Simulink Imlementation of a CDMA Smart Antenna System MOSTAFA HEFNAWI Deartment of Electrical and Comuter Engineering Royal Military College of Canada Kingston, Ontario, K7K 7B4 CANADA Abstract: - The main objective of this aer is to develo a new software testbed to aid our research on various smart antenna (SA) technologies roosed for future generation wireless communication systems. These models can be used as a latform to test the use of DSPs/FPGs in new generation CDMA-SA systems and will allow to evaluate the array attern (AP) and the Bit Error Rate (BER) erformance over a range of communication channel conditions, e.g., whether the fading is flat or frequency selective, and how much interference the user is getting from other users. This goal is achieved through modeling and simulation using the raid rototying environment based on Mathworks Simulink [5]. BER and AP erformances are evaluated for different models under various channel conditions to demonstrate the merits of the combined CDMA-SA system. Key-Words: - Smart antenna, Beamforming, DS-CDMA, Simulink, Raid rototying. Introduction Multiath fading and system caacity limitations are major hurdles in the delivery of 3G data rates and services as envisioned in IMT-2000 []-[2]. Limits on RF sectrum allocation means that there is only a finite bandwidth in which a wireless system can oerate, as such, increasing data rates and system caacity by increasing bandwidth is not an otion. Further comlicating the issue, CDMA techniques require strict ower management and using a higher transmitting ower, as a method of imroving link erformance is not a viable otion. Recently there has been much interest in CDMA schemes that use Smart Antenna (SA) technology in 3G high-seed alications [3]-[4]. What makes this SA concet such an exciting rosect is the ossibility of offering more efficient ower control and reduced transmitting ower which translates into increased coverage, increased system caacities and suort for higher data rates. These SA systems can automatically track the desired signal and null the interfering signal by adjusting the gain and hase of weights, which control the outut array element. Therefore, with combined CDMA-SA techniques, Multile Access Interference (MAI) and multiath fading issues faced by CDMA systems are mitigated by the higher recetion sensitivity and interference cancellation benefits of SA techniques, without imacting on the strict requirement for ower control in CDMA systems. The main objective of this aer is to develo a new software testbed to aid our research on various smart antenna (SA) technologies roosed for future generation wireless CDMA systems. This goal is achieved through modeling and simulation using the raid rototying environment based on Mathworks Simulink [5]. Simulink rovides a block diagram interface that is built on core MATLAB numeric, grahic, and rogramming functionalities. Simulink was chosen rimarily for its collection of alication-secific blocks that suort multile design discilines, including 50 hysical-layer communication system blocks for broadband wireless, broadcasting and satellite systems; and more secific to this work, an excellent real world mobile wideband CDMA model in the form of a comlete UMTS WCDMA system oerating over a multiath fading channel. However, the major design effort revolved around this study is the creation of new blocks that imlement the different modules of a smart antenna system, namely, the antenna array and the adative algorithm blocks. These new created subsystems allow extending the caability of Simulink beyond the canned functionality of generic blocksets. 2 System Model A ractical layout for imlementation of CDMA-SA system using a P-element linear antenna array is deicted in Fig.. The user data are transmitted using MPSK modulation and sread using direct sequence CDMA (DS-CDMA) technique. Each user in the CDMA system is assigned a unique seudorandom code sequence. The signals are received by the antenna array, desread and multilied by the adatation comlex weight.
Desired user Sread MPSK Data x x M 2 P x w w 2 w 3 w 4 Fig. Block diagram of a CDMA-SA System The weighted signals are then summed to form the array outut. The receiver is equied with a P- element linear antenna array. The received baseband signal at the -th element of the antenna array outut (before desreading oeration) is given by K L x ( t) b ( t τ ) s ( t τ ) a α + n ( ) () = k = l = k k, l k k, l k, l k, l t where L is the number of aths for each user, b k (m) is the k-th user s data symbol, α k, l is the comlex fading coefficient for the l-th ath of the k- th user, s k (t) is the sreading waveform of the k-th user, τ k, l is the time delay of the l-th ath of the k- th user. n (t) is an additive White Gaussian Noise (AWGN). a k, l is the -th entity of the k-th user s array resonse vector in the iminging direction θ of the l-th ath, k,l a k, l = e jπ sinθk, l... e jπ ( P )sinθk, l The k-th user s sreading waveform is k N n= k Τ (2) s ( t) = c ( n) h( t) (3) N where ck ( n) is the k-th user's sreading code, N n= is the rocessing gain and h (t) is a rectangular chi ulse in the interval [ 0, T c]. In order to ick out the desired user s signal, a digital matched filter containing the PN code c is alied to each element of the array to rovide the samled outut y (m). For simlicity, we assume that the aths from each mobile reach the antenna array with the same Direction Of Arrival (DOA), i.e. θ = θ k, l k. The beamformer needs to rovide the desired user (k=) with a beamforming weight vector to adjust the gain and the hase of the array to form the beam towards θ, w θ = [ w w w ] Τ 2... P (4) The antenna array outut can then be exressed as H y(m) = wθ y (m) (5) Where y( m) = [ y ( m) 2 y ( m)... P y ( m)] (6) 3 Simulink Imlementation of CDMA-SA Fig.2 shows the CDMA-SA Simulink model created in this study. The model can be divided into five main modules: Transmitter, Channel, Receiver, Interfering Users and Control/Dislay. Moreover, Given that the resonse of a bandass system to a bandass signal is identical to the resonse of the equivalent baseband system to the equivalent baseband signal, all modeling in this aer was done at the baseband because modeling at the baseband level is much simler and faster than modeling at the bandass level.
Fig. 2 Simulink Model of CDMA-SA System 3. Transmitter Module At the transmitter of each user, a random data was generated using MPSK modulation. The information was then sread and sent over the channel. As one of the elements of this study is the effect of MAI as it relates to the suort of multile rate services, OVSF codes were used in the CDMA model to sread user information and identify individual users across the communications link. This was achieved by using the Simulink OVSF block and assigning to each user a unique sreading code, Fig. 3. E reresenting the number of users transmitting simultaneously, Fig. 4. Fig. 4 Enable block 3.2 Channel Module The Channel Module is based on the Simulink Rayleigh fading and the additive white Gaussian Noise (AWGN) blocks. Fig. 3 DS-CDMA transmitter in Simulink To simulate multile users on the system, an interfering user module was created. The combined use of the Simulink Enable and Constant blocks defines and regulates the number of simulated users accessing the system. Adding an Enable block to a subsystem that executes only while the inut received at the Enable ort is greater than zero. Thirty-one users were modeled by defining a vector Fig. 5 Simulink Rayleigh fading and AWGN channel The Simulink Rayleigh fading channel block arameter field shown in Fig. 5 enables the maniulation of Maximum Doler Shift, the Samle Time, the Delay Vector, the Gain and the Initial Seed arameters. Together these blocks were used to simulate a wireless mobile channel found in a traditional urban environment.
Fig. 6 Simulink Rayleigh Fading Channel Block By defining the number of reflected multiath comonents and their relative delays and owers it was ossible to define whether the channel is undergoing flat or frequency selective fading. Fig. 7 Simulink imlementation of a linear antenna array 3.3 Receiver Module Based on the oeration of an SA system as described in the introduction the beamforming comonent was imlemented by creating two subsystems, an antenna array subsystem and an adative rocessing subsystem. Antenna Array Imlementation A Simulink Comlex Phase Shift block was used to roduce hase shifted versions of the signal that reresented the coies of the signal that would be incident on each element at the receiver, as shown in Fig. 7. It was then ossible to define the iminging angles of each signal from the individual users transmitting across the channel from the ersective of each receiver array element by using Simulink Go To blocks. Fig. 8 SA adative algorithm Block Adative Processing Imlementation The adative rocessing comonent of the SA technique was imlemented using the Simulink Least Mean-Square (LMS) blocks, Fig. 8. The LMS algorithm was chosen for its simlicity and the fact that it could sufficiently model the adative rocessing comonent of the SA system for the uroses of this study. Dislay and Control Modules The two-erformance criterions used to gauge the system erformance are the Bit-Error-Rate (BER) and the adative antenna array attern (AP). The Control/Dislay Module consists of the Set Model Parameters, the Array Pattern Plot, the BER vs. Eb/No Plot, and the BER Results Dislay. The arameters used to define the model simulation scenarios, i.e. those defined in the Set Model Parameter block are summarized in Table.
PARAMETER PG Delay Vector Gain Vector (db) Doler (Hz) Table Model Parameters Descrition Flat Fading Frequency Selective rocessing gain 64, 28, 64, 28, 256 256 Relative delay between the signal of interest and the delayed version [0 Tc] [0 0Tc] arriving at the receiver where Tc = Chi eriod. The relative gain that is assigned to the signal of interest [0 5] [0 5] and the delayed version. Positive scalar reresenting the maximum Doler 0 0 shift. 4 Simulation Results For this study, the following assumtions were made: ) each user was given unique user code; 2) there is a erfect ower control; 3) a 2-ray fading channel; and 3) a slow fading channel, i.e. the Doler effect was minimal. 4. Simulation results of CDMA model without smart antenna Fig. 9 shows the BER erformance of the CDMA model (without SA) for different rocessing gains (PGs) over a flat and a frequency selective fading channels. 0 0 C not result in any significant imrovement in erformance over a frequency selective channel. Moreover, it is noted that in the frequency selective case there was an overall degradation in link erformance caused by Inter-Symbol Interference (ISI) distortions. This validation results accurately reflect the exected behavior of a CDMA system oerating in a multiath fading environment. Validation of the ability of the CDMA model (without SA) to simulate the effects of MAI is shown in Fig. 0. It is noted that increasing the number of interfering users results in a degradation in the system erformance. This imairment in the system erformance is mainly due to the loss of orthogonality caused by the multiath fading channel. These results validate the model s ability to simulate the MAI effects. 0 0 0 - B E 0-2 R 0-3 CDMA Bit Error Probability Plot 2 Users 4 Users 22 Users 30 Users 0 - Frequency Selective Fading PG=64 PG=28 PG=256 0-4 0 2 4 6 8 0 2 4 6 8 20 BER 0-2 Flat Fading Fig. 0 CDMA Model BER Performance Plot for Multile Users 0-3 0-4 0 2 4 6 8 0 2 Fig. 9 BER erformance of CDMA system over flat and frequency selective fading channels It is shown that increasing the PG results in a better erformance across the flat fading channel and does 4.2 Simulation Results of the combined CDMA-SA Model The CDMA-SA model was run using the arameters defined in Table 2. Fig. and Fig. 2 show array atterns with signal s DOA of 20 and 40 degrees, resectively. For both DOAs it is shown that the main lobe is oriented toward the desired signal s DOA and nulls have been created towards interferers DOAs. In other words, the adative antenna array is able to track the signals of interest and reduce the effect of multiath fading and interferences.
Table 2 PARAMETER Antenna Array Elements 6 PG 64 Delay Vector CDMA-SA Model Parameters VALUE [0 Tc/0] Gain Vector (db) [0 5] Doler (Hz) 0 Desired User DOA (degrees) Interfering Users DOA Vector (degrees) 20 o and 40 o [30-60 -50-40 -30-20 -0 0 0] 0 20 for the BER and 0 db for the array attern lots 0 0 BER 0 - CDMA CDMA-SA 0 Gain (db) -0-20 -30-40 -50-60 -60-40 -20 0 20 40 60 80 00 Angle [Deg.] Fig. Array atterns for signal s DOA=20 deg. Gain (db) 0-0 -20-30 -40-50 -60-70 -80-60 -40-20 0 20 40 60 80 00 Angle [Deg.] Fig. 2 Array atterns for signal s DOA=40 deg. On the other hand, Fig. 3 shows a BER erformance comarison of the CDMA-SA and the CDMA only systems under MAI condition with 30 interferers. It is shown that the combined CDMA- SA technique, when comared with the CDMA only technique, erformed better in environments with an SNR <20dB. However, as the SNR increases the erformance gains fell to that delivered by the CDMA technique. 0-2 0 2 4 6 8 0 2 4 6 8 20 Fig. 3 BER Performance Comarison of CDMA- SA and CDMA systems with 30 MAI 5 Conclusion The main goals of this aer were to imlement a CDMA and CDMA-SA wireless schemes in a Simulink environment and use these models to evaluate their erformance. The main urose of the CDMA model was to rovide a baseline reference for erformance and system caacity so that an assessment of the effectiveness of the smart antenna technique could be made. Through simulation it was shown that, in environments with an SNR <20dB, the combined CDMA-SA technique rovided suerior erformance at higher system loads, roviding substantial imrovement in link erformance and system caacities, above that ossible with the traditional CDMA technique. References: [] M.R. Karim, and Mohsen Sarraf, W-CDMA and cdma2000 for 3G Mobile Networks, McGraw-Hill, 2002. [2] Tero Ojanerä and Ramjee Prasan, Wideband CDMA for Third Generation Mobile Communications, Artech House998. [3] J. Liberti and T.S. Raarort, Smart Antennas for Wireless Communications: IS- 95 & 3G CDMA, Englewood Cliffs, NJ, Prentice Hall, 999. [4] M. Hefnawi, G. Y. Delisle, Smart Antenna System for Wideband CDMA Signals, IEEE International Symosium on Antenna and Proagation, July 200,. 22-25. [5] The MathWorks Inc., htt://www.mathworks.com.