MIMO Communication

Module 3
Prepared By: Dr Vinod B Durdi
DSCE, TCE, Bangalore-78
Course Code: TE814

Space-Time Block Codes & Space-Time Trellis Codes

Space-Time Block Codes:-

In the previous chapter, we have seen that the capacity and constrained information rates over fading channels
increase significantly with the employment of multiple transmit and multiple receive antennas. In this chapter, we
will introduce the idea of space-time block coding which is a practical means to achieve the benefits offered by
MIMO systems. Space-time block coding is a simple yet very effective means of achieving transmit diversity
when other forms of diversity may be limited or non-existent, e.g., for quasi-static fading channels. Such codes can
be easily generalized to the case of multiple receive antennas as well, thus providing receive diversity in addition to
transmit diversity. Furthermore, they can be decoded efficiently at the receiver by simple linear processing of the set
of received signals at different receive antennas.

Our objective in this chapter is to study space-time block coding in detail. We beginwith the simple case of two
transmit antennas. We will detail the methods, give performance analysis results, and illustrate their performance by
both analytical tools and using simulations.

The chapter is organized as follows. We first describe the Alamouti scheme which is a simple way of obtaining
transmit diversity for the case of two transmit antennas. We then generalize this scheme to the case of more than two
transmit antennas. For both approaches, we consider the optimal receiver structures, theoretical performance analysis
in the presence of Rayleigh fading and performance evaluation by simulations. Furthermore, we study
quasi-orthogonal space-time block codes, and linear dispersion codes. Finally, we provide our conclusions and
suggestions for further reading.

Transmit Diversity with Two Antennas:

The Alamouti Scheme
As we have discussed in the previous chapter, it is relatively easy to obtain spatial diversity by employing multiple
receive antennas. Consider, for instance, the uplink of a cellular telephony system that is the transmission is from a
mobile to the base-station. Since the base-stations can be equipped with multiple antennas with sufficient separation
easily, the signal transmitted by the mobile unit can be picked up by multiple receive antennas and theycan be
combined using a diversity-combining technique, e.g., maximal-ratio combining, selection combining, equal-gain
combining, etc., to obtain receive diversity. However, if the situation is reversed (i.e., for downlink transmission)
achieving diversity gain is not that simple due to the fact that the mobile units are typically limited in size, and it is
usually difficult to place multiple antennas that are separated by sufficiently large distances for reception of multiple
copies of the transmitted signal through independent channels.Therefore, it is desirable to have a scheme where the
benefits of (spatial) diversity are exploited through “transmit diversity”. With this motivation, Alamouti (1998)
introduced away of obtaining transmit diversity when there are two transmit antennas.
Transmission Scheme
The Alamouti scheme is a simple transmit diversity scheme suitable for two transmit
antennas. Two symbols are considered at a time, say x1 and x2, and they are transmitted in
two consecutive time slots. In the first time slot, x1 is transmitted from the first antenna and x2
is transmitted from the second one. In the second time slot, −x∗2 is transmitted from the first
antenna, while x∗1 is transmitted from the second antenna. This process is illustrated in Figure 1.
The signals x1 and x2 are picked from an arbitrary (M-ary) constellation. Since two symbols are
transmitted in two time slots, the overall transmission rate is 1 symbol per channel use, or log2M
bits per channel use.

Figure 1 TheAlamouti scheme

Optimal Receiver for the Alamouti Scheme

Let us now derive the optimal receiver for the Alamouti scheme. We consider two separate cases, namely, a single
receive antenna and multiple receive antennas

Single Receive Antenna System

Consider the case of a single receive antenna. The received signal in the first time slot isThen

in the second time slot. We assume that the channel is Rayleigh fading, i.e., h1,1and h2,1 arezero
mean complex Gaussian random variables with unit variance (i.e., with variance ½ per
dimension), and they remain the same for two consecutive time intervals. We normalize the
power of each constellation to be 1/2, thus the total transmit power per channel use is unity. The
additive noise terms n1(1) and n1(2) are complex AWGN with variance ½ per dimension. With
these definitions, the signal-to-noise ratio becomes ρ.Define the vector of the received signals
(where the second signal is conjugated) as
Clearly, being linear combinations of jointly Gaussian random variables, n_ 1(1) and n1(2) are
jointly Gaussian. Since they are also uncorrelated, they are independent. Also, they both have
zero mean and a variance of 12(|h1,1|2 + |h2,1|2) per dimension.

Therefore, the optimal decisions ˆx1 and ˆx2 decouple, and simplify to the usual minimization of
the Euclidean distance between the possible transmitted symbols and the respective Components
vector HH y. That is,
 Clearly, being linear combinations of jointly Gaussian random variables, n
_

1(1) and n1(2) are jointly Gaussian. Since they are also uncorrelated, they are independent. Also,
they both have zero mean and a variance of 12 (|h1,1|2 + |h2,1|2) per dimension. Therefore, the
optimal decisions ˆx1 and ˆx2 decouple, and simplify to the usual minimization of the Euclidean
distance between the possible transmitted symbols and the respective components of the vector
HH y.
That is,
 As we will discuss later, the full-diversity of the MIMO system, which is two for the case
of two transmit and one receive antenna, can be obtained with the use of the Alamouti scheme.
Therefore, even if the receivers are not equipped with multiple antennas, spatial diversity can
still be achieved, which makes this scheme very beneficial.

Multiple Receive Antenna System:

The Alamouti scheme can easily be extended to systems with multiple receive antennas as well,
resulting in receive diversity in addition to the existing transmit diversity. For this case, the
available diversity order is twice the number of receive antennas, and it can also be achieved by a
simple linear receiver.

Assume that the received signal during the kth time slot at the jth receive antenna is yj (k), where
k = 1, 2, j = 1, 2, . . .,Nr . The channel coefficient from the ith transmit antenna to the jth receive
antenna is denoted by hi,j . We can then write

in the first time slot, and

in the second time slot. Here, nj (k) is the AWGN term at time slot k and receive antenna j.

Of course, the optimal decision rule can be simplified further if a constant energy constellation,
such as phase shift keying, is employed. For instance, for the case of BPSK modulation, it takes
the form

Performance Analysis of the Alamouti Scheme:

Consider a 2 × Nr MIMO system employing the Alamouti scheme. After linear
processing (which is shown to be optimal), as argued in the previous section, the decision
variable for the symbol xk is given by

From this expression it is clear that the average bit error probability of the Alamouti scheme
when BPSK modulation is employed decays inversely with the (2Nr )th power of the signal-to
noise ratio. Therefore, the diversity provided is 2Nr (i.e., full spatial diversity) over a Rayleigh
flat fading channel.

In this section, we present several error rate results for the Alamouti scheme used over Rayleigh
flat fading channels based on both simulations and performance analysis carried out in the
previous section. In Figure 2, we present the bit error rate obtained using BPSK modulation for
several cases. For comparison purposes, we also show the performance of the no-diversity case,
i.e., the single-input single-output system. For the two transmit and one receive antenna case, we
observe that the diversity order is two, and for the two transmit and two receive antenna case, it
 is four. We also observe that the simulations and theoretical results agree with each other.

Figure 2.Bit error rate performance of the Alamouti scheme with BPSK modulation
Orthogonal Space-Time Block Codes:
The Alamouti scheme is designed for two transmit antennas.

A question that naturally comes to mind is: How can this transmit diversity scheme be extended
to the case of more than two transmit antennas?

In this section, we describe what is known as the general space-time block codes
This based on the theory of orthogonal designs.

Let us present example of a space-time block code. Assume that we are to transmit the four
symbols {x1, x2, x3, x4} selected from a real signal constellation

Consider a system with Nt transmit antennas.

The objective is to design a set of Nt × Nt matrices

Assume that Nt = 4 and a.

We form the code matrix, A specific space-time block code can be described by the 4 × 4 matrix

Clearly, the columns of X1 are mutually orthogonal.

ith row of this matrix using the four transmit antennas.
Symbol interval i (1 ≤ i ≤ 4),
That is, the (i, j)th element of the matrix is transmitted from antenna j during the ith time slot.
This is a full-rate code example since we are able to transmit four symbols in four time slots,
Let us present another example of a space-time block code. Assume that we are to transmit the
four symbols {x1, x2, x3, x4} selected from a complex signal constellation.

A specific space-time block code can be described by the 8 × 4 matrix.

At time interval i (1 ≤ i ≤ 8), the ith row of X2 is transmitted using the four transmit antennas.
This code achieves full-diversity as well (this will be illustrated later); however, clearly, it is not
full-rate since only four symbols are transmitted using eight time slots.

This simple example clearly illustrates that if we relax the full-rate code design condition,
there are many potential designs one can employ.

In order to present general space-time block codes in a more concrete framework, let us
describe the important sub class of linear orthogonal designs in detail.
Linear Orthogonal Designs:

Let us now describe the general class of linear orthogonal designs in a mathematical framework
to combine different space-time block codes in a unified manner.

We denote the N × Nt code matrix that defines the space-time block code by X.

This matrix is used to transmit M symbols in N time slots (achieving a rate of M/N).

X is orthogonal.

We refer to this class of space-time block codes as linear orthogonal designs. We can
equivalently write the code matrix X in the form
where Am and Bm are N × Nt matrices, and {xm}, M are the set of symbols being transmitted.
This representation is general, and thus can be used to describe different codes studied earlier.
For example, for the Alamouti code, we have Nt = T = M = 2 with

For the space-time block code defined by X2 in expression (4.34), it is easy to see that the
matrices take the form

It is straightforward to show that the space-time block code matrix X defines an orthogonal
design if and only if the equalities

are satisfied (see Xu and Kwak (2005)). Here δij is the Kronecker delta (i.e., δij = 1 if i = j, and 0
otherwise), and Di is a diagonal matrix where the diagonal elements are strictly positive.
Decoding of Linear Orthogonal Designs
Let us now describe the decoding of space-time block codes based on the framework presented
in the previous section. Let yj denote the N × 1 received signal vector at antenna j whose kth
element, yj (k), shows the received signal at the kth time interval. We have

where the Nt × 1 vector hj shows the channel coefficients from all the transmit antennas to the jth
receive antenna, and the vector nj is the complex Gaussian noise term. The channel gains and
noise terms are normalized so that their variance per dimension is ½ and the signal constellation
is scaled so that its average energy is 1/Nt at each transmit antenna. Therefore, ρ denotes the
signal-to-noise ratio at each receive antenna.

Define the 2N × 1 vectors

Based on the structure of the space-time block codes based on the orthogonal designs discussed
in the previous section, we can write the equality.

Since the noise is AWGN, and the channel coefficients are assumed to be known at the receiver,
the optimal decision can be obtained by minimizing the squared Euclidean distance between a
candidate codeword (X(ˆx )) and the received signal, which is given by

With this set of equations, the optimal decoding rule which is simple linear combining is
obtained explicitly
Performance Analysis of Space-Time Block Codes

In this section, we calculate the error rates offered by space-time block codes over Rayleigh
fading channels. We again assume that the sub-channels between different antenna pairs fade
independently. Assume that the sequence of symbols {x1, x2, . . . , xM} are being transmitted.
Then, the decision variable um computed for xm at the receiver by linear processing is given by

By using the
properties of linear orthogonal designs given by , we can simplify this expression to

This expression can be approximated for large signal-to-noise ratios as

This expression clearly shows the diversity advantage provided by the space-time block coding
schemes. The bit error rate for the BPSK modulation scheme decays inversely with the (NrNt )th
power of the signal-to-noise ratio, i.e., the diversity order achieved is NrNt which is the full
spatial diversity that can be obtained. Clearly this generalizes the result obtained for the case of
the Alamouti scheme which was specific to the case of two transmit antennas.

which clearly demonstrates the diversity advantage that can be obtained

Examples
We now examine the performance of the two space-time block code examples of the previous
section through simulations. We first consider the case of four transmit antennas with the code
defined by X1. We present the bit error rate with BPSK modulation over a Rayleigh flat fading
channel in Figure 3. This is a full-rate code achieving one bit per transmission. We show both the
simulation results and the theoretical results derived in the previous section. We observe that the
diversity achieved is four for the case of one receive antenna, and it is eight for the case of two
receive antennas as expected.
Figure 3 Bit error rate of X1 with BPSK modulation
We consider the half rate space-time block code example given by X2 in Figure 4. In this case
there are four transmit antennas, and we assume that QPSK is employed resulting in a
transmission rate of one bit per channel use. We observe that the simulation results and the
theoretical expectations clearly match, and they demonstrate a diversity order of four
 Figure 4 Symbol error rate of X2 with QPSK modulation

Quasi-Orthogonal Space-Time Block Codes:

As we have seen earlier in this chapter, there are no full-rate, full-diversity codes (for complex
constellations) other than the Alamouti scheme, and such designs are very limited for real
constellations.

The quasi-orthogonal space- time block coding approach of Jafarkhani (2001) gives a way of
obtaining full-rate (or increased-rate) space-time block coding designs using smaller designs as
building blocks.
Let us illustrate the ideas via a simple example by presenting a full-rate design for the Nt = 4
scheme using the Alamouti code.

We consider the transmission of four symbols, say x1, x2, x3 and x4.

We use the Alamouti code to encode these symbols pair wise ie x1,x2 and x3 x4 resulting in two
2 × 2 matrices of the form
We then use these two matrices in another orthogonal design to obtain the 4 × 4 quasiorthogonal
space-time block code matrix as where (·) ∗ denotes the complex conjugate of a matrix.

Thus, we obtain

This is clearly a way of obtaining a full-rate code with complex constellations for the four
transmit antenna case. However, as we know it cannot achieve full-diversity as no such full-rate
full-diversity design exists. In fact, it is easy to show that the diversity order achieved with this
scheme is 2Nr .

Clearly not all columns of the code matrix are orthogonal.

To describe the decoding process, we observe that the first and fourth columns of the code
matrix are orthogonal to the second and third columns, respectively. Therefore, with appropriate
linear processing, the decisions for x1 and x4 can be decoupled from decisions for x2 and x3.

After that step, the pairs of codewords have to be decoded jointly, thus increasing the complexity
of the decoding algorithm compared with the case of linear orthogonal designs.
We conclude this section by emphasizing that other existing smaller orthogonal designs can also
be extended to a larger number of transmit antennas by a similar approach in a straightforward
way, giving rise to similar conclusions.

Linear Dispersion Codes:

We now describe another class of codes, called linear dispersion codes, which generalize the idea
of orthogonal and quasi-orthogonal space-time block codes.

As discussed earlier, orthogonal space-time block codes achieve full diversity over a MIMO
channel. However, the rate of transmission offered by them is not very high.

To alleviate this problem, Hassibi and Hochwald (2002) introduced linear dispersion codes which
basically remove the orthogonality constraint imposed before. Let us now describe this class of
codes in some detail, Assume that x1, x2, . . . , xM represent M symbols to be transmitted over N
time slots. The transmitted codeword is described by

where A and B are arbitrary N × Nt matrices, called dispersion matrices. This description looks
identical to the one in (4.35), however we are now talking about a much broader class of codes,
since the condition given in (4.37) to obtain linear orthogonal designs is not imposed here.
As another example, consider the scenario where the number of symbols transmitted is M =N Nt ,
and the space-time codeword is given by

which basically describes a very high-rate transmission scheme where at different time slots and
different antennas, different symbols are transmitted. Clearly, this is not an orthogonal design, or
any other design we have seen thus far. Since the special properties that were imposed in
orthogonal designs are lacking, decoding or analysis may not be simple. In fact, this scheme will
be referred as the vertical Bell Labs layered space-time (VBLAST) architecture. Maximum
likelihood decoding for linear dispersion codes (with parameters of interest) is out of the
question due to the very high complexity requirements. Linear processing (or one of its
variations) is also not feasible since there is no orthogonality constraint imposed on the
dispersion matrices.

Other things in SRBC

Examine the STBC.

Space–time block coding is a technique used in wireless communications to transmit multiple

copies of a data stream across a number of antennas and to exploit the various received versions
of the data to improve the reliability of data transfer. The fact that the transmitted signal must
traverse a potentially difficult environment with scattering, reflection, refraction and so on and
may then be further corrupted by thermal noise in the receiver means that some of the received
copies of the data may be closer to the original signal than others. This redundancy results in a
higher chance of being able to use one or more of the received copies to correctly decode the
received signal. In fact, space–time coding combines all the copies of the received signal in an
optimal way to extract as much information from each of them as possible.
In the case of STBC in particular, the data stream to be transmitted is encoded in blocks, which
are distributed among spaced antennas and across time. While it is necessary to have multiple
transmit antennas, it is not necessary to have multiple receive antennas, although to do so
improves performance.
An STBC is usually represented by a matrix. Each row represents a time slot and each column
represents one antenna's transmissions over time.
Only one standard STBC can achieve full-rate (rate 1) —
Alamouti'scode.
Here, Sij is the modulated symbol to be transmitted in time sloti from antenna j. There are
to be T time slots and T transmit antennas as well as R receive antennas. This block is
usually considered to be of 'length' T
The code rate of an STBC measures how many symbols per time slot it transmits on average over
the course of one block. If a block encodes k symbols, the code-rate is

Analysis of Orthogonal STBC..

This means that the STBC is designed such that the vectors representing any pair of columns
taken from the coding matrix is orthogonal.
Orthogonal STBC (O-STBC) has the simplest decoding complexity, as its ML decoding can be
achieved by linear detection.
Orthogonality
STBCs as originally introduced, and as usually studied, are orthogonal. This means that the
STBC is designed such that the vectors representing any pair of columns taken from the coding
matrix is orthogonal. The result of this is simple, linear, optimal decoding at the receiver. Its
most serious disadvantage is that all but one of the codes that satisfy this criterion must sacrifice
some proportion of their data rate.
has to be full-rank for any pair of distinct codewords c and e to give the maximum possible
diversity order of NtNr . If instead, B(c,e) has minimum rank over the set of pairs of distinct
codewords, then the space–time code offers diversity order . STBCs offer only diversity gain
(compared to single-antenna schemes) and not coding gain. There is no coding scheme included
here — the redundancy purely provides diversity in space and time. This is contrast with space–
time trellis codes which provide both diversity and coding gain since they spread a conventional
trellis code over space and time.

Hypothesize Quazi Orthogonal STBC.

These codes exhibit partial orthogonality and provide only part of the diversity gain. The quasi
orthogonal STBCs achieve higher data rates than Orthogonal STBC at the cost of inter-symbol
interference (ISI). Antenna system and has the coding matrix: where * denotes complex
conjugate.

The orthogonality criterion only holds for columns (1 and 2), (1 and 3), (2 and 4) and (3 and 4).
Crucially, however, the code is full-rate and still only requires linear processing at the receiver,
although decoding is slightly more complex than for orthogonal STBCs. Results show that this
Q-STBC outperforms (in a bit-error rate sense) the fully orthogonal 4-antenna STBC over a good
range of signal-to-noise ratios (SNRs). At high SNRs, though (above about 22 dB in this
particular case), the increased diversity offered by orthogonal STBCs yields a better BER.

Decoding of Orthogonal STBC.