Methodology of a System Level Design for a Cognitive Radio Receiver Application for IEEE 802.

22 Standard
Rami Ben Amira, Dorra Ayadi, Ins Kammoun and Mourad Loulou
Electronic and Communication Group, LETI-Laboratory, National School of Engineers Sfax, Tunisia E-mail: ramy.benamira@ieee.org Abstract This paper exhibits a methodology to achieve a system level design for Cognitive Radio (CR) receiver supporting IEEE 802.22 Standard. Based on superheterodyne receiver supporting Sub-Band Sensing, the overall radio system characteristics depicted from the IEEE 802.22 Standard specifications are dispatched through the different receiver blocks. Moreover, we developed the theoretical calculation and different techniques of system level design. Each of Radio Frequency (RF) characteristics such as noise figure (NF), gain and linearity (third order intercept point, (IIP3)) is simulated taking in account the analog-to-digital converter (ADC) full scale power constraint. On the one hand, our system level simulation results proved that the receiver architecture is capable to perform spectrum sensing and to allow communication processing. The receiver performs a total gain of 51 dB and an IIP3 of -19.8 dBm for low gain mode. It achieves 105dB as gain and 9.9 dB for noise figure for high gain mode. Index terms Receiver, Cognitive Radio, IEEE 802.22, Spectrum sensing, system level specification, budget analysis, linearity, noise, gain.

In this paper, we will focus on the methodology of a system level design for CR receiver supporting IEEE 802.22 standard. In section II, we will present an overview of CR, its services and we will introduce the first world-wide wireless Standard based on Cognitive radio which is IEEE 802.22. Moreover, a special interest will be considered to the sensing functionality offered by this standard. In section III, receiver characteristics are mentioned and computed. Then, in section IV, budget analysis and simulation results will be exhibited and analysed. At the end, a conclusion of our research will be given. II. IEEE 802.22 RECEIVER SPECIFICATION IEEE 802.22 is emerging as a new wireless air interface standard based on Cognitive Radio known as a Wireless Regional Area Network (WRAN) [4]. The IEEE 802.22 network is based on a Point to Multi Point (P-MP) interconnection. Indeed, it is composed of a Base Station (BS) and a number of CPEs. Each of them performs spectrum sensing. But the BS supervises when sensing is required, analyses all its results and then decides the availability of each television channel. This kind of decision is based on spectrum sensing results and also on auxiliary information shared between BSs. There are three various sensing thresholds which are: first, -116 dBm over a 6 MHz channel to sense DTV Primary Users (PU), Second, -94 dBm assessed for analog TV systems and Finally, -107 dBm calculated in a 200 KHz bandwidth to detect Wireless Microphones Users. IEEE 802.22 targeted first the commercial deployment of CRs in US regions which is reflected on the frequency operational range 54-862 MHz that belongs to the North American regions [4]. After that, to accommodate supplementary international regulatory requirements, this bandwidth could be extended to 41-910 MHz. In addition, different international TV channel with bandwidths of 6, 7, and 8 MHz have to be considered by 802.22 standard specifications [4]. Moreover, the IEEE 802.22 PHY shall provide high modulation and coding schemes flexibility. In this case, BS shall adjust dynamically bandwidth, modulation and coding schemes. The various WRAN base station and CPE receiver performances are specified in the IEEE 802.22 standard Draft; Indeed, in the downstream direction, equivalent receiver sensitivity is around -92 dBm as a minimum value [5]. Meanwhile, the other parameters are

I.

INTRODUCTION

Despite the fixed spectrum assignment policy that characterizes current Wireless networks, the biggest part of the available spectrum is not well exploited and even that 15% to 85% of the spectrum availability is under used [1]. This problem leads to the appearance of a new concept which is Cognitive radio (CR) as a promising solution allowing Opportunistic spectrum utilization that offers the possibility for Secondary users to take advantage of these spectrum holes while Primary Users are not communicating; This definition reflects one of the important Cognitive radio features which is Dynamic Spectrum Access (DSA) [2]. To identify these spectrum opportunities, Consumer Premise Equipments (CPEs) have to sense continuously their environment to inform the associated Base Station about these vacant spectral segments. This procedure represents one of the fundamental aspects of CR which is Spectrum Sensing. For our purpose, a CR receiver has to take into account this functionality. Thats why; one of the first tasks to do is system level specification since it aims to set, for each receiver block, its particular specifications. In order to fulfil this requirement, numerous standard requirements have to be considered such as sensitivity, noise figure (NF) and dynamic range (DR).

978-1-4244-5091-6/09/$25.00 2009 Crown

319

assessed such that the operational frequency band is 54862 MHz using a channel bandwidth of 6 MHz for communication mode and 50 MHz while sensing is performed and a QPSK modulation with rate of . Moreover, the noise figure is specified to be less than 10 dB with coupling and filter loss of around 2 dB. The required Eb/No is fixed to 6.07 dB which correspond to a BER less than 4.10-2 [5]. The maximal input signal, that could be detected and decoded, is also set to -41.8 dBm [6]. As specified, only QPSK modulation will be considered in this work. So, future study will address the impact of other modulation schemes, such as 16 QAM, 64 QAM and 256 QAM, on receiver performances. Therefore, adaptive modulation concept could be considered as a promising research area for CRs. III. CP1dB AND IIP3 COMPUTATION Due to instantaneous signal path fluctuations such as shadowing and path loss, the amplitude of transmitted and received signals could reveal a various range of discrepancy. Indeed, the noise figure performance fixes the lowest level of the dynamic range (DR) of any receiver; On the other hand, the highest signal level is limited by its linearity performance. To maintain constant signal strength at the input of the ADC, the Voltage Gain Amplifier (VGA) is used allowing an efficient sampling operation. In addition, among other RF blocks that may cover large DR, the LNA (Low Noise Amplifier) takes place as one of the first stages that the signal passes through. In order to reduce the large input signal levels, the gain of VGA have to be minimized to prevent any saturation effects that could affect the linearity requirements. To fulfill this demand, two switched gain modes have to be considered to provide the needed amplification that depends on the signal weakness or strength. A. Low Gain Mode (LGM) Owing to the signal robustness with sufficient power, in this low gain mode, a significant amplification is useless. Considering the fact that IEEE 802.22 standard sets OFDM as the modulation scheme retained, an input back-off (IBOmax) [7] of 12dB is assessed for this mode as the difference between the maximum input signal and the input- 1dB-compression point (CP1dB) [9]. Indeed, an OFDM signal has a varying envelope that is defined by the Peak-to-Average Power Ratio (PAPR) [2] which is associated to the IBO definition. Taking into account that the maximum input signal Pin,max is -41.8 dBm, CP1dB could be computed as follows: (1) CP 1dB _ min = P in _ max + IBO max = 29.8 dBm By virtue of the thumbs rule, the IIP3 is greater than CP1dB by 9~10 dB. So, the required IIP3 for IEEE 802.22 receiver in this mode is about -19.8 dBm with IIP2 of 32.4 dBm: (2) IIP 3 min = CP 1dB _ min + 10 = 19.8 dBm

B. High Gain Mode (HGM) Contrary to the low mode, in the high gain mode, an important amplification is needed to overcome the signal weakness. In fact, the minimum signal level Pin,min, specified in IEEE 802.22 standard while communicating, is about -92 dBm and the minimal back-off (IBOmin) from CP1dB is assessed to 8 dB [8]. Thus, the input-1dBcompression point is calculated as follows [9]: (3) CP 1dB _ max = P in _ max + IBO min = 33.8 dBm Then, the IIP3 is measured as:
IIP 3 max = CP 1dB _ max + 10 = 23.8 dBm

(4)

From these computations, it is obvious that linearity requirements are related to the gain mode and consequently to the signal level detected at the input of the RF receiver. For our case, to be able to decode accurately a signal with maximal power of -41.8 dBm (i.e. in low gain mode), the linearity of the receiver should be at least of -19.8 dBm. However, it becomes harder at 23.8 dBm for high gain mode. Therefore, we can affirm that linearity is more constraining in low gain mode and noise figure limits the receiver performance in high gain mode. IV. BUDGET SIMULATION A. Functional description of the CR receiver for IEEE 802.22 standard

Figure 1 exhibits the studied receiver architecture for the IEEE 802.22 specification. This receiver was simulated with ADS tool (Advanced Design System). It represents a superheterodyne receiver that has two Intermediate Frequency (IF) stages [6]. After being detected, the signal passes through RF filter that selects the useful bandwidth from 54 MHz to 862 MHz, after that the LNA helps to improve the SNR of the RF signal that will be mixed up to 1000 MHz using the first Local Oscillator (LO1). Its frequency range goes from 1054 MHz to 1862 MHz. In this case, the image frequency, which ranges from 2054 MHz to 2862 MHz, is not a constraining factor since it will be suppressed by IF1 filter without using any image frequency filter. While passing through the second transposition stage, the second LO2 mixes the IF1 signal at 1000 MHz with 970 MHz that leads to down convert it to a second IF2 at 30 MHz. Therefore, the second image frequency, which is around 940 MHz, will be suppressed by the second IF2 filter. Then, the VGA helps to maintain constant signal strength at the input of the ADC that requires 9 bits effective resolution [6]. Its full scale power is set to 13 dBm which represents an adequate ADC power level [10]. The overall DR at the receiver input is around 50 dB while processing communication signals. This wide band cognitive radio receiver supports sub-band sensing. In fact, the RF front end part is shared between sensing and communication processing. The difference is located at the base band part of this receiver where the digital signal is processed separately depending on the processing type. In addition, the sub-band spectrum

320

sensing technique allows dividing the whole bandwidth to different sub-band bandwidth of 50 MHz containing either eight sub-channels with a bandwidth of 6MHz or six sub-channels with a bandwidth of 8MHz [6]. Each sub-band is digitized separately with a sampling rate of about 100 Msps; The ADC effective bit number is set to only 4 bits with a dynamic range of 24 dB in order to detect the primary users. These characteristics are assessed depending on the sensing thresholds defined in the IEEE 802.22 requirements. For communication signals processing, the ADC digitizes each sub-channel alone with a bandwidth of 6, 7 or 8MHz. In this case, the ADC sampling rate needed is about 12 Msps and having a dynamic range of 54 dB. So, in this configuration, ADC would require 9 bits effective resolution and its SNRin_ADC is settled to 4.02 dB.

(a)

(b)
Figure1. IEEE 802.22 Receiver supporting sub-band sensing Figure 2. Evolutions of input third order intercept point TOI (a) and gain (b) among different receiver blocks for low gain mode

Each block characteristics, such as gain, NF, CP1dB, IIP3 and output power, are obtained by the RF system budget analysis while making the Budget simulations. All these parameters are correlated together referring to FRIIS and IIP3 equations [3]. The distribution of all the IEEE 802.22 requirements will be made among the different receiver blocks that are defined in figures 2 and 3. These blocks are listed as follows: the first block related to 0 is the band pass RF filter, block 1 is the LNA, block 2 represents the first mixer, blocks 3 and 4 are respectively the IF1 gain amplifier and its associated IF1 band pass filter, the second IF stage incorporates blocks 5, 6 and 7 which are respectively the second mixer, IF2 gain amplifier and IF2 band pass filter. The block 8 represents the VGA as the last block of the RF part of the receiver. The different budget simulations will be exhibited for the two gain modes in the next parts of this section. B. Budget simulation for low gain mode For this mode, linearity is the constraining factor since strong signals are considered here. Thats why, noise figure will not be considered. The budget simulations should reveal a total IIP3 around -19.8 dBm. Therefore, the VGA gain is at its minimum value. The Simulation results of IIP3 and gain evolution are presented in (figure (2-a)) and (figure (2-b)) respectively. Figure 2 proves that the linearity requirement is reached. In fact, at the input of the receiver, the IIP3 (represented by TOI in (figure (2-a))) is -19.74 dBm with a total gain around 51 dB.

C.

Budget simulation for high gain mode

In this case, linearity is not the most constraining issue since the input signal is weak. Only the VGA gain will be increased to meet the noise requirements without altering the blocks values already fixed. In addition to IIP3, gain and NF simulations, we will take care of the maximal output power (Pout_max) delivered to the ADC input to meet the full scale power requirement. Figure 3 reveals respectively the progress of IIP3 (figure (3-a)), Gain (figure (3-b), NF (figure (3-c)) and Pout_max (figure (3-d)). At the input of the receiver, the value of IIP3 is about -19.74dB which fits clearly the specifications as represented on figure (3-a). Figure (3-b) proves that the receiver gain reached 105dB at the input of the ADC. Figure (3-c) demonstrates that the total NF is around 9.9dB. This figure shows also a difference of 2dB at the first bloc. In fact, it represents the coupling and filter losses, which are specified by the IEEE 802.22 requirements [5], and are taken into account when simulating the ADS receiver model, as the RF filter insertion losses. Moreover, figure (3-d) demonstrates that Pout_max is around 13dBm which is equal to the full scale power requirement at the ADC input.

321

All the receiver characteristics are recapitulated in table I. Finally, the Cognitive radio requirements, which are fixed at the beginning of this work in terms of IIP3, NF and Pfs referring to IEEE 802.22 specifications, are satisfied. V. CONCLUSION As a conclusion, a methodology of a system level design supporting IEEE 802.22 standard is presented in this paper. In addition, a concise overview of the Cognitive Radio state of art is avowed and some basic design techniques are introduced to fulfill the budget simulation. We described also the wide band spectrum sensing receiver specifications taking the interest on the ADC requirements while processing communication or sensing. This receiver supports IEEE 802.22 as the first Cognitive Radio standard that tends to reuse the fallow TV spectrum without harming incumbent users. After that, the distribution of the RF specifications such as linearity, gain and noise are presented and the relationship that correlates them is discussed. Finally, to meet the challenge between linearity improvement and noise reduction, two gain modes are detailed. Simulations prove that the receiver provides 51 dB gain and -19.8 dBm IIP3 in the low gain mode. In the high one, the receiver performs up to 105 dB gain, -19.8 dBm IIP3 and 9.9 dB for Noise Figure. REFERENCES
[1] Ian F. Akyildiz, Won-Yeol Lee, Mehmet C. Vuran and Shantidev Mohanty NeXt generation/dynamic spectrum access/cognitive radio wireless networks: A survey, Computer Networks, Elsevier editor, Volume 50, Issue 13, 15 September 2006, Pages 2127-2159 R. Rajbanshi OFDM-Based Cognitive Radio for DSA Networks, Technical report ITTC-FY2008-TR-31620-05, The University of Kansas, September 2007 B. Razavi, RF Microelectronics, Prentice Hall Communications Engineering and Emerging Technologies Series, 1998. C. Cordeiro, K. Challapali, and D. Birru, S. Shankar N, IEEE 802.22: An Introduction to the First Wireless Standard based on Cognitive Radios, Journal of communications, Academy Publisher, vol. 1, no. 1, pp. 3847, April 2006. G. Chouinard 22-08-0327-02-0000- Wran-receiverperformance, Communications Research Centre, Canada (CRC), December 2008 Blaschke, V.; Renk, T.; Jondral, F.K. Cognitive Radio Receiver Supporting Wide-Band Sensing, IEEE International Conference on Communications, ICC workshop, 19-23 May 2008 Page(s):499 503. P.L G. Pinal Multi look-up table digital predistortion for RF power amplifier linearization, Ph.D. Department of Signal Theory and Communications, University Polytechnic of Catalonia Barcelona, December 2007. S. Shellhammer Spectral Mask Implications, IEEE P802.22 Wireless RANs, October 2008. Ayadi, D. Rodriguez, S. Loulou, M. Ismail, M., System level design of radio frequency receiver for IEEE 802.16 standard, 2008. 3rd International Design and Test Workshop. IDT, 20-22 December 2008. Pellon, L.E High dynamic range multibit ADC based receiver prototype employing dynamic error correction, Compound Semiconductor Integrated Circuit Symposium, 2005. CSIC '05. 30 Oct.-2 Nov. 2005

(a)

(b)

[2]

(c)

[3] [4]

[5] [6]

(d)
Figure3. Evolution of IIP3 (TOI) (a), gain (b), noise figure NF (c) and Pout_max (d) among receiver blocks for high gain mode TABLE I. Para Gain
(dB)

[7]

BLOCKS LEVEL SPECIFICATIONS IF1 Gain Amp 20.79 8 18 Mix 2 1 10 21 IF2 Gain Amp 20.79 8 18 VGA
LGM HGM

[8] [9] 54 11 32.6 [10]

LNA 9.42 1.5 -5.5

Mix 1 1 10 21

NF
(dB)

IIP3
(dBm)

322