JOURNAL OF TELECOMMUNICATIONS, VOLUME 8, ISSUE 2, MAY 2011 18

Multi-Band 4G Wireless sub-sampling receiver Based on: ISPD synthesizer and New Multi band IIR Filter Bilel GASSARA and Nouri MASMOUDI Abstract—— In 4G wireless receivers, the sub-sampling technique used in frequency translation offers to the receiver: a simple architecture, the multiband ability and gives a discrete time signal which can be treated easily. In previous work, we focused the research on the frequency synthesizer which presents the key assumption of multiband receiver and we profited the advantages of the Inverse Sine Phase Detector (ISPD) PLL with our new model of the inverse sine function, in addition, we designed a new multiband VCO applied in the synthesizer architecture. In this paper, we present our total and extended work by including all the designed parts in the receiver architecture and presenting a new design of an auto-reconfigurable base band anti-aliasing digital filter applied in GSM/DCS/WLAN-WiMax receiver, without using any added blocks like decimators and switches. The ADS simulation results show its ability to be used in multiband application without SNR degradation. Index Terms— Receiver, GSM, DCS, UMTS, WiMax, ISPD PLL, Multi-Band, Sub-sampling, Synthesizer, Wireless, IIR.

——————————
——————————

1 INTRODUCTION

T

he remarkable evolution in the wireless communication terminals and market leads many research activities to define RF system receiver architectures that can support various standards [1][2], with multimode operation, and to make a challenge in order to have a low cost, reduced size and increased battery life of user equipment hardware. This challenge open the door more widely for the software radio applications, so, the typical and faster way to reconfigure the multi-standard receiver hardware is the use of software tools with maximum hardware functionality sharing between various standards, that can be obtained by digitalizing the receiver architecture. Today, the push in digital receivers is to sample the RF waveform as close to the antenna as possible and eliminate problems caused by traditional mixerbased down-conversion architectures. In other hand, the sub-sampling frequency translation technique offers, at the time, the simplicity to the receiver architecture and the discrete time criterion to the RF signal. So, bandpass sampling, does not use any tuner / mixer to down convert signals, but instead takes advantage of digital aliasing to down convert a Nyquist band. Down conversion via direct RF bandpass sampling has a number of advantages over traditional homodyne approaches, including simplified hardware design with fewer analog components [2]. However, the RF sample clock must be tuneable or selectable, for thus, in previous work, we have designed multiband frequency synthesizer based on Inverse Sine

Phase detector Phase Locked Loop (ISPD PLL) with new model of the inverse sine function and a new multiband Voltage controlled Oscillator, in order to generate an RF carrier used to demodulate the received signal, and we demonstrated in previous publications [3],[4],[5],[6] its performances and its ability to be used in multiband homodyne receiver. In other hand, the performances of RF sample clock are not enough to have well reception; other parts of receiver are needed to be in high performances especially the base band anti-aliasing filter which may be reconfigurable and integrable. The recent works [1] use many controlled decimators and switched blocks to set the filter reconfigurable and to control its performances in each received band. However, the added blocks need to be controlled and synchronized by the DSP part of the receiver and need supplementary hardware parts like switches and clock, all that can generate negative effect, noise and additional response time and affect the filter performances. In this paper, we present our total and extended work by including all the designed parts in the receiver architecture, in addition, we describe a new theoretical design of the base band anti-aliasing filter, by using a digital IIR filter which profits the variability of the sampled signal cadence to be auto-reconfigurable, without using additional switched blocks. To show its performances, the complete receiver architecture are designed and simulated in GSM/DCS/ UMTS/WiMax –WLAN application by using the ADS tools.

————————————————

• B.GASSARA, Laboratory of Electronics and Technologies of Information (LETI), Circuits and Systems Group, National Engineering School of SFax, 3038 Sfax, Tunisia. • N. MASMOUDI, Laboratory of Electronics and Technologies of Information (LETI), Circuits and Systems Group, National Engineering Schoo of Sfaxl, 3038 Sfax, Tunisia. © 2011 JOT http://sites.google.com/site/journaloftelecommunications/

19

2 MULTISTANDARD RF SUBSAMPLING RECEIVER ARCHITECTURE

The benefits of simplicity of the homodyne architecture and its ability to be used in multiband-monochip applications pushed us to profit it in our multiband receiver architecture, but its diadvantages and especially the DCoffset problem make a gap which must be resolved. Today, there are many solutions to reduce and cancel the DC-offset problem [7],[8] which can be applied in RF or in base band receiver part. The sub-sampling technique used in frequency translation can be one of these solutions, in addition to its simplicity; it can minimize the DC-Offset at the mixer [9] in homodyne architecture because there is not subtraction of tow equal frequency, it is based simply on the frequency translation du to the sampling effect. The proposed architecture of the multiband receiver combines both the simplicity of homodyne architecture and the advantages of sub-sampling mixer. Barrak et al [1] describe a multi-standard direct RF sampling architecture for GSM/UMTS/802.11g that uses a track-and-hold followed by a discrete time band pass decimation filter followed by an ADC. In this work, we adopt the same architecture by interesting more on the two keys assumption of the multiband receiver which are the frequency synthesizer and the low pass digital filter. Base-Band low pass filter

RF Filter

LNA Programmable

E/B S/H

Divider

S&H

Sin -1 (x) -

π 2

Command band Step variation command

Image Reject Filter

E/B S/H

Band n

Divider ACG

IIR

Base-Band low pass filter

Multi band VCO

ADC Multiband Frequency Synthesizer

0°

RF Filter

Reference signal

Band1 Band2 Band3

Step selection

IIR

90° RF Filter

the variation of the step selected to make the synthesizer able to operate at different bands which are usually didn’t have the same characteristics. Compared to the synthesizer based on standard PLL, we have demonstrated in previous publications [3] that the ISPD offers a large coverage range synthesizer without using any controlled gain block that simplify the circuit’s architecture and make it amenable for monolithic integrated application. Further more, the mains parts of the proposed frequency synthesizer are the Inverse Sine Phase Detector (ISPD) circuit and the multi-band VCO circuit, so, the performances of the synthesizer are depending to the performances and robustness of there parts. Therefore, the design of the ISPD circuit is involved with employing a new mathematical description of the ISPD characteristic. A new mathematical model of the Inverse Sine function approximation is already studied and designed in previous study [4]. So, the conception of the proposed synthesizer is reduced to conceive and optimize the Inverse Sine Phase Detector (ISPD) and the Multi-Band VCO which will determine the performances of the Local Oscillator circuit.

ADC

selector

Fig. 2. Architecture of the proposed multi-band ISPD frequency synthesizer

ACG

Fig. 1. Multiband sub-sampling receiver Architecture

So, the architecture of multiband receiver is based on track and hold component controlled by RF multiband frequency synthesizer based on Inverse Sine phase detector (ISPD) PLL and followed by low pass multiband IIR filter, Automatic controlled gain and analog to digital converter. In this paper, we will remind our proposed frequency synthesizer architecture and we will focus on the base band digital filter by giving new approach which profits the variability of the sampling signal cadence to set the filter auto-reconfigurable. Therefore, the challenge of the proposed architecture is to use only one digital filter (IIR) to select the base band of each received standard without using many controlled decimators.

2.1 Multiband ISPD frequency synthesizer The proposed multi-band synthesizer use the ISPD PLL architecture associated with digital controlled multi-band tuned LC VCO and two other digital controlled blocks which control the steps selection in the selected band and

Table 1 summarizes the simulated performances of the proposed multiband frequency synthesizer such as the lock up time, keep range and capture range in each band, that shows that the synthesizer can cover widely all the desired bands. TABLE 1 FREQUENCY SYNTHESIZER PERFORMANCES Band WiMax DCS/DECT/UMTS GSM

Lock up time 8ns 12.7ns 14ns

Keep range

Capture range

1.4GHz 587MHz 73MHz

0.7GHz 101.4MHz 30MHz

2.2 Inverse Sine Phase Detector (ISPD) architecture The ISPD frequency synthesizer is based on Inverse sine function block to give directly the phase error of input sampled signal compared to π/2. Therefore, the ISPD circuit must be having many well criteria like simplicity, rapidity, integrability and especially high fidelity in the inverse sine function calculation. So, we have been study carefully the inverse sine function approximation and we have developed a new model

20

[4] to approximate this function with lowest error and simplest mathematical description compared to the current model [10]. The proposed model of the inverse sine function is based on a combination between a square root function and a third order polynomial function, the coefficients are determined carefully to give a function curve which follows precisely the inverse sine function, so, the proposed model is described by the following expression: (1) f ( x ) = 1.37 1 + x − 1 − x − 0.37 x − 0.025 x3

(

)

In second hand, we have studied the robustness of our model by controlling its divergence from the inverse sine function due to the coefficients variations and fluctuations which can be occurred at the electronic conception and fabrication phase of the ISPD circuit, and we demonstrated that our proposed model is stable and robust [4]. Finally, we have conceived the functional block diagram and the electronic transistor schema of ISPD and we verified its performance in demodulation and in the proposed frequency synthesizer applied in GSM/DCS/ UMTS/WiMax-WLAN receiver. Input Signal (x) Port P1

VMult

VMult

Attenuator

V_DC att Vdc=0.4 V

av3

Output signal Arcsin(x)

VMult VSum

vsqrt Port P2

0. 4.x + 0. 01. x 3 vsum1

VSum

VCVS SRC7

V_DC Vdc=1 V

VCCS

1.4 ´ nmos4 MN1

ues of phase noise and Factor Of Merit (FOM) (using Fong expression [11]) of the proposed circuit at 1MHz Offset frequency, at the different generated bands. As can be seen in these results, the best phase noise of the proposed multi band circuit is -127dBc at 2.3GHz and the best FOM is -186 dB. TABLE 2 PHASE NOISE AND FOM SIMULATED VALUES OF THE PROPOSED MULTI BAND VCO Bandes

Frequency GHz 2.27 3.5 1.69 2.21 0.83 1

WiMax DCS/UMTS GSM

Phase Noise @1MHz -127.155 dBc -124.1 dBc -122.885 dBc -118.416dBc -120.236 dBc -117.246 dBc

FOM @1MHz -186.08 dB -186.786 dB -179.247 dB -177.108 dB -170.422 dB -169.05 dB

Figure 4 shows the simulated oscillation frequency versus the control voltage of the proposed VCO as a function of the respective switching combinations. As can be seen in Figure 4, the VCO can cover the GSM standard at tow separate bands (from 0.83GHz to 0.938 GHz and 9.43GHz to 1GHz) and the DCS/DECT/UMTS standards at one band (from 1.69GHz to 2.21GHz) and the WiMax fixe standard also at one band (from 2.27GHz to 3.5GHz). So, the overall frequency tuning range is 20% at the GSM band and 30% at the DCS/DECT/UMTS band and 78% at the WiMax band. So, the proposed multiband VCO can cover wide frequency range.

(1+ x) - (1 - x)

sqrt1

sqrt2

VCCS nmos4 MN2

VCVS

m2 vtune=0.950 freq[1]=3.505E9 Bit2=0.000000, Bit1=0.000000 m2

Fig. 3. ADS Simulation’s diagram of the ISPD

3.365

m1 vtune=2.975 freq[1]=2.274E9 Bit2=0.000000, Bit1=0.000000

2.3 Multiband quadrature LC tuned VCO The LC VCO is characterized by its stability and its well phase noise performances compared to the other oscillator architecture. Therefore, we adapted the LC VCO architecture in multiband frequency synthesizer application by adding switched capacities blocks to select the desired band. The adaptation of the LC VCO architecture, for the GSM/DCS/DECT/UMTS and WiMax application, reside mainly on the determination of the number of switched capacitors branches needed to cover all the desired bands and the capacitors values. The adapted VCO core consists of cross-coupled transistors, to drive the LC tanks to generate full-swing output voltages, spiral inductor and a capacitance block to induce the required oscillation frequencies. The capacitance block is made by a varactor to assures the continuously variation of the frequency and a switched capacitors to assure the discrete variation of the bands. The VCO architecture and performances are already published [3],[5], and table 2 summerizes the simulated val-

freq[1], GHz

WiMax

2.865

1.865 1.365

m1

m4

2.365

DCS/UMTS

m8m6

vc2 vc1 m3

m4 vtune=1.231 freq[1]=2.217E9 Bit2=0.000000, Bit1=3.000000

GSM900

m3 vtune=3.312 freq[1]=1.697E9 Bit2=0.000000, Bit1=3.000000

0 1 m7 m5

0.865 0.5

1.0

1.5

m6 vtune=1.006 freq[1]=9.434E8 Bit2=3.000000, Bit1=3.000000 m8 vtune=0.931 freq[1]=1.009E9 Bit2=3.000000, Bit1=0.000000

2.0 vtune

2.5

0 0

3.0

1 0 1 1

3.5

m5 vtune=3.425 freq[1]=8.838E8 Bit2=3.000000, Bit1=3.000000 m7 vtune=3.369 freq[1]=9.380E8 Bit2=3.000000, Bit1=0.000000

Fig. 4. Frequency versus the control voltage of the proposed VCO as a function of the respective switching combinations

3 NEW MULTIBAND BASE BAND FILTER DESIGN In order to filter the base band signal, we can take advantage of the discrete criterion of the output subsampled signal by using a digital second order IIR filter. Our idea basis on using the same filter to receive all the

21

proposed standards, with the same architecture without adds supplementary blocks. So, this filter will be autoreconfigurable due to the variation of the sampling frequency. At the beginning, we remind that this filter is an antialiasing low pass filter which select the base band signal and attenuate all others aliased bands due to the subsampling and avoid image aliasing. Let's Fc and Fr be the low-pass filter cut-off and rejection frequencies normalized by half RF sampling frequency (fs/2). Blocker level

Amin fc

fr

fs

Fig. 5. Base band – anti aliasing filter mask

Figure 5 shows the low pass filter mask; "Amin" is the required attenuation at the rejection frequency to reject the blocker level in the appropriate band. The following table shows the required Amin for each band:

To design our filter, we determined its coefficients to satisfy both, the largest pass band and the lowest sampling frequency to cover all over the frequency range. Let remind that in our application, the largest pass band frequency is occupied by the WiMax and Wifi standard (22MHz) and the lowest sampling frequency is there of the first band of the standard WiMax (B1) which is 2305MHz. So, the normalized cut-off frequency is described by: B fc _ 3dB = 1.3 × ch _ max 1.3 ×11 (4) 2 Fc = = = 0.0124 f s _ min 2305 / 2 2 The filter coefficients are determined by using the MATLAB tools (function BUTTER for a Butterworth low pass filter), and the transfer function is represented by the following equation: 0.3696 + 0.7393 × z −1 + 0.3696 × z −2 (5) H ( z) = ×10−3 1 − 1.9449 × z −1 + 0.9464 × z −2 In order to make sure of the filter stability, we have studied the Nyquist diagram of its transfer function, so it presents two poles (Z1=0.9724+0.0268i and Z2=0.9724 0.0268i), which are both inside the unit circle diagram (figure 6), that prove the stability of the designed filter. Now, lets study the sampling frequency variation effect on the pass band filter, and we must remind that the multiband VCO applied in the frequency synthesizer can cover a frequency range from GSM standard to the sixth band of WiMax (3.4GHz).

TABLE 3 REQUIRED MINIMAL ATTENUATION AT REJECTION FREQUENCY

Root Locus 0.3

Standard

Amin (dB)

GSM

DCS

UMTS

802.11g

85

47.3

51.8

60

0.1π/T

0.2

System: H Gain: 0 0.2 0.1 0.3 Pole: 0.972 + 0.0268i 0.5 0.4 0.6 Damping: 0.7 0.8 Overshoot (%): 0.9 Frequency (rad/sec):

0.1 Imaginary Axis

In other hand, the rejection frequency is described as following as: B (2) f r = f s − ch 2 The cut-off frequency (Fc) must be grater than the half of the selected channel bandwidth (Bch) with a required margin of 30% [12], so it will be described by: B (3) f c ≤ 1.3 × ch 2 In other hand, the sampling frequency depends of the central frequency of the received standard, so, the cadence of the sampled signal will be variable. The recents works use a reconfigurable decimation filter blocks to make the cadence signal constant, however in this work, the idea is to lets the cadence signal variable by removing the decimation filter and profits the cadence variation to reconfigure the band pass and the cut-off frequency of the base band filter. So, by using the same architecture, our filter will has the same coefficients, the reduction of the sampling frequency produce the reduction of the cut-off frequency and the pass band of the filter, consequently the filter will adapt itself to the received standard.

0 System: H Gain: 0 Pole: 0.972 - 0.0268i Damping: Overshoot (%): Frequency (rad/sec):

-0.1

-0.2 0.1π/T 0.85

0.9

0.95

1

1.05

Real Axis

Fig . 6. Nyquist Diagram of the transfer function filter

So, in the WiMax standard, the sampling frequency is located as following as: 2305MHz ≤ f e _ WiMaX ≤ limVCO = Bande6 = 3.4GHz (6) Therefore, the filter pass band will be flanked by: (7) 14.29 MHz ≤ B pass _ WiMaX ≤ 21.08MHz The last equation shows that the minimal value (14.29MHz) of the pass band filter in the WiMaw stan-

22

WiMax WLAN UMTS DCS GSM

Required Minimal pass band 11 MHz + 30% 11 MHz + 30% 3.84 MHz + 30% 1.4 MHz + 30% 0.1 MHz + 30%

Minimal pass band 14.29 MHz 14.88 MHz 13.082 MHz 11.19 MHz 5.8 MHz

Maximal Pass band 21.08 MHz 15.397 MHz 13.45 MHz 11.65 MHZ 5.95 MHz

Table 4 proves that the minimal pass band of the filter in each band is grater than the required pass band. But this criterion is insufficient to characterize our filter; we must verify the good selectivity of the filter and the rejection of the appropriate blocker in each standard. For that, we have simulated the gain vs frequency response of the filter at every standard; the results are given by the figure 7. The rejection frequency of the filter depend of the sampling frequency (fr=fs-Bch/2), in other hand, the homodyne architecture of receiver make the sampling frequency (fs) is too grater than the half of pass band (Bch/2), consequently, the rejection frequency will be pushed to the central frequency of received signal, so that the blockers will be strongly attenuated. fcGSM

dB(WiMax) dB(UMTS) dB(WLAN) dB(DCS) dB(GSM)

fcWiMax freq=14.37MHz dB(WiMax)=-3.048

fcGSM freq=5.802MHz dB(GSM)=-3.007

-40 Amin_UMTS Amin_802.11g

-60

AttGSM

-80

Amin_GSM

-100 1E6

-60 -70 -80 -90 -100 2.40

2.41

2.42

2.43

2.44

2.45

2.46

2.47

2.48

freq, GHz Fig. 8. 802.11g input Spectrum

fcWiMax

0

-3dB

-20

-55

1E7

AttGSM freq=360.6MHz dB(GSM)=-85.420

1E8

4E8

dBm(Output_Filter_IIR) dB m(Sortie_Filtre_ IIR) dB m(Sortie_Echantillo nneur) dBm(Output_Sample&Hold)

Standard

tools and we have determined the Noise figure ratio (NF) and the bit error rate (BER) in each band. The WiFi signal is modelled by 64-QAM signal having a central frequency 2442MHz and a 22MHz band width. The input power is -65dBm and an added noise of 174dBm/Hz, according the specifications of 802.11g standard with 54Mbps bits rate. Figure 8 shows the 802.11g input spectrum. Figure 9 shows the sub-sampled signal which makes obvious the frequency translation from RF band to the base band and the response of the designed filter. The signal to noise ratio measured at the filter base band signal is SNRout=33.38dB, that make a Noise Figure NF=6.62dB which is lower than the required NF (Table 5).

dBm(WLAN)

dard is grater than the half of the required pass band (22MHz) with a margin of 1.3%. We have verified this criterion according to the other standards (GSM/DCS/UMTS/WLAN), the results are showed at the following table: TABLE 4 PASS BAND VARIATION OF THE DESIGNED FILTER

0 -50 -100 -150 -200 0.0

0.5

1.0

1.5

2.0

2.5

3.0

freq, GHz

freq, Hz Fig. 7. Gain vs Frequency response of the designed filter

Figure 7 shows that all the minimal required attenuations (Table 3) are obtained at a frequency range: 100MHz-360MHz. In conclusion, the designed low pass filter obeys the specifications of the all standards filter mask, and can be used in the multiband receiver.

4 GSM/DCS/UMTS/WLAN-WIMAX

APPLICATION

To evaluate its performances, we have simulated the whole architecture of the multiband sub-sampling receiver based on RF ISPD synthesizer and the proposed multiband base band filter in GSM/DCS/UMTS/WLANWiMax application. Therefore, we have used the ADS

Fig. 9. 802.11g Sub-sampled translated signal and filter response

Figure 10 shows the 64-QAM constellation diagram of emitted signal and received signal; and by using the Error Vector Magnitude (EVM) simulation tool, we determined that the simulated EVM is about 4.61% which is lower than the 802.11g required EVM (5.6%). The simulated BER (for 100 QAM symbols) is 5.10-3 which is grater than the required BER (10-5) but this value is obtained without adding a correctors blocks, so it can be performed and we can evaluate it as acceptable.

23

dBm(Sortie_Echantillonneur) dBm(Output_Sample&Hold)

imag(constellation_out) imag(constellation_In)

1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -1.5

-1.0

-0.5

0.0

0.5

1.0

-70 m1 -80 -90 -100

-120 -130

dBm(Sortie_Filtre_IIR) dBm(Output_Filter_IIR) dBm(Sortie_Echantillonneur) dBm(Output_Sample&Hold)

m3 -100

-150

m3 m3 freq= 2.140GHz freq=2.140GHz dBm(Sortie_Echantillonneur)=-104.724 dBm(Output_Sample&Hold)=-104.724

-250 1.5

2. 0

2

3

4

5

6

7

8

9

freq, MHz

The simulated BER is 59.10-3, is grater than the required BER (10-3) but it can be ameliorated by adding a baseband corrector Blocks. The following table summarizes the simulated Noise Figure in each desired band. Compared to the required NF, the table shows that the obtained values are acceptables. TABLE 5 NOISE FIGURE SIMULATED VALUES REQUIRED NF SIMULATED NF

7

-50

1.0

1

Fig. 12. GSM baseband signal spectrum

The UMTS signal is modelled by QPSK signal having a central frequency 2140MHz and a 3.84MHz band width. The input power is -90dBm and an added noise of 174dBm/Hz, according the specifications of UMTS standard by using CDMA spread spectrum technique. Figure 11 shows the sub-sampled signal which makes obvious the frequency translation from RF band to the base band and the response of the designed filter. The signal to noise ratio measured at the filter base band signal is SNRout=32dB, that make a Noise Figure NF=8dB which is lower than the required NF (Table 5).

0.5

0

1.5

Fig. 10. 64-QAM Constellation diagram

0.0

m2 m2 freq= 375.0kHz freq=375.0kHz dB m(Sortie_Ec hantillonneur)=-108.521 dBm(Output_Sample&Hold)=-108.521

m2

-110

real(constellation_In) real(constellation_out)

-200

m1 m1 freq=25.00kHz f req= 25.00kHz dBm(Sortie_E chant illonneur)=-81.372 dBm(Output_Sample&Hold)=-81.372

2. 5

3. 0

freq, GHz

Fig. 11. UMTS Sub-sampled translated signal and filter response

The simulated BER is 29.10-3, is grater than the required BER (10-3) but it can be ameliorated by adding a baseband corrector Blocks. The GSM signal is modelled by GMSK signal having a central frequency 940MHz and a 200KHz band widh. The input power is -102dBm and an added noise of 174dBm/Hz, according the specifications of GSM standard. Figure 12 shows the GSM baseband signal. The simulated SNR is SNRout=27dB, according to 30dB input SNR, that make a Noise Figure NF=3dB which is lower than the required NF.

GSM 10DB

UMTS 9.4 DB

802.11B 14.8 DB

802.11G 7.5 DB

802.16E 8 DB

3DB

8DB

7DB

6.6DB

6.8DB

CONCLUSION

New base band multiband filter stage architecture is proposed in RF sub-sampling multi-standard radio receiver. The proposed architecture is based only on an IIR filter which profits the variation of the sampling frequency to set its parameters as needed as the received standard requirements, without adding any switched block or decimator stage. Using into account, the simplicity and the stability of the integrated system, the transfer function of the base band IIR filter is determined. So, the filter model is implemented and used in GSM/DCS/UMTS/WLANWiMax receiver application. The ADS simulation results showed an optimum receiver performance without SNR degradation obtained in the other approach. Moreover, the proposed ISPD synthesizer, the subsampling techniques, and the proposed baseband filter give the receiver architecture more simplicity and allow saving area and power in favour of 4G programmable single chip - multimode - very high scale integration.

24

REFERENCES [1]

R. Barrak, A. Ghazel, F. Ghannouchi, "Discrete-time Decimation filter Design for Multistandard RF Sub-sampling Receiver", ICECS 2007. 14th IEEE International Conference on Electronics, Circuits and Systems, , pp. 1396-1399, 11-14 Dec. 2007. [2] G. L. Fudge, Mark A. Chivers, Sujit Ravindran, Ross E. Bland, Phillip E. Pace, "A Reconfigurable Direct RF Receiver Architecture", IEEE International Symposium on Circuits and Systems, ISCAS 2008, pp.2621-2624. [3] B. Gassara, M. Abdellaoui, N. Masmoud, " 4G Wireless Receiver Architecture based on ISPD PLL Supporting GSM/DCS and WiMax", Transactions on Systems, Signals & Devices, Vol. 4, No. 3, pp.351-368, 2009. [4] M. ABDELLAOUI, B. GASSARA, N. MASMOUDI, "A new model of an Inverse Sine Phase Detector to design ISPD PLL Demodulator without using any filters", AEUE - International Journal of Electronics and Communications, Vol. 61, no.1, pp.10-21, 2007. [5] Bilel Gassara, Mahmoud Abdellaoui, Nouri Masmoudi, "Multi Band Frequency Synthesizer Based on ISPD PLL with Adapted LC Tuned VCO", INTERNATIONAL JOURNAL OF ELECTRONICS, CIRCUITS AND SYSTEMS, vol.1, No. 3, pp. 150-156, 2007 ISSN 1307-4156. [6] B. GASSARA , M. ABDELLAOUI, N. MASMOUDI, "4G Wireless Receiver Architecture based on ISPD PLL Supporting GSM/DCS and WiMax", Fourth International Multi-conference on System, Signals & Devices, March 19-22 2007 Hammamet Tunisia. [7] Y. Zheng, J. Yan , Y. P. Xu, "A CMOS VGA with DC offset cancellation for direct-conversion receivers", IEEE Transactions on Circuits and Systems Part I: Regular Papers, vol.56, Issue.1, pp.103-113, January 2009. [8] S.B. Park, M. Ismail , "DC offsets in direct conversion multistandard wireless receivers: Modeling and cancellation", Analog Integrated Circuits and Signal Processing, vol.49, Issue.2, pp.123 – 130, October 2006. [9] I. Mostafa et al, "Subsampling RF Receiver architecture", United States Patent, No. 7,110,732 B2, 19. Septembre 2006. [10] J.K.Seon, "Analyse et conception d’un circuit PLL utilisant un détecteur de phase Arcsinus". Thesis at High National School of Telecommunications. Paris- France, ENST Paris, 2000. [11] N.Fong et al, "Design of wide band CMOS VCO for multiband wireless LAN applications", IEEE J. Solid State Circuits, Vol.38, no.8, pp1333-1342, Aug. 2003 [12] C. Rebai, M. Ben-Romdhane, P. Desgreys, P. Loumeau, A. Ghazel. , "Pseudorandom signal sampler for relaxed design of multistandard Radio Receiver", MicroelectronicsJournal. Elsevier, Vol.40, Issue 6, pp. 991999, June 2009.

Bilel GASSARA was born in Sidi Bouzid, Tunisia, on September 28, 1980. He received his Dipl. Engineer degree in Electrical Engineering from the National Engineering School of Sfax (ENIS) in 2004, and his Master Degree in Electronic Telecommunication from ENIS, in 2005, and his Ph.D. degree in Electrical Engineering from ENIS in 2010. In September 2007, he joined the Department of Electrical and Automatic of the National Engineering School of Gabes as an assistant teatcher. His research interests include the areas of the RF circuits and systems: working with theory, analysis and design of multiband demodulator based on ISPD PLL with CMOS process technology.

Nouri MASMOUDI was born in Sfax, Tunisia, in 1955. He received his Electrical Engineering degree from the Faculté des Sciences et Techniques de Sfax, Tunisia, in 1982, his DEA degree from the Institut National des Sciences Appliquées de Lyon and Université Claude Bernard de Lyon, France in 1982. From 1986 to 1990, he prepared his thesis at the laboratory of Power Electronics (LEP) at the Ecole Nationale d’Ingénieurs de Sfax (ENIS). He then received the ‘Thèse de 3ème Cycle’ at the Ecole Nationale d’Ingénieurs de Tunis (ENIT), Tunisia in 1990. From 1990 to 2000, he was an Assistant Professor at the Electrical Engineering Department at the ENIS. Since 2000,

he has been an Associate Professor and Head of the group ‘Circuits and Systems’ of the SFAX Laboratory of Electronics and Information Technology. Currently, he is responsible for the Electronic Master Program at ENIS. His research activities have been devoted to several topics: Design, Telecommunication, Embedded systems and Information technology.