WiMAX/WLAN Radio Receiver Architecture for Convergence in WMANS Bahar Jalali Farahani
Mohammed Ismail
IEEE Student Member Analog VLSI Lab, the Ohio State University Columbus, OH 43210
[email protected]
Fellow IEEE Analog VLSI Lab, the Ohio State University Columbus, OH 43210
[email protected]
Abstract— The trend toward increasing the data rate and providing the customers with wireless broadband access has initiated WiMAX systems. IEEE802.16e, the mobile version of IEEE802.16, provides WMAN (Wireless Metropolitan Area Network). Such system would need to work in the hostile environment of cellular phones and hence has the tough blocker requirements of GSM while providing the quality of service of WLAN (IEEE802.11a/g). Combination of these factors imposes stringent requirements on system and block level. Convergence of WiMAX and WLAN facilitates the broadband accessibility by using the already existing wireless networks wherever there is a WLAN hot spot. This paper proposes a fully integrated solution for a multi-band WiMAX/WLAN system based on a zero-IF architecture. System requirements are analyzed and block level specifications are derived. Challenges in system and block level are discussed and solutions are offered based on the current submicron CMOS technologies.
spot and switch to WiMAX mode everywhere else inside the metropolitan area. The paper proposes a zero IF solution that incorporates different calibration and compensation techniques to provide the flexibility of adapting to different bands and data rates. Block level challenges are discussed and solutions are suggested based on current technology.
I.
II.
Figure 1 shows the available RF bands that can be used for WiMAX. Although figure shows licensed and unlicensed bands, IEEE802.16e specifies only licensed bands for mobile application of WiMAX. Therefore, the system that we design here covers MMDS bands (2.5-2.69 GHz, 2.7-2.9 GHz) as well as 3.4-3.6 GHz. WiMAX uses OFDM with modulations that can be adaptively changed among BPSK, QPSK, 16QAM and 64QAM. Data rate is increased by using more complex modulations (64QAM, 256QAM is also optional) and within a specific modulation, it can be further increased by increasing coding rate. In order to have a comparable data rate with WLAN system, the bandwidth is chosen to be 20 MHZ. Table 1 shows some of the key system parameters of WiMAX.
INTRODUCTION
The new developed IEEE 802.16 standard [1] or, as it is known in industry, WiMAX (Worldwide Interoperability Microwave Access) is a new revolution in wireless broadband access. The new extension to this standard, IEEE 802.16e [2] fills the gap between very high data rate of WLAN (Wireless Local Area Networks) and high mobility of cellular systems. WiMAX standardizes the wireless broadband connectivity and provides a wireless alternative to DSL (Digital Subscriber Line) and cable [3]. IEEE 802.16e can reach further range than WLAN and provides point to multipoint wireless access in MAN systems. Using adaptive modulation, WiMAX supports BPSK, QPSK, 16QAM and 64QAM adaptively to exploit highest available data rate based on link quality. Although the future WiMAX systems can work in 2 to 11 GHz range of frequency, the first generation of mobile WiMAX is going to develop in MMDS and 3.5 GHz licensed bands.
III.
BLOCKING PROFILE
IEEE802.16 standard specifies the receiver adjacent and alternate channel rejection requirements. These requirements are more stringent for a 64QAM with 3/4 coding ration; hence, we consider this case in our design. Adjacent channel and alternate channel rejection should be measured under minimum and maximum signal strengths. The interference tests are depicted in Figure 2. This figure shows three scenarios, a weak signal with weak interferer (a), strong signal with strong interferer (b) and finally weak signal and strong interferer (c), where (c) imposes the hardest trade-off between noise and linearity. The required carrier to noise ratio (CNR) given in table 1 should be satisfied under all interferes tests. The blocker profile for IEEE 802.11b/g is brought in Figure 3. Comparing Figure 2 and 3, the difference between the strength of signal and interferer in WLAN is more than WiMAX however the requirement of BER is more relaxed (10-3
This paper is studying a single chip that supports both WiMAX and WLAN at 2.4 GHz (IEEE 802.11b/g) standards. The specifications given in these two standards are translated into the block level parameters. Such a system can use the available local area network whenever it is near a WLAN hot
0-7803-9197-7/05/$20.00 © 2005 IEEE.
SYSTEM REQUIREMENTS
1621
compared to 10-6 in WiMAX). That makes the overall requirement on both systems close to each other.
-39 -34
Unlicensed
-58
Unlicensed ISM (802.11b/g) 2400-2480
-30
WRC (new) 5470-5725
-62
3300-3400
f0
Wi-Fi
f0+20
f0
f0+40
(a)
f0+40
(b)
-30
Initial WiMAX Profiles
f0+20
Future WiMAX Profiles
2
US WCS MMDS 3 2305-2320 2500-2690 2345-2360 2700-2900
3.5 GHz Band 3400-3600
Licensed
Licensed
5 Low & Mid U-NII Bands 5150-5350 (802.11a)
4
-53
Upper 6 U-NII /ISM Band 5725-5850
f0+20
f0
f0+40
(c) WRC: World Radio Conference MMDS: Multic hannel Multipoint Distribution Service
Figure 2. Adjacent and alternate channel interferer tests in IEEE802.16.
Figure 1. 2 GHz to 6 GHz centimeter bands available for BWA (Broadband Wireless Access) [4]. Table 1. Different modulation scheme and key system parameters for OFDM WiMAX.
Modulation
Sensitivity [dBm]
NF [dB]
CNR [dB]
Max. Signal [dBm]
BPSK-1/2
-83.05
7
10.94
-30
QPSK-1/2
-80.05
7
13.94
-30
QPSK-3/4
-78.25
7
15.74
-30
16QAM-1/2
-73.05
7
20.94
-30
16QAM-3/4
-71.25
7
22.74
-30
64QAM-2/3
-66.75
7
27.24
-30
64QAM-3/4
-65.05
7
28.94
-30
-10 dBm In Band -23 dBm
-35 dBm -70 dBm f0
f0+25
f0+50
f0+425
Figure 3. IEEE802.11b/g blockers profile
Modulation
RX SNR [dB]
Signal BW [MHz]
NFFT
Nused
PSK-1/2
6.4
20
256
200
QPSK-1/2
9.4
20
256
200
QPSK-3/4
11.2
20
256
200
16QAM-1/2
16.4
20
256
200
16QAM-3/4
18.2
20
256
200
64QAM-2/3
22.7
20
256
200
64QAM-3/4
24.4
20
256
200
IV.
RECEIVER ARCHITECTURE
High level of integration and low power consumption suggest the zero-IF architecture as the best choice for multiband receiver. However, problems of zero-IF architecture, i.e. DC offset and sensitivity to I/Q mismatch shall be addressed carefully. A summary of the system specifications in WiMAX and WLAN (IEEE 802.11b) are given in Appendix I. Based on system requirements, link budget design was done using Excel spread sheets and specifications for individual blocks in the receiver chain were derived. Block diagram of the whole transceiver is given in Figure 4.
1622
hand, using a part of subcarriers and not all, will ease the requirements on PLL phase noise and DC cancellation at the price of lower data rate. Using all the 192 data carriers with 64QAM modulation can provide data rate of 100Msps which is almost twice the maximum data rate available in WLAN (54Msps). Therefore, there is a trade-off between number of used subcarriers and data rate.
ADC 90
Antenna
RF Filter ADC
DSP
Vin
DAC 90
Filtering & Amplifying
Vout
A
DAC
Pre-Amp for WiMAX modes Power Control
Σ + _ Vf
DC Offset Calibration
Gm
TX/RX AGC
Figure 4. Block diagram of the proposed zero-IF transceiver for WiMAX/WLAN
V.
DC Extractor
DC OFFSET CANCELLATION
Figure 5. DC Offset Cancellation [7]
DC Offset is one of the important challenges in zero-IF architecture. Usually DC offset is removed by high pass filtering the signal. This method can be easily used for WLAN where the information at DC is not significant. In IEEE 802.11b, the 22 MHz channel bandwidth is fully used to carry the information, but if the corner frequency of the Notch filter is %0.1 of the data rate, the increase in BER would be insignificant [5].
Pilot Carriers (8)
Data Carriers (192)
Guard Bands
Guard Bands
IEEE 802.11g uses 64FFT OFDM and subcarrier spacing is 312.5 kHz, therefore, a corner frequency of 156.25 kHz would not affect the information carrying carriers [6]. In WiMAX, however, using 256FFT OFDM makes a small subcarrier spacing (90 KHz for a 20 MHz bandwidth). Such a small cutoff frequency results in a long settling during the transients (e.g. TX/RX switching or switching between different modes). The cut-off frequency can be selected higher during the preamble for fast settling time. Another problem is that the drift of filter center frequency with respect to the LO frequency will result in undesired filtering of the near-zero subcarriers which degrade the BER significantly. This problem can be removed by using a mixed mode AFC (Automatic Frequency Correction) [7].
-128 -127
-101
101
126 127
Figure 6. 256 FFT- OFDM symbol in WiMAX shown in frequency domain
VI.
FREQUENCY SYNTHESIZER
Sub channelization also relaxes the requirement on PLL phase noise without which a technique for compensation of phase noise is needed or else the ICI (Inter-carrier Interference) would degrade the receiver performance significantly. Compensation and cancellation of phase noise in OFDM systems have been discussed in many papers [8]-[10]. The phase noise requirement for PLL was obtained from link budget analysis to be -110 dBc/Hz at 1MHz offset frequency. The settling time should be better than 5µs which is dictated by the TX/RX turn over time in IEEE802.11b/g.
In [8], a GmC filter is used to extract the DC offset and subtract it from the output of the mixer. The diagram of the DC extractor is shown in Figure 5. The cut-off frequency of the filter is tuned by changing the resistor or capacitor. This method is used for Bluetooth and 802.11b and could provide as low as 1KHz cut-off frequency for Bluetooth application. Figure 4 shows the OFDM channel in WiMAX. It uses 256 subcarriers from which 192 carry the data. If all the 192 available data carriers are used, the carrier spacing of 90 KHz for a 20 MHz bandwidth is resulted that would force very stringent requirements on PLL phase noise and also need a high Q high pass filter for DC offset cancellation. On the other
VII. FILTERING AND ANALOG TO DIGITAL CONVERSION A 3th order Butterworth filter can be used for channel selection and also as anti-aliasing filter. Analog to digital converter needs to work with high speed (more than 50Msps) but needs relatively low resolution. It can be implemented using Pipeline or a broadband sigma delta ADC.
1623
VIII. CONCLUSION This paper proposes a zero-IF receiver for multi-band
Table 3. Block requirements for zero-IF WiMAX/WLAN receiver
WiMAX/WLAN receiver. The convergence of these two standards enables the wireless system to maintain the broadband connection almost everywhere inside a metropolitan area. A comprehensive system design is done to derive the specifications of different blocks in the receiver. WiMAX systems need to work in a hostile environment similar to cellular phones and still maintain the high service quality of local area networks, these entail challenges in system design. WiMAX also provides data rates as high as 100Msps by using 256FFT OFDM. Using high points OFDM although increases the data rate but needs a very low phase noise PLL so that the resultant ICI (Inter-carrier Interference) products do not drastically degrade the radio performance. High pass filtering the DC offset is becoming more and more challenging as the higher point FFTs are used.
RF-Filter
LNA
Mixer
Gain high [dB] low
-1.5
18 0
10 5
IIP3 [dBm]
53
-7
8.6
IIP2 [dBm]
80
29.9
64.5
Block Input Referred Noise [v/Hz^0.5]
4.5E-10
3.50E-10
1.00E-9
Table 4. Block requirements for zero-IF WiMAX/WLAN receiver (cont.)
REFERENCES
Channel Select Filter
VGA
Channel Select Filter
ADC
-3
50 20
-3
0
Gain high [dB] low
[1] [2] [3]
IEEE 802.16 Rev d/D5-2004. IEEE 802.16e D4. D.J. Johnston, M. LaBrecque, “IEEE 802.16 WirelessMAN Specification Accelerates Wireless Broadband Access”, technology Intel magazine, August 2003. [4] “RF system and circuit challenges for WiMAX”, Fujitsu Microelectronics America Inc. [5] Ho-Kowan.Yoon, “Multi-standard receiver for Bluetooth and WLAN applications”, Ph.D thesis, the Ohio State University, 2004. [6] IEEE 802.11g [7] A. R. Behzad et al, “ A 5-GHz direct-conversion CMOS transceiver utilizing automatic frequency control for the IEEE 802.11a wireless LAN standard”, IEEE Journal of Solid-State Circuits, Vol. 38, No. 12, December 2003, pp.2209-2220. [8] Y. Juang et al., “A 2.4-GHa 0.25-um CMSO dual-mode directconversion transceiver for Bluetooth and 802.11b”, IEEE Journal of Solid-State Circuits, Vol. 39, No. 7, July 2004, p. 1185-1190. [9] J. Tubbax et al., “Compensation of IQ imbalance and phase noise in OFDM systems”, IEEE Tran. On Wireless Communication, Vol. 4, No. 3, May 2005, pp. 872-877. [10] G.Liu, “Compensation of phase noise in OFDM system using an ICI reduction scheme”, IEEE Trans. On Broadcasting, Vol. 50, No. 4, Dec. 2004, p. 399-407. [11] S.Wu, “A phase noise suppression algorithm for OFDM-based WLANs”, IEEE Communications Letters, Vol. 6, No. 12, Dec. 2002, pp.535-537.
IIP3 [dBm]
12
12
12
22.5
IIP2 [dBm]
64.5
64.5
64.5
75
1.00E-9
1.00E-9
1.00E-6
1.00E-6
-
-
-
55
Block Input Referred Noise [v/Hz^0.5] ADC Dynamic Range [dB]
Figure 7. Different signal levels in proposed zero-IF receiver for WiMAX system. Maximum Signal Level
0 1
2
3
4
5
6
7
8
9
10
Minimum Signal Level
-20 Signal level for Test1
-40
Signal Level Test2, adjacent channel
-60
Signal Level, Test2, alternate channel
-80
Blocker, Test1, adjacent channel
-100
Blocker, Test1, alternate channel
-120
Blocker, Test2 Block Input Refered Noise
-140
Figure 8. Different signal levels in proposed zero-IF receiver for WLAN system.
APPENDIX Table 2. System specifications of WLAN and WiMAX
Signal Levels for WLAN 0
Standard
IEEE 802.11b/g
IEEE 802.16
Frequency Bands
2.4~2.4835 GHz
2.5~2.69, 2.7~2.9, 3.4~3.6
Channel Bandwidth
22 MHz
20 MHz
Modulation scheme
DBPSK, DQPSK, QPSK
BPSK, QPSK, 16QAM, 64QAM
Multiple Access
DSSS, CCK
OFDM
QoS
FER
