Enhancement of HIPERLAN/2 Systems using Space-Time Coding Mikael Gidlund Radio Communication Systems Group Department of Signals, Sensors and Systems Royal Institute of Technology (KTH) SE-100 44 Stockholm, Sweden Email: mikael.gidlund@mh.se ABSTRACT The development of broadband wireless communi- cation systems must cope with various performance- limiting challenges that include channel fading as well as size and power limitations at the mobile units. As a promising method dealing with these challenges, space- time coding is effective in supporting reliable, high-data rate transmissions: the major goal in broadband wire- less communications. Space-time coding relies on multi- antenna transmission that are combined with appropri- ate signal processing at the receiver to provide a diver- sity gain. In this paper, we investigate the performance of using space-time coding in HIPERLAN/2 which is a European-standard for high-speed Wireless Local Area Network (WLAN) operating in the 5 GHz frequency band. Software-simulated physical layer performance results are presented and first results show that Packet-Error- Rate (PER) performance is enhanced by almost 4 dB when using space-time coding in the HIPERLAN/2 sys- tem. 1 INTRODUCTION Broadband wireless access to multimedia supporting backbone networks has been rapidly drawing attention toward ubiquitous communications scenario. Recently, a couple of standardizing bodies and research institutes have been actively working to establish high-speed Wire- less Local Area Networks (WLANs) [1], [2]. These stan- dards will be operating in the 5 GHz frequency band. Both the IEEE 802.11a and HIPERLAN/2 is designed to provide high-speed communication, up to 54 Mbit/s, be- tween portable devices attached to an IP, ATM or UMTS backbone network. In this paper we focus on the HIgh PErformance Radio Local Area Network (HIPERLAN/2) which is defined by the ETSI BRAN. HIPERLAN/2 will be capable of supporting multimedia applications and the typical environment is indoors with restricted user mobil- ity. Furthermore, such as a wireless access network shall be able to provide Quality of Service (QoS), including re- quired transfer data rate, delay and blocking error proba- bility similar to what users can expect from a wired LAN. The key to successfully deploying broadband WLAN is to employ a transmission technique to secure a low Packet Error Rate (PER) over the frequency selective fading channels: high-rate transmission incurs multipath delays that can range over several times the clock period even in indoor environments [3]. In the physical layer, HIPER- LAN/2 employs a transmission scheme called Orthogo- nal Frequency Division Multiplexing (OFDM) which has been selected due its excellent performance to combat frequency selective fading in highly dispersive channels and randomizes the burst error caused by the fading chan- nel [4]. A key feature of the physical layer is to pro- vide several modes with different coding and modulation schemes (Table I) which are selected by link adaption [5]. It enables the system to match the physical layer mode to the required radio link quality in order to reach desired QoS [8]. In recent years, space-time coding has gained much attraction as an efficient transmit diversity technique to combat fading in wireless communications and improve the capacity of wireless networks. Space-time coding relies on multi-antenna transmission that are combined with appropriate signal processing at the receiver to pro- vide a diversity gain. For a fixed number of antennas, their decoding complexity at the receiver increases expo- nentially with the transmission rate. To reduce decod- ing complexity, orthogonal space-time block codes with two transmit antennas were first introduced by Alamouti [6] and later generalized to an arbitrary number of trans- mit antennas in [7]. An attractive property of space-time block codes is that maximum-likelihood (ML) decoding can be performed using only linear processing. For com- plex constellations, space-time block coding with two transmit antennas is the only block code that provides full diversity without loss of transmission rate [7]. In this paper we implemented the Alamouti’s space- time coding scheme in a HIPERLAN/2 system to im- prove the system performance. We consider a system with two transmit antennas and one receive antenna. Our results shows that using space-time coding in HIPER- LAN/2 increases the PER performance substantially and we achieve a lower PER. The organization of this paper is as follows: In Section II a short review of HIPERLAN/2 standard are presented and in Section III we briefly de- scribe the space-time coding and signal model. In Section IV we discuss the obtained simulation results and finally in Section V we conclude the work. Table 1: Physical layer modes of HIPERLAN/2 Mode Modulation Code rate PHY bit rate 1 BPSK 1/2 6 Mbps 2 BPSK 3/4 9 Mbps 3 QPSK 1/2 12 Mbps 4 QPSK 3/4 18 Mbps 5 16QAM 9/16 27 Mbps 6 16QAM 3/4 36 Mbps 7 16QAM 3/4 54 Mbps 2 THE HIPERLAN/2 STANDARD The HIPERLAN/2 standard is split into three layers: the Data Link Control (DLC) and Physical (PHY) lay- ers, which are core network independent, and a set of Convergence Layers (CLs), which are network-specific. The technical specifications define a radio access network that is able to operate at rates up to 54 Mbit/s, provides support for multimedia QoS parameters and, through the various CLs, can flexibly interconnect with various wired core networks. The Medium Access Control (MAC) protocol is a part of the DLC layer, and uses a Time Division Multiple Access (TDMA) Time Division Duplex (TDD) approach [8]. A centralized scheduling algorithm, implemented in the Access Point (AP), controls the medium access, de- termining how the data transmission resources provided by the PHY layer are shared between the mobile termi- nals (MT’s) connected to the AP. The MAC mechanism is based around 2 ms frames, within which time slots are assigned for broadcast, DL, UL payload, and resource request transmissions. Preambles are transmitted regu- larly in order to allow accurate Channel State Information (CSI) estimation. The physical layer is based on OFDM as mentioned before. The convergence layers (CL) adapt the core network to the HIPERLAN/2 DLC layer. The CL provides all functions needed for connection set-up and support mobility in the core network. For each sup- ported core network a special CL is designed. Support for packet based networks like Ethernet as well as cell based networks like ATM and UMTS will be available. The convergence layers available at the AP/CC are announced via broadcast. Mobile terminal and AP/CC negotiate one of them during association. 3 SPACE-TIME CODING AND SIGNAL MODEL In order to improve the physical layer of the HIPER- LAN/2 system, we will implement a space-time coding scheme with two transmit antennas. This Space-Time Coding (STC) scheme was first proposed by Alamouti in 1998 and it achieves full diversity for the two trans- mit antennas [6]. This scheme supports maximum likeli- hood detection based on linear processing and can easily Figure 1: Parts of an OFDM system including space-time coding be combined with arbitrary outer coding schemes and re- quires little additional complexity. In figure 1 the parts concerning space-time block code of an OFDM system are depicted. The input symbols to the Space-time block code en- coder are divided into a group of two symbols each; i.e. two OFDM symbols are used to generate one STC code word. At a given symbol period, the two symbols in each group fc1;c2g are transmitted simultaneously from the two antennas nT. From antenna 1 the signal c1 is trans- mitted and from antenna 2, c1 is transmitted. During next symbol period the signal c 2 is transmitted from antenna 2 and c 1 from antenna 1. It is assumed that the OFDM signal of each transmit antenna is transmitted over a slowly multipath Rayleigh fading channel characterized by its time impulse response [9] hi( ;t) = LX l=1 hi;l(t) ( i;l); i = 1;2 (1) where L is the total number of paths, f i;lg are the dif- ferent time delays, and fhi;l(t)g represent the different complex path gains which are modelled as wide sense stationary uncorrelated complex Gaussian processes with normalized total channel power LX l=1 jhi;l(t)j2 = 1; i = 1;2: (2) We define h1 and h2 as the channels from the first and second transmit antennas to the receive antenna nR, re- spectively. Here we assume that both h1 and h2 are con- stant over two consecutive symbol periods. It is further assumed that the antennas are well separated in space such that the transmitted signals pass through a indepen- dent multipath fading process. At the receiver we assume that only one single receiver antenna, and we denote the received signals over two consecutive symbol periods as r1 and r2. The received signals can be written as: r1 = h1c1 +h2c2 +n1 (3) r2 = h1c 2 +h2c 1 +n2 (4) where n1 and n2 represent the AWGN and are mod- elled as iid complex Gaussian random variables with zero mean and power spectral density N0=2 per dimension. We define the received signal vector r = [r1;r 2]T, the noise vector n = [n1;n 2]T , and the code symbol vector c = [c1;c2]T . Then we can rewrite the equations (3) and (4) so they can be represented in matrix form as r = H c+n (5) where the channel matrix H is defined as H = h 1 h2 h 2 h 1 (6) The vector n is a complex Gaussian random vector with zero mean and covariance N0 I. We define C as the set of all symbol pairs c = fc1;c2g. We assume that all symbol pairs are equiprobable, and since the noise vector n is assumed to be multivariate AWGN, then the opti- mum maximum likelihood decoder become ?c = arg min ?c2C jjr H ?cjj (7) Then we can simplify the ML decoding rule in (5) by realizing that the channel matrix H is orthogonal and, hence, H H = I where = jh1j2 +jh2j2. Consider the modified signal vector ?r given by ?r = H r = c+ ?n (8) where ?n = H n. In this case the decoding rule becomes ?c = arg min ?c2C jj?r ?cjj2 (9) Since H is orthogonal, we can easily verify that the noise vector ?nwill have zero mean and covariance N0 I, i.e. the elements of ?n are independent and identically dis- tributed. Hence, it follows immediately that by using this simple linear combining, the decoding rule in equation (7) reduces to two separate, and much simpler, decoding rules for c1 and c2. For the above 2 x 2 space-time block code, only two complex multiplications and one complex addition per symbol are required for decoding. When the receiver uses M receive antennas, the re- ceived signal vector rm at receive antenna m is rm = Hm c+nm (10) where nm is the noise vector and Hm is the channel ma- trix from the two transmit antennas to the mth receive antenna. In this case the optimum ML decoding rule is ?c = arg min ?c2C MX m=1 jj?r ?cjj2 (11) As before, in the case of M receive antennas, the de- coding rule can be further simplified by pre-multiplying the received signal vector rm by H m. In this case, the diversity order provided by this scheme is 2M [10]. c y h n Figure 2: Equivalent channel model for space-time block code and linear combiner at the receiver 3.1 Bit Error Probability of Space-Time Block Codes The transmission of space-time block code matrix together with linear combining at the receiver corre- sponds to transmission over the Single-Input-Single- Output (SISO)-AWGN channel with the per bit SNR b which can be described by an equivalent channel model as depicted in figure 2. The bit error probability (BER) Pb( b) can be calculated using the well known expres- sions for an AWGN channel. The density function of the average per bit SNR b af- ter combining becomes [11] f b( b) = 1 (nTnR 1)! (ij)nTnRb;0 nTnR 1b e b (ij)b;o : (12) However, first we investigate the density function f ( ) of the SNR b after combining, normalized to its expected value b;0. From = PnT i=1 h (ij)nRjh(ij)j2 nTnR (13) it follows the density function f ( ) = (nTnR) nTnR (nTnR 1)! nTnR 1e nTnR (14) The density function f ( ) is depicted in figure 3 for different diversity levels nTnR. It can be observed how the variance of the SNR decreases with increasing diver- sity level and the fading channel is transformed towards a gaussian channel as it is well known from maximal ratio combining. Due to the chosen normalization, the plots in figure 3 describe any diversity scheme with diversity level nT nR, no matter which particular diversity method is applied. Using the SNR density function f b( b) given in (12), we can now calculate the bit error probability Pb of a space-time block code in quasi-static fading with inde- pendent complex Gaussian channel taps from Pb = Z 1 0 Pb( b)f b( b)d b: (15) For BPSK and QPSK with Gray mapping, we obtain Pb( b) = 12 erfc(p b): (16) Figure 3: Probability density function of per bit SNR af- ter combining normalized to its expected value According to [12], there exists the closed form solution for BPSK and QPSK Pb = 12 " 1 b nTnR 1X k=0 2k k 1 2 b 4 k!# (17) for (16), where b = vu ut (ij)b;0 1+ (ij)b;0 = s 1 1+ nTN0E b : (18) For higher order modulation, there exist no closed form solution. However, for high SNR we can use the approx- imation [11] Pb( b) 1log 2 M erfc p b log2 M sin M (19) 4 SIMULATION AND RESULTS In this section, we present some results of our simu- lations. A fully compliant HIPERLAN/2 physical layer simulation has been developed. Each OFDM symbol comprises 48 data-bearing and 4 pilot subcarriers, and modulation and demodulation can be implemented by means of a 64 point Fast Fourier transform (FFT) oper- ation. The sampling rate is set to 20 MHz and the sub- carrier spacing is 0.3125 MHz. Forward Error Control (FEC) is performed by Convolutional Code (CC) of rate 1=2 and constraint length of seven. The further code rates are obtained by puncturing. The multipath radio channel considered in this paper is specified as in [13]. It contains different channel models, representing different environ- ments, with tapped delay lines modelled as Rayleigh or Rician as indicated in Table II. Channel time variance was modelled with a classical Jake’s Doppler spectrum corresponding to a terminal speed of 3 m/s on each tap of the channel impulse response. This corresponds to the maximum Doppler rate = 53:5 Hz at 5350 MHz, which is the highest frequency in the band designate for indoor operation of HIPERLAN/2. In the simulations we have used omni-directional antennas and we assume that the distance between the two antennas are enough separated. It is possible to include two MT antennas a 5 GHz with sufficient low correlation over the subcarriers. Typically an antenna spacing of =2 (i.e. 2.3-3 cm) gives a corre- lation lower than 0.5. It has been shown that signals are close to iid for an antenna separation of one wavelength [14]. Furthermore, we assume that perfect Channel State Information (CSI) is applied [7][15][16]. The PDU error rate (PER) versus C/N has been adopted here as a suitable measure of performance. The PDU train in the HIPERLAN/2 standards contains 54 bytes of data. Figure 4 shows the PER performance of mode 1 (6 Mbps) and mode 2 (9 Mbps) of the HIPER- LAN/2 PHY layer with space-time coding for channel model A. It is clearly shown that the case with space-time coding outperforms the regular case with HIPERLAN/2 without diversity. For mode 1 the difference is almost 4 dB difference in PER. Figure 5 shows the performance of space-time coding for all HIPERLAN/2 modes with a delay of 2 ms between up-link and down-link. We clearly see that PER performance is improved. The improvement is in the range 3-6 dB depending on which mode we con- sider. In figure 6 we have considered 2 Tx antennas and 2 Rx antennas. We clearly se a significant improvement when increasing the number of antennas. If we compare the results obtained here with those obtained in [17], we conclude that using 2 Tx and 2 Rx antenna elements can almost double the system capacity. During simulations we observed that space-time block code is more sensitive to channel time variance than to AWGN on the CSI. If the delay is 6 ms or more between the DL and UL transmissions, the performance improve- ment falls to less than 2 dB, which also was observed in [18]. Nevertheless, given the MAC frame duration of 2 ms and assuming regular UL and DL transmission, we conclude that the performance improvements that space- time block code gives are still significant. Besides the improvement in PER performance, space-time coding is effective in reducing peak power required for the ampli- fier at the AP. Table 2: ETSI BRAN channel models Name rms ds Characteristics Environment A 50ns Rayleigh NLOS B 100ns Rayleigh NLOS C 150ns Rayleigh NLOS D 140ns Rician (K=10dB) LOS E 250ns Rayleigh NLOS 5 CONCLUSION In this paper, we have implemented a simple space- time coding scheme in HIPERLAN/2 and showed by computer simulation that the PER performance signifi- cantly improves. When using space-time coding we gain 3-6 dB compared to a HIPERLAN/2 system with only one antenna. The space-time coding scheme discussed here do not require much more hardware architecture than an ordi- nary system with only one antenna. If we compare with Maximum Ratio Receiver Combining (MRRC) [6] and the total radiated power is to remain the same, the space- time coding scheme has a 3 dB disadvantage because of simultaneous transmission of the two distinct sym- bols from two antennas. If one assume equal radiated power, the space-time coding scheme requires two half- power amplifiers compared to one full power amplifier for MRRC, which can be advantageous for system im- plementation. The key factor that will determine whether or not using space time coding in HIPERLAN/2 is of practical use is whether or not the benefits that space-time coding that they yield outweight the increased cost of the MT. REFERENCES [1] ETSI, “Broadband Radio Access Networks (BRAN); HIgh PErformance Radio Local Area Network (HIPERLAN) type 2; Requirements and architectures for wireless broadband access,” ETSI TR 101 031, V2.2.1, January 1999. [2] IEEE,"Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications IEEE 802.11-a, 1999. [3] H. Hashemi, “The Indoor Radio Propagation Channel,” Proceeding of the IEEE, vol. 8, No. 7, pp. 943-968, 1993. [4] R. V. Nee and R. Prasad, “OFDM for Wireless Multimedia Communications,” Artech House Pub- lishers, December 1999. [5] ETSI, ”Broadband Radio Access Networks (BRAN); HiperLAN/2; Physical (PHY) layer,” ETSI TS 101 475,v1.1.1, April 2000. [6] S. M. Alamouti, “A Simple Transmit Diversity Technique for Wireless Communications,” IEEE JSAC Vol. 16, no. 8, October 1998. [7] V. Tarokh, H. Jafarkhani, and A. R. Calderbank, “ Space-time block codes from orthogonal designs,” IEEE Trans. Inform. Theory, vol. 45, pp. 1456- 1467, July 1999. [8] ETSI, “Broadband Radio Access Networks (BRAN); HiperLAN/2; Data Link Control (DLC) layer,” ETSI TS 101 761-1, V1.1.1, April 2000. [9] W. C. Jakes, Mirowave Mobile Communiations, IEEE Press, 1974. [10] A. Naguib and R. Calderbank, "Space-Time Cod- ing and Signal Processing for High Data Rate Wireless Communication," Wireless Communica- tions and Mobile Computing, Wiley, Vol.1, no. 1, 2001. [11] G. Bauch, J. Hagenauer and N. Sehadri, "Turbo TCM and Transmit Antenna Diversity in Mul- tipath Fading Channels," In Proc. International Symposium on Turbo Codes, Brest, France, Sept. 2000. [12] J. G. Proakis, Digital Communications, 3rd Edi- tion, McGraw-Hill, New York, 1995. [13] J. Medbo and P. Schramm “Channel Models for HiperLAN/2,” ETSI/BRAN document no. 3ERI085B, 1998. [14] D. McNamara, M. Beach, P. Fletcher and P. Karlsson, “Initial Investigation of Multiple- Input Multiple-Output Channels in Indoor En- vironments,” IEEE Benelux Chapter Symposium on Commmunications and Vehicular Technology, Leuven, Belgium, October 2000. [15] V. Tarokh, N. Seshadri and A. R. Calderbank," Space-Time Codes for High Data Rate Wireless Communications: Performance criterion and code construction." IEEE Trans. Inform. Theory, Vol. 44, no.2,pp. 744-765, Mars 1998. [16] J. Guey, M. Fitz, M. Bell and W. Kuo, " Signal De- sign for Transmitter Diversity over Rayleigh Fad- ing Channels," IEEE Trans.Commun.", Vol. 47, no. 4, pp. 527-537, April 1999. [17] A. Doufexi, S. Armour, M. Butler, A.Nix and D. Bull, "A Study of the Performance of HIPER- LAN/2 and IEEE 802.11a Physical Layers," IEEE VTC’01-Spring, Rhodes, Greece, May 6-9 2001. [18] M. Butler, A. Nix, D. Bull and P. Karlsson, "The Performance of HIPERLAN/2 Systems with Mul- tiple Antennas," IEEE VTC’01-Spring, Rhodes, Greece, May 6-9 2001. 0 2 4 6 8 10 12 14 1610?2 10?1 100 C/N [dB] PER PER Performance of H/2 with 2 Tx Antennas STC 6 MbpsH/2 6 Mbps STC 9 MbpsH/2 9 Mbps Figure 4: Performance of HIPERLAN/2 PER vs C/N ra- tio for channel model A. We are using 2 Tx and 1 Rx antenna 1 2 3 4 5 6 70 5 10 15 20 25 30 35 Different modes UL C/N in [dB] required for PER=10 ?2 H/2 with Space Time CodingHIPERLAN/2 Figure 5: UL C/N required to achieve a PER of 10 2 for different HIPERLAN/2 modes for both space-time cod- ing and without any diversity. We are using 2 Tx and 1 Rx antenna 1 2 3 4 5 6 70 5 10 15 20 25 30 Different modes UL C/N in [dB] required for PER=10 ?2 PER Performance of H/2 with 2 Tx antennas and 2 Rx antennas H/2 with Space Time CodingHIPERLAN/2 Figure 6: UL C/N required to achieve a PER of 10 2 for different HIPERLAN/2 modes for both space-time cod- ing and without any diversity. We are using 2 Tx and 2 antennas