Multi-Antenna Techniques for OFDM based WLAN Mursalin Habib, Ayon Quayum

Fakhrul Alam

Nazrul Islam

Technical Division, GrameenPhone Ltd. Gulshan – 2, Dhaka 1212, Bangladesh [email protected] [email protected]

IIMS, Massey University, Albany Private Bag 102 904, NSMC, Auckland New Zealand F. [email protected]

Department of EEE, BUET Dhaka 1000, Bangladesh [email protected]

Abstract— Wireless Local Area Networks (WLANs) have been one of the great communications technology success stories of the past few years. Orthogonal Frequency Division Multiplexing (OFDM), a multi-carrier based transmission technique, is currently being used for several popular WLAN systems. In this paper we investigate several techniques that employ adaptive antenna array to improve the performance of an OFDM based WLAN. We employ the transmission format specified by the IEEE 802.11g standard. We demonstrate the improvement in performance with the aid of computer simulations.

I. INTRODUCTION Low expense for infrastructure and high data rate capability is making Wireless Local Area Networks (WLANs) very successful. Orthogonal Frequency Division Multiplexing (OFDM) has become the de facto transmission technique for WLANs. Among the IEEE’s 802.11 [1] family of WLANs there are two OFDM based standards namely IEEE 802.11g [2] and 802.11a [3]. IEEE has also started the process of standardizing OFDM based IEEE 802.11n [4], which will provide 100 Mbps+ date rate by applying the Multiple Input Multiple Output (MIMO) [5] techniques. Space-time processing techniques based on adaptive smart antenna systems [6] are becoming popular for improving the capacity, throughput and link quality of wireless systems. These techniques operate by taking advantage of array gain, diversity gain, spatial multiplexing and interference reduction [5]. In this paper we will investigate several spatial processing techniques that can be employed to improve the performance of an OFDM based WLAN. Here is how the rest of the paper is organized. Section II provides a brief discussion on OFDM, and the IEEE 802.11g transmission format that was implemented for simulation purposes. In section III we provide an overview of antenna array, beamforming and diversity techniques and the Minimum Mean Square Error (MMSE) [7] criterion for combining signals. Section IV introduces the multi-antenna techniques. Simulation results and discussion on the results are provided in section V. Section VI concludes this paper. II.

OFDM AND IEEE 802.11G

A. OFDM OFDM is a multi-carrier technique where the data is converted into multiple parallel streams and orthogonal subcarriers are employed to each of these streams [8]. The transmitter can multiplex data symbols onto sub-carriers by

employing the computationally efficient Inverse Fast Fourier Transform (IFFT) operation. An OFDM receiver can perform the demodulation with a simple Fast Fourier Transform (FFT) operation. This eliminates the banks of sub-carrier oscillators and coherent demodulators required by conventional frequency division multiplexing systems. B. IEEE 802.11g Transmission Format IEEE 802.11g operates at the 2.4 GHz Industrial, Scientific and Medical (ISM) band whereas IEEE 802.11a operates at the 5 GHz Unlicensed National Infrastructure (U-NII) bands. Both these standards have the same Media Access Control (MAC) and similar physical (PHY) layer. In this paper the PHY layer of the IEEE 802.11g was implemented for simulation purposes. However we should emphasize that the results and discussions apply to 802.11a as well as any generic OFDM based WLAN. Serial information bits are converted to 48 parallel data streams and 4 extra known data bits are added for estimation and detection purposes. The central frequency is intentionally made zero. There are another 11 zero padded sub-carriers which are required to construct the 64 point IFFT block. In order to mitigate the effect of inter symbol interference (ISI), cyclic prefix (CP) of length 16 is added. The first four time slots or OFDM symbols do not carry any information bits; all the available 52 sub-carriers contain training symbols. These identical training symbols are later used as the reference signal for computing the weight vectors employing the MMSE algorithm. The details of the transmission format for IEEE 802.11g can be found in [2]. Table 1 summarizes the parameters of the transmitted signal. Table 1: WLAN Parameters FFT Points (Sub-carriers) Data Sub-carriers Pilot Sub-carriers Modulation Scheme Sub-carrier Spacing Guard Interval (CP) Symbol Duration with CP Payload

64 48 4 16 QAM 312.5 kHz 0.8 µs 4 µs 480 bytes

III.

SPACE-TIME PROCESSING & MULTI-ANTENNA TECHNIQUES

A. Antenna Array An antenna array is necessary to collect the signals for spatial processing. It consists of a set of antenna elements that are spatially distributed at known locations with reference to a common fixed point [9]. Uniform Linear Array (ULA) is an array configuration where the centers of all the antenna elements are aligned along a straight line and the spacing between the elements are equal [6]. In this paper we have employed an antenna array with two isotropic elements separated by half the carrier wavelength. B. Receive Beamforming One of the most common application of a receive antenna array is beamforming [6]. When the desired user and the interferer are in different spatial location, the spatial separation is utilized by a beamformer. Signals obtained from the antenna elements are combined to reduce the effect of co-channel interference and improve the quality of the desired signal. The output of the antenna array is given by y (t ) = wH (t ) x (t )

(1)

Here w = [ w1 w2 ... wN ] is the N × 1 weight vector, N being the number of antenna elements and H denotes Hermitian transpose. The weight vector is chosen to optimize some beamforming criterion. Popular beamforming criteria include Maximum Signal to Interference and Noise Ratio (MSINR), Constant Modulus (CMA) Minimum Mean Square Error (MMSE), etc [6]. T

C. Receive Diversity Receive diversity techniques with an antenna array can be applied to mitigate fading [5]. This is in addition to the interference cancellation attained from steering beams towards the desired user and/or steering nulls in the direction of interferers. The signal envelopes observed across the elements of an antenna array should have very small cross-correlation in order to achieve spatial diversity gain [5]. In this scenario, if the signal at one of the elements is going through a fade, it is highly unlikely that the signals at the other elements are encountering that at the same time. So there is nearly always good signal reception on one of the antenna elements. Therefore combining the signals from various elements will reduce the fading and improve signal fidelity. Equation 1 describes diversity combinig at the receiver. However for beamforming the weight vector is chosen to reduce co-channel interference whereas for diversity combining the weight vector is chosen so that the fading is reduced. D. Transmit Diversity The receive diversity approach is very easily applicable for the WLAN Access Points (AP). However the cost, size and power requirements make it less attractive for small terminals like a wireless Network Interface Card (NIC).

Transmit diversity schemes shift the burden of complex processing and multiple antenna arrays from the receiver to the transmitter. As a result transmit diversity schemes are attractive for a wireless network where a single access point serves a large number of small wireless terminals. One of the most popular transmit diversity scheme is the so called Alamouti code which improves the signal quality at the receiver by simple processing across two transmit antennas on the transmitter side [10]. The most attractive feature of this scheme is the complete absence of feedback from the receiver to the transmitter and the small computation complexity. Alamouti code is considered to be a 2x1 orthogonal space-time block code [11].This scheme has been adopted by several wireless communication systems including the WCDMA based Third Generation (3G) standard [12] for cellular systems. The details of the Alamouti scheme can be found in [10]. E. MMSE The MMSE criterion intends to find a weight vector that will minimize the Mean Squared Error (MSE) between the combined signal and some desired (or reference) signal. The weight vector is given by the Wiener solution [7] −1

w MMSE = R xx r xd

(2)

Here R xx is the received signal covariance matrix defined as E ( x x H ) , where x is the received signal vector and H

denotes Hermitian transpose, r xd is the correlation of the received signal vector x with the desired signal (or the reference signal) d(k) and is defined as E [ x (k ) d (k ) ] . The pilot and training symbols as specified in the IEEE 802.11g standard (see section II B) are utilized as the reference signal to estimate the weight vector. In this paper we apply the Direct Matrix Inversion (DMI) [6] technique to compute the weight vectors given by Equation 2. However practical systems may adopt computationally simple iterative algorithms [7] like the Least Mean Square (LMS) algorithm. After the initial training with the training symbols at the beginning of the frame, the weights can be updated with the help of the pilot symbols. IV.

MULTI-ANTENNA TECHNIQUES FOR WLAN

A. Receive Beamforming for WLAN The Receiver employs a ULA with two omni directional elements. The spacing between the elements is half the carrier wavelength. At 2.4 GHz this separation is less than 5 inches. Each antenna has its own set of receiver blocks, for removing cyclic prefix, extracting pilot and training symbols and performing the FFT operation. The weight vector of the beamformer is computed with the aid of the extracted training symbols based on the MMSE criterion. The receiver configuration is shown in Figure 1. Only one set of weight vector is necessary for all the sub-carriers at the absence of multipath. However in a multipath environment, different sub-carriers may experience

different channel frequency response and consequently will require different set of weight vectors. Depending on the coherence bandwidth of the channel several contiguous subcarriers may experience similar channel response and may be served by the same set of weight vectors [13], [14].However in such a scenario, the transmission format has to be suitably modified to allocate the appropriate number of training and pilot symbols for the different subcarrier groups.

B. Receive Diversity combiner for WLAN Similar to the beamformer described in section IV A, the receiver employs a Uniform Linear Array (ULA) with two omni directional elements. But diversity gain rather than interference mitigation is the goal here. We employ the MMSE algorithm with the aid of the extracted training symbols after the DFT operation to compute the weight vectors. MMSE combining is equivalent to the MRC [15] if there is no interference present. However the benefit if MMSE lies in the fact that apart from spatial diversity gain, the combiner will also be able to cancel interference as long as the inter element separation satisfies the spatial sampling theorem[9]. Each additional user will consume one order of diversity [16]. We assumed that the spacing between the elements is half the carrier wavelength. This small separation works for a large angle spread of the channel. If the angle spread of the channel is narrow, larger separation might be necessary to achieve low correlation between the received signals at the antenna elements [5]. However separation larger than half the carrier wavelength will violate the spatial sampling theorem for a ULA [9] and interference reduction will no longer be possible with this antenna array. Another limiting factor is that the number of antenna elements may become prohibitively large if a large number of interferers are present.

Figure 1: Receive beamforming/diversity combining for WLAN

C. Transmit Diversity combiner for WLAN Before transmitting the OFDM symbols, a 2 branch coding scheme based on the Alamouti code can be employed for achieving transmit diversity. With the Alamouti coding scheme, two consecutive OFDM symbols are transmitted according to Table 2 where T is the OFDM symbol duration and i = 1, 3, 5, ….

Table 2: STC for the WLAN Time

Antenna 1

Antenna 2

iT

x(t ) (i −1)T

xi −1

xi

( i +1)T

− xi∗

xi∗−1

x(t ) iT

Figure 2 shows the block diagram for the STC based transmit diversity. The weights which are the channel coefficients are computed by employing pilot symbol assisted MMSE algorithm. Although in section III E we have discussed the MMSE algorithm in the context of receive antenna array, it can easily be applied to estimate the channel. The channel is assumed to be stationary for two consecutive OFDM symbols. This is a reasonable assumption at a pedestrian velocity of 5 km/h.

Figure 2: STC block diagram We can also apply the Alamouti coding scheme across two consecutive sub-carriers. This code satisfies complex orthogonality both in the frequency and the spatial domain. The coding is performed before the IFFT operation i.e. in the frequency domain. So often this coding scheme is termed as space-frequency coding (SFC) [17]. In order for the SFC scheme to work, the channel must not change across two contiguous sub-carrier i.e. the coherence bandwidth of the channel should be larger than two subcarriers. This assumption is usually true for an indoor channel. MMSE based channel estimation is applied to compute the channel coefficients. Figure 3 shows the block diagram of the SFC scheme applicable for WLAN.

Figure 3: SFC block diagram V. SIMULATION RESULTS Figure 4 shows a typical beampattern when the desired user is at 300 with respect to the array broadside [5] and the interfering user is at 600. The interferer is an 802.11g transmitter identical to the desired user. The average power level of signal received from the interferer is the same as that of the desired user. We have not considered any multipath channel. The only impairment is Additive White Gaussian Noise (AWGN).We can clearly observe the null at

90 120

0

10

-1

10

-2

10

BER

the direction of the interferer and high gain at the direction of the interferer. Figure 5 shows the Bit Error Rate (BER) vs. SNR performance for the aforementioned scenario. The baseline curve is generated for the case of a single antenna receiver. We can clearly observe that at the presence of a co-channel interferer the performance of the single antenna receiver is unacceptable. However the receiver equipped with the beamformer is performing very well. This indicates that the channel can be concurrently used by two different users and there is a potential for bandwidth reuse. Multiple access and channel reservation protocols should therefore be suitably modified to increase the capacity and throughput. WLANs have to share the unlicensed spectrum with many other applications. For example IEEE802.11g and Bluetooth [18] both operate at the unlicensed 2.4 GHz ISM band. Therefore beamforming may facilitate the co-existence of several co-located wireless systems. Also notice that the performance of the beamformer improves if the desired user moves to 00 with respect to the array broadside. Since the separation between the desired user and the interferer has increased, the beamformer can cancel the interference better. The width of the beam depends on the size of the array. As the number of elements increases, the spatial resolution of the array [6] increases and the beampattern becomes narrower. So, in order to discriminate very closely spaced users, we may require larger antenna array. A larger array can cancel more interferers and accommodate higher number of users. However we may be limited by the size of the antenna array that can be employed for a small wireless terminal.

-3

10

-4

10

desired user at 300 desired user at 00

-5

10

single antenna -6

10

0

2

4

6

8

10

12

14

16

18

20

Eb/N0

Figure 5: BER vs. SNR for the beamformer. The interferer is at 600. Figure 6 shows the BER vs. SNR performance for the receive diversity scheme. We employed a vector channel modified from Jake’s Model [19] considering pedestrian motion, as 802.11g WLAN would be used only indoor. Table 3 shows the parameters used for generating the complex coefficients of the vehicular channel model for our study. Table 3: Parameters of the vector channel model Doppler Spread Normalized Antenna Separation Angle Spread of Sub-paths Number of Unresolved Multipath

10 Hz 0.5 60° 10

60

150

30

180

0

330

210 Desired user interferer

300

240 270

Figure 4: Sample beampattern for a 2 element beamformer. The desired user is at 300 and the interferer is at 600.

The baseline curve is generated for the case of a single antenna receiver. We can clearly observe that the performance of the single antenna receiver is inferior compared to the receiver equipped with the diversity combiner. The performance can be further improved by increasing the number of elements. The size of the ULA might be prohibited large for a small wireless terminal and thus has not been considered during the simulations. However antenna array of different geometry and/or application of polarization diversity [6], [9] may make the employment of more elements feasible. Figure 7 provides the BER vs. SNR performance of the transmit diversity schemes. Two sets of weight are calculated for the corresponding OFDM symbols and subcarriers for STC and SFC respectively. These weights are actually channel estimates in their respective domains. These estimated weights are used to equalize the received signal of the corresponding time slots or sub-carriers. Pilot symbol assisted MMSE algorithm is applied to estimate the weights in both cases. The modified Jake’s model described by Table 3 is employed to model the channel from the two transmitter antenna to the single receive antenna. We can observe that the performance of both the STC and the SFC is identical. This is consistent with the findings of [17].The performance of the receive diversity scheme is better because of the array gain. The transmitter power for each

antenna is half of that for the receive diversity to ensure that the transmit diversity schemes employ the same transmitter power as that of the receive diversity scheme. However the slope of the two curves is similar since the obtained diversity order of the Alamouti scheme is equal to applying MRC with two antennas at the receiver [10]. We can observe by comparing the Figures 6 and 7 that the transmit diversity schemes are outperforming the single antenna receiver. 0

10

2 element RX diversity single antenna

[2] [3] [4] [5] [6] [7] [8]

-1

10

[9] -2

10

BER

[10]

-3

10

[11] [12]

-4

10

-5

10

0

2

4

6

8

10

12

14

16

18

20

[13]

Eb/N0

Figure 6: BER vs. SNR for the receive diversity combiner.

[14]

0

10

STC SFC 2 element RX diversity

-1

10

[15] [16] [17]

-2

BER

10

-3

10

[18]

-4

10

[19]

-5

10

0

2

4

6

8

10

12

14

16

18

20

Eb/N0

Figure 7: BER for the Alamouti schemes VI.

CONCLUSIONS

In this paper we have investigated several multi-antenna based techniques that can be applied to improve the performance of an OFDM based WLAN significantly. In view of the increase in performance attained by the spatiotemporal processing, it may be worthwhile to further investigate their application for use in WLAN systems. REFERENCES [1]

IEEE 802.11-1999, “Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications,” 1999.

IEEE 802.11g-2003, “Wireless Medium Access Control (MAC) and Physical Layer (PHY) specifications: Amendment 4: Further Higher Data Rate Extension in the 2.4 GHz Band,” 2003. IEEE 802.11a-1999, “Wireless Medium Access Control (MAC) and Physical Layer (PHY) specifications: High Speed Physical Layer in the 5 GHz Band,” 1999. IEEE 802.11n PAR: Wireless LAN MAC and PHY specifications: Enhancements for Higher Throughputs. A. Paulraj, R. Nabar and F. Gore, Introduction to Space-Time Wireless Communications, Cambridge University Press, 2003. L. C. Godara, Smart Antennas, CRC Press, 2004. S. Haykin, Adaptive Filter Theory, 4/e, Prentice Hall, 2001. L. J. Cimini, Jr., “Analysis and simulation of a digital mobile channel using orthogonal frequency division multiplexing,” IEEE Trans. Commun., vol. 33, pp. 665-675, July 1985. W. L. Stutzman and G.A. Thiele, Antenna Theory and Design. John Wiley & Sons, New York, 1981. S. M. Alamouti “A simple transmit diversity technique for wireless communications,” IEEE Journal on Select. Areas in Commun (JSAC)., Vol. 16, Issue 8, pp.1451 – 1458, Oct. 1998. S. Haykin and M. Moher, Modern Wireless Communications, Prentice Hall, 2004. “TS 25.211 V2.1.0 (1999-06),” a 3GPP Technical Specification document on physical channels and mapping of Transport channels onto physical channels (FDD), 1999. F. Alam, B. L. P. Cheung, R. Mostafa, W. G. Newhall, B. D. Woerner, and J. H. Reed, “Sub-band beamforming for OFDM system in practical channel condition,” Proceedings of IEEE 60th Vehicular Technology Conference, 2004. VTC2004-Fall., Vol. 1, pp. 235-239, Sept. 2004. P. Vandenameele, L.Van der Perre, M. G. E. Engels, B.Gyselinckx, and H.De Man, “A Combined OFDM/SDMA Approach,” IEEE Journal on Select.. Areas in Commun (JSAC), vol. 18, No. 11, pp. 2312- 2321, Novemebr 2000. J.G. Proakis, Digital Communicatons, 4/e, Mcgraw-Hill, 2000. J. H. Winters, J. Salz, and R. D. Gitlin, “The impact of antenna diversity on the capacity of wireless communication systems,” IEEE Trans. Commun., vol. 42, pp. 1740-1751, Feb./Mar./Apr. 1994. K. F. Lee, and D. B. Williams, “A space-frequency transmitter diversity technique for OFDM systems,” Proceedings of IEEE Global Telecommunications Conference, 2000. GLOBECOM '00., Vol. 3, pp. 1473-1477, Sept. 2000. IEEE 802.15.1-2002 “Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Wireless Personal Area Networks (WPANs),” 2002. R. Michael Buehrer, Achilles G. Kogiantis, Shang-chieh Liu, Jiannan Tsai, and Dirck Uptegrove, “Intelligent Antennas for Wireless Communications-Uplink,” Bell Labs Technical Journal, pp. 73-103, July-September 1999.