and standard technique.

The simulation system is developed using Matlab. The established network consists of multiple access points (APs) with point-to-multipoint link configuration. The conceptual diagram used for the experiment is shown in Fig. 2.6 [8]. Two key performances are greatly improved.

the link QoS due to optimized channel selection.

the utilization of the spectrum due to interference avoidance at each cognitive network node.

8

Fig. 2.6 Conceptual diagram of experiment.

2.5 Dynamic Spectrum Access in WLAN Channels:

Dynamic spectrum access in the time-domain relies on the existence of sufficient whitespace between bursty transmissions. This case is near to practical situation. The baseband signal recorded by a signal analyzer for a Voice-over-IP conference session over WLAN shown in Fig. 2.7 [10].

Fig. 2.7 Baseband signal of a WLAN supporting a VoIP conference session.

In [10] dynamically sharing of the spectrum in the time-domain has been considered by using whitespace between the transmissions of a primary user, represented by an 802.11b-based wireless LAN (WLAN). This model is based on measurements taken at 2.4GHz using a signal analyzer.

2.6 A Real Time Cognitive Radio Testbed for Physical and link layer Experiments:

Since the idea of cognitive radio implies in utilizing the licensed band of frequency by the unlicensed (secondary) users or the cognitive radio users when the licensed (primary)

9

users are not using it, it becomes necessary to detect whether the particular band of frequency is currently being used by licensed users or not. There are various techniques applied for the detection of primary users. Now there is a need to develop some test to prove the reliable detection primary user by a cognitive radio and to vacate the frequency band when it is reclaimed by primary user.

Experiment has been done using sensing algorithms and schemes, to get the results for indoor cognitive radio network at the Physical and Network layers [1]. Here Berkeley Emulation Engine 2 (BEE2) has been used to get a set of tests to measure the performance of various sensing techniques.

10

Chapter 3

Methodology

3.1 Introduction Our model is based on actual measurements in the 2.4GHz ISM band using a signal analyzer to collect complex baseband data. We look at the energy-detection sensing strategy to identify the user behavior by sensing the spectrum. The data that will be collected from this experiment will be used to describe the idle and busy periods of the channel.

3.1.1 Energy-based detection

For detecting the primary user, we are using the energy detection technique by using the spectrum analyzer. We are using the power spectral density (PSD) to identify the presence of an active primary user. We are using a power threshold level for this purpose. If the PSD value is greater more than the defined threshold, then we consider the presence of a primary user (Fig. 3.1).

PSD

Freuency

PTH

Primary user detected

Fig. 3.1 PSD of the received signal, showing threshold power level

Let, the received signal is

( ) ( ) ( ) ( )y t x t h t t= + (3.1)

where,

( )x t is the primary user transmitted data.

( )y t represents the received data at AP,

is channel response and ( )h t

11

( )t is additive white Gaussian noise

A simple decision strategy is applied for sensing the primary user by using binary hypothesis test with

(3.2) 01

: ( ) ( ) (signal absent): ( ) ( ) ( ) ( ) (signal present)

H y t tH y t h t x t t

= = +

Now let is the power spectrum density (PSD) of the received signal and is threshold power level (this value can be taken from a testbed or real time implementations). Now detection of primary user by using hypothesis will become

( )yG f THP

(3.3) 01

: ( ) (signal absent)

: ( ) (signal present)y TH

y TH

H G f PH G f P

>

The frequency agility of cognitive radio is limited by the number of samples taken. By taking large number of samples can miss the use of short duration gaps in spectrum band [3].

3.2 Experimental setup In order to capture the transmissions of the WLAN discussed above we use an AirMagnet Enterprise 8.0 [15]. The AirMagnet Spectrum Analyzer Sensor integrates AirMagnets advanced spectrum-sensing hardware and analytical and visual display software into one application. AirMagnet Enterprise system will be used to monitor and collect spectrum data to show the user behavior. The entire AirMagnet Enterprise system consists of three major components as shown in Figure 3.2 below:

AirMagnet Enterprise Server AirMagnet Enterprise Console, and AirMagnet SmartEdge Sensor

12

Fig. 3.2 Topology of the AirMagnet Enterprise system

3.3 Schedule

The proposed timeline for the project is shown in Fig. 3.3

Fig. 3.3 Proposed timeline of the APT project

13

Chapter 4

Experimental Results

4.1 Location We analyze one-week trace of a local area wireless network installed in Telecommunications (TC) building at Asian Institute of Technology, Thailand. There are two access points at TC building, using channels 1 and 6. It is two stories building, with one AP at each floor. Its ground floor has labs, while first floor has two class rooms, a lounge, coffee shop and faculty offices. The access point we analyzed using channel 1, placed at senior lab. We find that users are divided into distinct location-based sub-communities, each with its own movement, activity, and usage characteristics. Most users exploit the network for web-surfing, session-oriented activities and chat oriented activities. The high numbers of web-surfing oriented activities are observed. These results will help in presenting the behavior of user for this particular local area wireless network and user community. We believe that similar environments may exhibit similar behavior and trends.

The overall user behavior, network traffic, load characteristics and traffic characteristics from the user point of view is analyzed.

4.2 RF Signal Quality This report contains data on the overall RF signal quality of the 802.11 network in terms of signal strength, noise level, and signal-to-noise ratio. It is very critical to note that there is sufficient coverage for all the devices and the RF environment has the minimum amount of noise possible. WLAN reliability and efficiency depend on the quality of the RF media. Be it 802.11b/g at 2.4GHz or 802.11a at the 5GHz RF spectrum, they are all susceptible to RF noise impact. A cordless phone, Bluetooth devices, microwave, wireless surveillance video camera, or baby monitor can all emit RF energy to disrupt WLAN service. Malicious attacks can manipulate the RF power at 2.4GHz or 5GHz spectrum with a high gain directional antenna to amplify the attack impact from a distance. Excessive noise causes WLAN devices in the target area to be out of wireless service.

14

4.3 User Behavior For the data sent over this period of one week, we record:

The users identity The users location (current AP) The source and destination ports The application in use The size of packet

Figure 4.1 shows the power density function of inter-arrival time of the users.

15

PDF of Inter-arrival Time

00.050.1

0.150.2

0.250.3

0 ---- 100 100 ----200

200 ----300

300 ----400

400 ----500

500 ----600

Inter-arrival Time (msec.)

f(x)

f(x)

Fig. 4.1 Inter-arrival Time

16

APPENDEX A

CHANNEL AP SIGNAL CROSS CHANNEL INTERFERENCE

NOISE SIGNAL/NOISE

1 -89 -89 -95 60

2 -100 -87 -95 80

3 -100 -86 -95 90

4 -100 -100 -95 00

5 -100 -73 -95 220

6 -73 -73 -95 220

7 -100 -74 -95 210

8 -100 -100 -95 00

9 -100 -90 -95 50

10 -100 -100 -95 00

11 -85 -87 -95 100

12 -100 -87 -95 80

13 -100 -100 -95 00

14 -100 -100 -95 00

17

APPENDEX B

INTERVALS X-OCCURENCES F(X) 0 ---- 100 292 0.148676171

100 ---- 200 273 0.139002037 200 ---- 300 526 0.267820774 300 ---- 400 431 0.219450102 400 ---- 500 345 0.175661914 500 ---- 600 97 0.049389002

total 1964 1

18

References

[1] S. M. Mishra, D. Cabric and R. W. Brodersen, A Real Time Cognitive Radio Testbed for Physical and Link Layer Experiments, IEEE international symposium on New Frontiers in Dynamic Spectrum Access Networks, 2005. [2] D. Cabric, S. M. Mishra, and R. W. Brodersen, Implementation issues in spectrum sensing for cognitive radios, in Asilomar Conference on Signals, Systems, and Computers, 2004. [3] A. Sahai, N. Hoven, R. Tandra, Some Fundamental Limits on Cognitive Radio, Proc. of Allerton Conference, Monticello, Oct 2004. [4] NTIA, U.S. frequency allocations. Available: www.ntia.doc.gov/osmhome/allochrt.pdf [5] R. W. Broderson, A. Wolisz, D. Cabric, S. M. Mishra, and D. Willkomm, White paper: CORVUS: A Cognitive Radio Approach for Usage of Virtual Unlicensed Spectrum. Technical report, 2004. [6] Kawadia, V. and Kumar, P.R. (2005) A cautionary perspective on cross-layer design. IEEE Wireless Communications, 12(1), 311. [7] Mitola III, J. (2000) Cognitive Radio: An Integrated Agent Architecture for Software Defined Radio, PhD thesis, Royal Institute of Technology, Sweden. [8] Maldonado D., Le B., Hugine A., Rondeau T.W., Bostian C.W., Cognitive Radio Applications to Dynamic Spectrum Allocation, IEEE, p 596-600, 2005. [9] IEEE Std 802.11g/D1.1-2001, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) speei5cations: specifications: Further Higher-Speed Physical Layer Extension in the 2.4 GHz Band. [10] S. Geirhofer, L. Tong, and B. M. Sadler, A measurement-based model for dynamic spectrum access in WLAN channels, in Proc. Military Communi. Conf. (MILCOM), Washington, DC, Oct. 2006. [11] H. V Poor, An Introduction to Signal Detection and Estimation, 2nd ed. Springer-Verlag, 1994. [12] C. Chang, J. Wawrzynek, and R. W. Brodersen, Bee2: A high-end reconfigurable computing system, IEEE Design and Test of Computers, vol. 22, no. 2, pp. 114125, 2005. [13] Cabric D., Mishra S.M., Willkomm D., Brodersen R. and Wolisz A., A Cognitive radio approach for usage of virtual unlicensed spectrum, in: Proc. 14th IST Mobile and Wireless Communications Summit, June,2005 www.bwrc.eecs.berkeley.edu/MCMA

19

[14] Nandagopalan S. (2007), Squeezing the Most Out of Cognitive Radio: A Joint MAC/PHY Perspective, Acoustics, Speech and Signal Processing, 2007. ICASSP IEEE International Conference, Volume 4, 15-20 April, 2007. [15] Available at http://www.airmagnet.com/products/enterprise/ cts/enterprise/

20

IntroductionOverviewObjective of the study

Literature ReviewTypes of Cognitive RadioFunctions of Cognitive RadioSensing StrategiesEnergy-based detectionFeature-based detection

Cognitive Radio and IEEE 802.11b/g service PHY standardDynamic Spectrum Access in WLAN Channels:A Real Time Cognitive Radio Testbed for Physical and link la

MethodologyIntroductionEnergy-based detection

Experimental setupSchedule

Experimental ResultsLocationRF Signal QualityUser Behavior

APPENDEX AAPPENDEX BReferences