Multiple Antenna Technologies
Manar Mohaisen | YuPeng Wang | KyungHi Chang
The Graduate School of Information Technology and Telecommunications
INHA University
ABSTRACT the receiver. Alamouti code is considered as the simplest transmit diversity scheme while the receive diversity includes maximum ratio, equal gain and selection combining Multiple antenna technologies have methods. Recently, cooperative received high attention in the last few communication was deeply investigated as a decades for their capabilities to improve the mean of increasing the communication overall system performance. Multiple-input reliability by not only considering the multiple-output systems include a variety of mobile station as user but also as a base techniques capable of not only increase the station (or relay station). The idea behind reliability of the communication but also multiple antenna diversity is to supply the impressively boost the channel capacity. In receiver by multiple versions of the same addition, smart antenna systems can increase signal transmitted via independent channels. the link quality and lead to appreciable On the other hand, multiple antenna interference reduction. systems can tremendously increase the channel capacity by sending independent signals from different transmit antennas. I. Introduction BLAST spatial multiplexing schemes are a good example of such category of multiple Multiple antennas technologies proposed antenna technologies that boost the channel for communications systems have gained capacity. much attention in the last few years because In addition, smart antenna technique can of the huge gain they can introduce in the significantly increase the data rate and communication reliability and the channel improve the quality of wireless transmission, capacity levels. Furthermore, multiple which is limited by interference, local antenna systems can have a big contribution scattering and multipath propagation. to reduce the interference both in the uplink Through shaping the antenna radiation and the downlink by employing smart pattern and adaptively adjusting the antenna antenna technology. weight vector, smart antennas improve the To increase the reliability of the communication link quality by increasing communication systems, multiple antennas the received signal power and suppressing can be installed at the transmitter or/and at the interference.
Fig. 1. Multiple antenna technologies.
Besides, on-line calibration technique is II. MIMO Diversity also adopted to correct the errors due to the distortions and nonlinearity of the radio In communication systems, we have to frequency components in the antenna array increase the reliability of the communication system. operation between transmitter and receiver while maintaining a high spectral efficiency. Fig. 1 summarizes the different multiple The ultimate solution relies in the use of antenna technologies and gives some diversity, which can be viewed as a form of examples of these technologies. redundancy [1]. There are many diversity techniques that can be applied to This paper is organized as follows: in communication systems; we mention herein section II we present multiple antenna time diversity, frequency diversity, and diversity schemes employed at the spatial diversity or any combination of these transmitter or/and at the receiver. Spatial three diversities. In time diversity, the same multiplexing presented by BLAST schemes information-bearing signal is transmitted in is detailed in section III. Section IV is different time slots where a good gain can be dedicated to some advanced multiple input achieved when the duration between the two multiple-output (MIMO) systems including slots, in which the same symbol is multi-user MIMO and cooperative transmitted, is greater than the coherence communications. While techniques related time of the channel. In frequency diversity, to the smart antennas such as phased the same information-bearing signal is antenna array, switched beam antenna array, transmitted on different subcarriers where a and adaptive antenna array are described in good diversity gain can be achieved when Section V. Finally, we conclude in Section the separation between subcarriers is greater VI. than the coherence bandwidth. Finally, in spatial diversity, the same (MS) receiver complexity while improving information-bearing signal is transmitted the detection performance. or received via different antennas where The pioneering work in the transmit the maximum gain can be achieved when diversity was done by Alamouti where he the fading occurring in the channel is proposed his famous 2×1 space-time code. independent (or low correlated). In the Alamouti scheme achieves diversity gain receiver, diversity gain can be achieved by while requiring only a linear decoder. combining the redundant signals arriving Later on, Tarokh et al. proposed a via independent (or lowly correlated) generalized theory of the complex channels. orthogonal space-time codes. Based on Tarokh work, more than two antennas can Fig. 2 shows some possible combinations be used and the code rate can be fractional. of transmit diversity which can be In the following we present the two achieved when employing multiple different types of space-time codes. transmit antennas. 2.1.1 Complex Orthogonal Space- In the following section, we present some Time Codes famous space-time block codes applied at the transmitter side. We present also the For this type of space time codes, the combining techniques used when different following conditions must be satisfied versions of the information-bearing signal are received. Finally, we present a scheme Square transmission matrix (number that includes transmit and receive of transmit antennas Nt equal to diversities. number of used time slots m) A unity code rate (number of used
time slots m equals to number of 2.1 Space Diversity at the transmitted symbols l) Transmit Side Orthogonality of the transmission matrix in the time and space The basic idea of the use of transmit domains ( SSHH =S S ) where SH is diversity is to reduce the mobile station the conjugate transpose of S .
Fig. 2. Transmit diversity. ⎡ s12s ⎤ S= ⎢ **⎥ ⎣-s21s ⎦
Fig. 3. Alamouti scheme example with QPSK modulation.
As said before, the simplest complex At the receiver the following signals are orthogonal space-time code is the received (with applying the complex
Alamouti code which uses two transmit conjugate to the received signal at t2) antennas and one receive antenna. Furthermore, Alamouti scheme requires ⎡ yhhsn⎤⎡ ⎤⎡⎤⎡⎤ 11211=+ (1) that the fading channel envelope remains ⎢ ***⎥⎢ ⎥⎢⎥⎢⎥ * ⎣yhhsn22122⎦⎣− ⎦⎣⎦⎣⎦ constant over two time slots. The linear combiner multiplies the Fig. 3 shows an example of the encoding received symbols by the Hermitian process of Alamouti scheme with QPSK transpose of the channel matrix (for modulation [2], [3]. simplicity, we consider that channel is Fig. 4 shows the receiver structure used perfectly estimated). The output of the for decoding the combined received linear combiner is then given by symbols. ⎡x ⎤⎡⎤⎡⎤sw 111=+⎡⎤hh22 + (2) ⎢x ⎥⎢⎥⎢⎥⎣⎦12sw ⎣ 222⎦⎣⎦⎣⎦
Maximum-Likelihood (ML) decoder is then applied to get the transmitted symbols. As one can see, the simplicity of the receiver is due to the spatio-temporal orthogonality of the transmission matrix.
A complex orthogonal space-time code hˆ 1 using 4 or 8 antennas was proposed by ˆ Tarokh et al. in [4]. h2
2.1.2 Generalized Complex Orthogonal Space-Time Codes sˆ sˆ 1 2 The search for space-time codes with more Fig. 4. Alamouti code receiver. than two antennas was started by Tarokh, Jafarkhani, and Calderbank. Their work has built the basis for a theory of frequency-Time domains. These coding generalized complex orthogonal designs. schemes are thus known as ST, SF, and Generalized complex orthogonal designs STF coding, respectively [7]. are distinguished from Alamouti code by the following 2.1.3 Cyclic Delay Diversity (CDD)
A non-square transmission matrix CDD can be considered as a very simple (number of used time slots ≠ number transmit diversity scheme. CDD can of Tx antennas) achieve transmit artificial frequency A fractional code rate (number of diversity by selecting appropriate transmit transmitted symbols < number of delays. In this method, multiuser diversity, used time slots) obtained by scheduling based on Orthogonality of the transmission frequency domain channel response, can matrix is only guaranteed in the time be improved by adjusting the delay spread sense. (at the transmitter) which is done by controlling the delay values dependent on As a consequence of these characteristics, the channel condition [8], [9]. the spectral efficiency is reduced and the number of time slots over which the 2.2 Space Diversity at the channel should be constant is increased. Receive Side
The transmission matrix of a generalized In space diversity at the receive side, complex space-time code with 3 antennas, multiple antennas are used in the receiver 4 transmitted symbols and 8 used time with sufficient spacing between antennas slots is given by [5] in such a way mutual correlation between antennas is reduced and as consequence
⎡⎤s13s 2 s diversity gain is increased [10]. To get ⎢⎥−−s s s diversity gain at the receiver, received ⎢⎥241 signals from different antennas are ⎢⎥−s31s 4 s ⎢⎥ combined. There are four combining −s −s s G = ⎢⎥423 (3) methods, namely, select combining (SC), 3 ⎢⎥*** s13s 2 s maximal-ratio combining (MRC), equal- ⎢⎥*** ⎢⎥−−s 24s1 s gain combining (EGC), and square-law ⎢⎥−s **s * s combining. The first three schemes are ⎢⎥314 linear while the last requires a non-linear ⎢⎥−s **−s * s ⎣⎦423 receiver.
In the literature, more research was done Fig. 5 shows a simplified block diagram to increase the rate of the space-time codes. of the linear combining schemes which For more details refer to [6]. differ in the weighting vector w. In SC, the Table 1 summarizes the difference signal at the branch with maximum signal between Alamouti and space-time code to noise ratio (SNR) is selected and other characterized by the transmission matrix received signals are discarded.
G3. The weighting vector w = (w1, w2, …, th These shown coding schemes can be wM) is the N column of the identity th transmitted in the space-time domains, matrix of size M where the N branch has space-frequency domains or in space- the maximum SNR.
Table 1. Comparison between Alamouti and generalized complex space-time codes.
Space-time Number of Number of Number of Orthogonality Rate = l/m code Tx antennas transmitted used time of Tx matrix symbols, l slots, m S 2 2 2 Spatio- 1 temporal sense
G3 3 4 8 Only temporal 1/2 sense Table 2. Comparison between diversity combining schemes.
Scheme Requiring CSI Outage Probability Application F(x)
SC No M No constraints −xy/ c ⎣⎡1−e ⎦⎤ k MRC Yes M −1 No constraints −xy/ 1 ⎛⎞x 1−e c ∑ ⎜⎟ k =0 ky!⎝⎠c EGC Yes No closed form for M > 2 No constraints SLC No - FSK or DS-CDMA frequency-shift keying (FSK) or direct- As one can see, the SC scheme does not sequence CDMA [5], [11]. require any channel information except Table 2 summarizes a comparison that of SNR. On the other hand, MRC between combining schemes. It is known and EGC schemes require the channel that, from a performance point of view, state information (CSI) or a part of it MRC is optimum and gives the best (channel envelope, phase, delay). MRC performance among the pre-described scheme weights the received signals combining schemes. according to their reliability; a more w1 reliable signal has a high weight while a y1 less reliable signal has a small weight. x
Also, the channel phase distortion is y2 x yc compensated. Finally, signals are aligned Linear . . then combined. On the other hand, EGC . . combiner . w . scheme can be viewed as a simplified . 2 . y version of MRC where signals are M x weighted equally (i.e. the weighting vector wM w = [1, 1, …,1]M) then aligned before being combined coherently. In practice, the phase at different branches can’t be Fig. 5. Simple block diagram of linear combining schemes. often estimated. So, EGC and MRC can’t be employed. In such situations, square 2.3 Combined Transmit/Receive law combining (SLC) can be applied to Diversity obtain spatial diversity without requiring phase estimation. Unlike linear combining Spatial diversity schemes explained in schemes, SLC scheme can only be applied the previous two sections can be combined to modulation schemes which preserve together to achieve diversity at both some sort of orthogonality including receive and transmit sides. An example of such a hybrid spatial diversity scheme is the 2×2 Alamouti/MRC MIMO system 3.1 Diagonal-BLAST [12]. D-BLAST was originally proposed by III. Spatial Multiplexing Foschini [13]. In D-BLAST, the symbols to be transmitted are arranged on the As shown in the previous section, diagonals of the space-time transmission MIMO diversity can be used in the matrix where elements under the diagonal transmitter or the receiver sides or in both are padded with zeros. Fig. 6-a depicts the to increase the reliability of the structure of the D-BLAST transmitter for communication. In this section we talk four transmit antennas. At first, the bit about spatial multiplexing schemes which stream is de-multiplexed into four parallel are for goal to increase the channel streams which are encoded and modulated capacity. independently. Encoded-modulated The most known spatial multiplexing streams are cycled over time. Equation (4) schemes are the BLAST family which is an example of the transmission matrix includes Vertical-BLAST, Diagonal- when using four transmit antennas. BLAST, and Turbo-BLAST. The acronym BLAST stands for “Bell Laboratories Layered Space-Time”.
⎡⎤ss1,1 1,2 s 1,3 sss1,4" 1,KK− 1 1, 000 ⎢⎥ 0 ssssss" 00 S = ⎢⎥2,1 2,3 2,4 2,KK−− 2 2, 1 2, K (4) ⎢⎥00s sssss3,2" 3,KKKK−−− 3 3, 2 3, 1 3, 0 ⎢⎥3,1 ⎣⎦000ssssss4,1" 4,44,3KKKKK−−−− 4,24,1 4,
Fig. 6. Transmitter block diagrams for BLAST family using four transmit antennas.
Table 3. MIMO systems diversity orders. MIMO Configuration Diversity order
STBC Nt×Nr BLAST Nr - Nt + 1
The first diagonal of S is transmitted via The structure of the RLST encoder, with the first antenna; the second diagonal is periodical cyclic space-time interleaving is transmitted via antenna 2, and so on. depicted in Fig. 6-d. For optimal performance of the RLST code, the receiver should employ the maximum a 3.2 Vertical-BLAST posteriori probability (MAP) decoding A simplified version of D-BLAST was algorithm. Nevertheless, the complexity of proposed by Wolniansky known as the MAP decoding algorithm is very high Vertical-BLAST or V-BLAST [14]. In V- (increases exponentially with Nt). To BLAST, incoming data stream is de- decrease the complexity of the receiver, multiplexed into Nt streams each of which the near-optimal turbo-like receiver can be is encoded and modulated independently used. This near-optimal turbo-like receiver and sent on an antenna of its own. V- is known as iterative detection and BLAST high-level diagram is depicted in decoding (IDD) receiver. Fig. 6-b where four antennas are used at Before going further, we list in Table 3 a the transmit side. Compared to D-BLAST, comparison between diversity order of the V-BLAST does not include cycling over different space-time coding and the time, the complexity is significantly BLAST family schemes. reduced. In addition, unlike D-BLAST, V- BLAST does not include any space-time IV. Advanced Topics wastage. At the receiver, transmitted symbols can be decoded using ordered 4.1 Single and Multi CodeWord serial interference-cancellation (OSIC) MIMO detector. For the OSIC to work properly, In single codeword (SCW) MIMO, an the number of receive antennas Nr must be at least as large as the number of transmit encoded packet is distributed across many antennas. streams to form the MIMO transmission. Feedback is used to control the rank of the MIMO transmission (number of streams 3.3 Turbo-BLAST used) as well as the overall rate of Turbo-BLAST was first described by transmission. In multiple codeword Sellathurai and Haykin [15]. The Turbo- MIMO, several separately encoded BLAST transmitter structure is depicted in packets are transmitted independently over Fig. 6-c. The data stream bits are firstly the multiple streams. Here the rate of each demultiplexed into Nt parallel streams stream can be controlled with feedback which are encoded independently using [16] and [17]. the block encoder (outer encoder) (i.e. channel coding). The output streams of the 4.2 Single-User MIMO and outer encoder are interleaved Multi-User MIMO independently and passed to the inner In single-user MIMO, already explained encoder. The mission of the outer encoder techniques in previous sections are used is to achieve random-layered space-time where the channel capacity grows linearly (RLST) coding. with min(Nt, Nr) [18]. For multi-user MIMO, which is of high station its own data bits and some of other interest research topic, it was shown that mobiles (sometimes called partner) for Nt transmitting antennas (at the base information bits. Fig. 7 shows a station) and Nr users, the same overall cooperative cellular system where for capacity can be achieved. This later work simplicity we consider three cooperative was encouraged by applying dirty paper users and one base station [21]. As coding [19] where results showed that if depicted in Fig. 7, user 1 cooperates with the transmitter knows the interfering users 2 and 3 to send its own information. signal, then the channel capacity will not As a result, the overall cooperative system be affected by the presence of the can be seen as virtual-MIMO (V-MIMO) interference [20]. On the other hand, and in the above example it is 3×NBS multi-user MIMO can integrate MIMO system (for the uplink) where NBS beamforming to apply spatial division is the base station’s number of receive multiple access (SDMA). antennas. Users 2 and 3 can simply amplify and forward user 1 received 4.3 Cooperative Communication and information or detect and forward [22]. Virtual MIMO Another method of cooperation is the In cooperative communication, a mobile coded cooperation where different coded can act as both a user and relay. As portions are sent via different fading consequence, mobile sends to the base channels [23].
Fig. 7. Cooperative communication and virtual-MIMO. 4.4 Pre-Coded MIMO with Rank and coding schemes per antenna. Fig. (8) Adaptation shows a general PARC transmitter structure with 4 transmit antennas [24]. 4.4.1 Per Antenna Rate Control (PARC) 4.4.2 Per Group Rate Control (PGRC) PARC can be considered as a closed-loop MIMO system where transmitter uses In PARC, a CQI feedback is necessary for channel quality indication (CQI) fed by each transmit antenna. This increases the the receiver to select the best modulation uplink overhead.
Fig. 8. Per Antenna Rate Control (PARC).
To solve this problem, PGRC is used wireless transmission, which is limited by where a feedback is required per group of interference, local scattering and multipath antennas. This reduces the feedback propagation [26], [27]. Smart antennas information while maintaining almost the offer the following main applications in same performance of PARC [24]. high data-rate wireless communication systems [28], [29]: 4.4.3 Per User Unitary Rate Control (PU2RC) Spatial Diversity Co-channel interference reduction 2 PU RC is a multi-user closed-loop MIMO Angle reuse or space division system. Each user feeds back the CQI to multiple access (SDMA) the base station. The base station uses the Spatial multiplexing CQI to determine the modulation and coding schemes per user. In addition, base Smart antenna system can be station can apply unitary pre-coding and categorized into three main groups: adaptively select the number of transmit Phased antenna array system, switched antennas (rank adaption) [25]. beam systems, and adaptive antenna array system. To match the characteristics in V. Smart Antennas each radio frequency chain of the transmitter and receiver, on-line 5.1 Introduction calibration is required in smart antenna systems. On-line calibration technique Smart antenna was born in the early can compensate the errors such as the 1990s when well developed adaptive distortions of radio frequency antenna arrays originate from Radar components due to small environment system. Later, Smart antenna technique is changes, the nonlinear characteristics of applied in wireless communications mixer, amplifier and attenuator, I/Q system. Recently, Smart antenna technique imbalance errors, etc. has been proposed as a promising solution to the future generations of wireless 5.2 Phased Antenna Array System communication systems, such as the Fourth-Generation mobile communication Phased antenna array is a group of systems, broadband wireless access antennas in which the relative phases of networks, where a wide variety of services the respective signals feeding the antennas through reliable high-data rate wireless are varied in such a way that the effective channels are expected. Smart antenna radiation pattern of the array is reinforced technique can significantly increase the in a desired direction and suppressed in data rate and improve the quality of undesired directions. Phased antenna array system is usually utilized in radio frequency (RF) or intermediate frequency (IF) with the system central frequency larger than 10 GHz, such as satellite communication system [30]. There are two main different types of phased arrays, also called beamformers. There are time domain beamformers and frequency domain beamformers. Fig.9. Functional block diagram of switched beam system. 5.3 Switched Beam System
The switched beam method is considered as an extension of the current sectorization scheme. In the switched beam approach, the sector coverage is achieved by multiple predetermined fixed beam patterns with the greater gain placed in the centre of a beam [30]. When a mobile user is in the vicinity of a beam, then the signals at the output ports will be given as in (5). This enables the switched beam system to select the signal from the output port corresponding to that beam. As the mobile moves to the coverage of another Fig.10. Produced antenna pattern of beam during the call, the system monitors switched beam system with 4 antennas. the signal strength and switches to other Switched beam systems can offer output ports as required. A basic switched several advantages, including beam antenna architecture is shown in Fig. 9. And Fig. 10 illustrates the produced Low complexity and cost. Since antenna pattern with 4 antennas. switched beam system only requires a beamforming network, RF switches, L (5) and simple control logic, they are ytiiillil()= stGt () ()∑ ItG () (θ ) l=1 relatively easy and cheap to implement. where yi(t) is the total signal appearing at Moderate interaction with base port i, si(t) is the signal source, Il(t) is the station receivers. In practice, interfering signal source located at switched beam system can simply arbitrary angles θl, Gi is the transfer replace conventional sector antennas function between signal source along the without requiring significant main beams and their corresponding modifications to the radio base output ports, Gli is the transfer function station antenna interface or the between interference signal l and port i. baseband algorithms implemented at the receiver. Coverage extension. The antenna array aperture gain will boost the link budget, which could be translated to a Transmit beamforming provides a coverage extension. powerful method for increasing downlink capacity [32]-[35]. The idea of TxBF is 5.4 Adaptive Antenna Array similar to the pre-coded MIMO technique System (AAA) but with different strategies to calculate the transmit weight vector. TxBF adjusts Adaptive antennas date back to 1959. the antenna main lobe towards to the The original work was attributed to L. C. desired user and reduce the interference to Van Atta’s work, Electromagnetic other users. A simple illustration of TxBF Reflection. Since then, adaptive is shown in Fig. 11. beamforming techniques have been employed to remove unwanted noise and jamming from the output, mainly in military applications. With the thriving commercial wireless communication industry and the advancing microprocessor technologies, the adaptive beamforming techniques have found their applications in commercial wireless communications. With powerful digital Fig.11. An illustration of TxBF. signal processing (DSP) hardware at the base-band, algorithms could control Eigenvector TxBF Algorithm antenna beam patterns adaptively to the real signal environment, forming beams Eignenvector TxBF algorithm is widely towards the desired signals while forming used for TxBF. The eigenvector of the nulls to co-channel interferers. Thus, the spatial covariance channel matrix is system performance is optimized in terms calculated as of link quality and system capacity [31]. Adaptive antenna array can be utilized in Rss = λH (6) the transmitter side, which is known as
transmit beamforming (TxBF) or in the where Rss is the autocovariance matrix of receiver side, which is called receive the desired user`s signal, and H is the beamforming (RxBF). spatial covariance channel matrix. The eigenvector λ which corresponds to the 5.4.1 Transmit Beamforming (TxBF) max largest eigenvalue will be selected as the weight vector [36]. One example of beam The implementation of adaptive antenna pattern for 4 uniform linear array elements array technique in a handset is difficult is shown in Fig 12. with today`s hardware due to its limitations in size, cost, and energy storage capability, while it is feasible to adopt antenna arrays at base stations.
Fig. 12. Example beam pattern of 4 antenna elements in a sectorized system for a single sector (main beam direction is 240。).
Transmit Adaptive Array (TxAA) (RxBF). Through RxBF, smart antenna Algorithm system can receive predominantly from a desired direction (direction of the desired Transmit adaptive array (TXAA) is a source) compared to some undesired technique in which the user periodically directions (direction of interfering sources). sends quantized estimates of the optimal This implies that the digital processing has transmit weights to the BS via a feedback the ability to shape the radiation pattern to channel. The transmitter weights are adaptively steer beams in the direction of optimized to deliver maximum power to the desired signals and put nulls in the the user. The optimal transmit weights are direction of the interfering signals. This given by enable low co-channel interference and large antenna gain to the desired signal. HH wH= / HH (7) Based on the reference signals adopted in the beamforming algorithms, RxBF can where w is the transmit weight vector and be classified into spatial reference H is the channel matrix. beamforming (SRB), temporal reference The weights are normalized so that the beamforming (TRB), and signal structure total transmitted power is not altered. In reference beamforming (SSRB). the case of multipath channels emanating from each antenna, the optimal weights Spatial Reference Beamforming will be given by the principal eigenvector (SRB) of the channel correlation matrix HHH . Spatial reference beamforming method 5.4.2 Receive Beamforming (RxBF) is sometimes referred as direction of arrival (DoA) method. SRB estimates the Beamforming also can be applied in the direction of arrival of the signal based on uplink to improve the link quality and the spatial reference signal, using any of suppress the co-channel interference, the techniques like multiple signal which is known as receive beamforming classification or estimation of signal
Fig. 13. A general structure of SRB.
parameters via rotational invariance Collect received samples and techniques algorithms or their derivatives. estimate the covariance matrix of the They involve finding a spatial spectrum of received samples. the antenna/sensor array, and calculating Perform eigen-decomposition of the the DoA from the peaks of this spectrum covariance matrix. [37]. A general architecture of SRB Calculate spatial spectrum. algorithm is shown in Fig. 13. The general Estimate DoA by locating peaks in steps of SRB method are shown as the spectrum. follows: Estimation of signal parameters via DoA Estimation rotational invariance techniques Arbitrary Array: MUSIC, etc. (ESPRIT) is also well known for the SRB Linear Array: ESPRIIT, etc. method. In addition, ESPRIT has many Beam Synthesis important advantages over MUSIC Gram-Schmidt, etc. algorithm [38]: Combining EGC, MRC, Wiener Filter, etc. No knowledge of the array geometry and element characteristics are Multiple signal classification (MUSIC) required. algorithm estimates the DoA of the Much less complex on computation. desired signal by using an eigen-space No calibration of the array is required. method based on a spatial reference The algorithm simultaneously signal. MUSIC requires intensive estimates the number of sources and calculation of eigenvalues and DoA`s eigenvectors of an autocorrelation matrix of the input vectors from the receiving antenna array. A general step of MUSIC algorithm is shown below:
Fig. 14. A general structure of TRB.
Temporal Reference Beamforming Criterion of Adaptive Weight (TRB) Calculation Temporal Reference Beamforming In the minimum mean-square error shown in Fig. 14, is a method used to (MMSE) criterion, the weights are chosen create the radiation patter of the antenna to minimize the mean-square error (MSE) array by adding constructively the phases between the beamformer output and the of the signals in the DoA of the desired temporal reference signal. While in the user, and nulling the pattern of the maximum signal-to-interference ratio interfering users based on the temporal (MSIR) criterion, the weights are chosen reference signal [39]. Based on the to directly maximize the signal-to- temporal reference signal and some interference ratio (SIR). And The predefined adaptive weight calculation minimum variance (MV) criterion chooses criterion, some adaptive algorithms such the weights that minimize the variance of as LMS (Least Mean Square), RLS the output power. All the above three (Recursive Least Squares), and DSMI criterions has the same form of (Direct Sample Matrix Inverse) algorithms, −1 are used to adjust the weight vector of the wRvopt= β i (8) antenna array to improve the link quality. where R −1 is the inverse of the covariance The general characteristics of TRB are as i follows: matrix of the interference signals received in the antenna array and V is the antenna Good performance in multipath array propagation vector [40]. channel environment Let us assume that d(t) is the Computationally inexpensive transmitted temporal reference signal and
Requiring Training sequence Ru is the covariance matrix of interference Difficult to apply TxBF because of signals at the output of the beamformer. the absence of DoA information The calculation of β for MMSE, MSIR Table 4. β Calculation.
Criterion MMSE MSIR MV
Ed{()}2 t 2 g Ed{()} t H β 21H − vwopt H −1 1{()}+ Ed t vRvi SIR vRvu
and MV criterions are summarized in problem by normalizing with the power of Table 4. the input. The weights updating function of NLMS algorithm is shown as Adaptive Beamforming Algorithm μˆ ww(1)()tt+= + x () tε * (10) The least mean square (LMS) algorithm axt+ ()2 uses the temporal reference signal to update the weights at each iteration. In the Recursive least square (RLS) algorithm LMS algorithm, we are searching for the is derived to overcome the drawback of optimal weight that would make the array slow convergence speed in the LMS output either equal or as close as possible algorithm, when the eigenvalue spread of to the reference signal, which is the weight the correlation matrix R of received signal that minimizes the MSE. Since the MSE vector x is large. RLS algorithm replaces has a quadratic form, moving the weights the step size μ with the inverse of R. The in the negative direction of the gradient of weights are then updated using (11). the MSE should lead us to the minimum of the error surface. The weight update −1* wwRx(1)()ttt+= − (1) +ε (11) equation is shown in (9) [30].
wwx(1)()ttt+=−+μ (1)ε * (9) where μ is a constant, called the step size, which determines how close the weights approach the optimum value after each iteration and it controls the convergence speed of the algorithm. Andε is the error signal between the temporal reference signal and the received signal at the beamformer output. x(t+1) is the received signal vector at the antenna array at time t+1. The main drawback of the LMS Fig. 15. Performance comparison algorithm is that it is sensitive to the among LMS, NLMS, and RLS algorithms. scaling of its input. This makes it very Fig. 15 shows a simple performance hard (if not impossible) to choose a step comparison of the above three algorithms size μ that guarantees stability of the in OFDMA system under the Rayleigh algorithm. The normalized least mean fading channel with 8 antenna elements square (NLMS) algorithm is a variant of [41]. From this figure, we see that RLS the LMS algorithm that solves this algorithm performs best due to its faster convergence speed than LMS and NLMS References algorithms. [1] A. Goldsmith, Wireless Communica- Signal Structure Reference tions, Cambridge Univ. Press, 2005. Beamforming (SSRB) [2] M. Alamouti, “A simple transmit SSRB method is based on inherent diversity technique for wireless structure of the transmit signal of the communications”, IEEE J. Sel. Area implicit kind reference signal. Algorithms Commun., 16, pp. 1451-1458, 1998. such as blind beamforming, least squares, [3] G. Tsoulos, MIMO System Technology and constant modulus algorithms, are for Wireless Communications, Taylor based on the SSRB method. SSRB method and Francis, 2006. is robust against different propagation [4] V. Tarokh et al., “Space-time block conditions and does not require the array codes from orthogonal designs”, IEEE manifold knowledge. But the convergence Trans. Inf. Theory, vol. 45, no. 5, pp. problem becomes the main drawback of 1456-1467, 1999. the SSRB method. [5] S. Haykin and M. Mohr, Modern Wireless Communications, Pearson Prentice Hall, 2005. VI. Conclusions [6] W. Su and X-G. Xi, “Two generalized complex orthogonal space-time block In this paper we introduced the multi codes of rates 7/11 and 3/5 for 5 and 6 antenna technologies which can be transmit antennas”, IEEE Trans. Inf. considered as one of the most vivid area of Theory, vol. 49, no. 1, pp. 313-316, research. Multiple antenna technologies Jan 2003. were categorized into two main groups [7] K. Suto, and T. Ohtsuki, “Space-time- where in the first group we introduced frequency block codes over frequency some techniques related to spatial selective fading channels,” IEICE diversity and spatial multiplexing by Trans. Commun., vol. E87-B, no. 7, pp. outlining the gain achieved by these 1939-45, July 2004. schemes. Furthermore, we introduced the [8] NTT DoCoMo, “Multi-degree cyclic smart antenna techniques and the up-to- delay diversity with frequency-domain date research progress in this field. channel dependent scheduling”, 3GPP The advantages of multiple antenna TSG-RAN WG1 meeting #44bis, R1- systems make of them a very strong 060991, Mar. 2003. candidate to increase link reliability, [9] Samsung, “Adaptive cyclic delay increase channel capacity and reduce diversity,” 3GPP TSG-RAN1 43#, R1- interference in both uplink and downlink. 051354, 7th – 11st, Nov. 2005, Seoul, Korea. [10] G. Stüber, Principles of Mobile Comm-unication, Kluwer, 2001.
[11] M. K. Simon and M-S. Alouini, Digital Communication over Fading Channels, Wiley, 2005. [12] E. Biglieri et al., “Diversity,
interference cancellation and spatial multiplexing in MIMO mobile diversity in wireless networks,” Proc. WiMAX systems”, to appear in IEEE IEEE ISIT, June 2001, pp.294. WiMAX ‘07, 2007. [23] T. Hunter and A. Nosratinia, [13] G. Foschini, “Layered space-time “Diversity through coded architecture for wireless cooperation,” IEEE Trans. on communication in fading environment Wireless Commun., vol. 5, no. 2, pp. when using multi-element antennas”, 1-7, Feb. 2006. Bell Labs Technical Journal, Autumn [24] Texas Instruments, “MIMO OFDMA 1996. E-UTRA proposal for different [14] P. Wolniansky et al., “V-BLAST: an antenna configurations,” 3GPP TSG architecture for realizing very high RAN1 WG1 #43, R1-051315, Nov. data rates over the rich-scattering 2005. wireless channel”, URSI International [25] Samsung, “Downlink MIMO for Symposium on Signals, Systems and EUTRA,” 3GPP TSG RAN1 WG1 #43, Electronics, 1998. R1-051353, Nov. 2005. [15] M. Sellathurai and S. Haykin, [26] S. Ohmori, Y. Yamao, and N. “TURBO-BLAST for wireless Nakajima, “The future generationsof communications: theory and mobile communications based on experiments”, IEEE Trans. Signal broadband access technologies,” IEEE Processing, vol. 50, no., 10 pp. 2538- Commun. Mag., vol. 38, pp. 134–142, 2546, Oct. 2002. Dec. 2000. [16] A. Hottinen et al., “Industrial embrace [27] K. Sheikh, D. Gesbert, D. Gore, and A. of smart antennas MIMO,” IEEE Paulraj, “Smart antennas for Wireless Commun. Mag., Aug. 2006. broadband wireless access networks,” [17] A. Jette et al. “UMBFDD candidate IEEE Commun. Mag., vol. 37, pp. proposal for IEEE 802.20,” IEEE 134–142, Nov. 1999. C802.20-07/09, Mar. 2007. [28] A. Lozano, F.R. Farrokhi, and R.A. [18] A. Goldsmith et al., “Capacity limits Valenzuela, “Lifting the limits on of MIMO channels,” IEEE J. on Sel. high-speed wireless data access using Areas in Commun., vol. 21, no. 5, pp. antenna arrays,” IEEE Commun. Mag., 684-702, Jun. 2003. pp. 156–162, Sept. 2001. [19] M. Costa, “Writing on dirty paper,” [29] Jeffrey H. Reed,” Smart antennas: A IEEE Trans. Info. Theory, vol. 29, no. system level overview for software 3, pp. 439-441, May 1983. defined radios for creating an API”, [20] Q. Spencer et al., “An introduction to SDRF-04-I-0057-V0.00, Software the multi-user MIMO downlink,” Defined Radio Forum, Jan. 2004. IEEE Commun. Mag., pp. 60-67, Oct. [30] Ahmed EI Zooghby, Smart Antenna 2004. Engineering, Artech House, 2005. [21] A. Nosratinia and A. Hedayat, [31] Michael Chryssomallis, “Smart “Cooperative communication in antennas”, IEEE Antennas and wireless networks,” IEEE Communi- Propagation Mag., vol. 42, pp.129- cations Magazine, pp. 74-80, Oct. 136, , June 2000. 2004. [32] H. Boche and M. Schubert, [22] J. Laneman et al., “An efficient “Theoretical and experimental protocol for realizing cooperative comparison of optimization criteria for downlink beamforming,” European Trans. on Telecomm., vol. 12, no. 5, OFDMA/TDD system," in Proc. ISAP, pp. 417–426, 2001. Aug. 2005, pp.27-30. [33] R. M. Buehrer, A. G. Kogiantis, S.-C. Manar Mohaisen Liu, J. Tsai, and D. Uptegrove, “Intelligent antennas for wireless July, 2001: BS, communications -uplink,” Bell Labs Communication and Control, Technical Journal, vol. 4, no. 3, pp. University of Gaza, Gaza, Palestine 73–103, 1999. Sep., 2005: MS， [34] H. Holma and A. Toskala, WCDMA The School Polytechnic of for UMTS, John Wiley & Sons, 2000. Nice University, Sophia- [35] A. Yener, R. D. Yates, and S. Ulukus, Antipolis, France “Interference management for CDMA Feb., 2006 ~ Present: Ph.D. student, The Graduate School of IT & T, INHA University systems through power control, 2001 ~ 2003: The Palestinian Telecommunication multiuser detection, and Company (JAWWAL) beamforming,” IEEE Trans. on
[36] S.J. Ko, J. Heo, and K.H. Chang, "An effective downlink resource allocation July, 2004: BS, for supporting heterogeneous traffic Communication Engineering, data in an OFDM/SDMA-based Northeastern University, cellular system," in Proc. IEEE Shenyang, China GLOBECOM, Nov. 2006, WLC27-5. July, 2006: MS, The Graduate School of IT [37] Lal C. Godara, “Application of & T, INHA University antenna arrays to mobile Sep., 2006 ~ Present: Ph.D. communications, part II: beam- student, The Graduate School of IT & T, INHA forming and direction-of-arrival University considerations”, IEEE Proc., vol .85,