Unit 2

Diversity

When the path is in a deep fade, any communication scheme will likely suffer from errors. A
natural solution to improve the performance is to ensure that the information symbols pass
through multiple signal paths, each of which fades independently, making sure that reliable
communication is possible as long as one of the paths is strong. This technique is called
diversity, and it can dramatically improve the performance over fading channels.

There are many ways to obtain diversity. Diversity over time can be obtained via coding and
interleaving: information is coded and the coded symbols are dispersed over time in different
coherence periods so that different parts of the codewords experience independent fades.
Analogously, one can also exploit diversity over frequency if the channel is frequency-
selective. In a channel with multiple transmit or receive antennas spaced sufficiently far
enough, diversity can be obtained over space as well. In a cellular network, macrodiversity can
be exploited by the fact that the signal from a mobile can be received at two base-stations.
Since diversity is such an important resource, a wireless system typically uses several types of
diversity.

Types of Diversity

a. Time Diversity: Time diversity is obtained by re-transmitting the same signal at

separate interval of time. The re-transmission interval should be less than the coherence
time. Time Diversity also does not require any increase in the transmission power.
b. Antenna Diversity: Antenna diversity, or spatial diversity, can be obtained by placing
multiple antennas at the transmitter and/or the receiver. If the antennas are placed
sufficiently far apart, the channel gains between different antenna pairs fade more or
less independently, and independent signal paths are created. The required antenna
separation depends on the local scattering environment as well as on the carrier
frequency.
c. Frequency Diversity: Frequency diversity is obtained by transmitting the same
information on more than one carrier frequency or transmitting the same signal over
different carrier frequency. In Frequency Diversity only one antenna is needed.
d. Polarization Diversity: Polarization Diversity require two transmitter and two receiving
antennas with different polarization. The transmission wave with two different
polarizations constitutes two different paths. This provides only two different diversity

Dr.HMV
 branches. Polarization diversity uses half power by dividing the power between two
different polarized antennas.

Rayleigh fading

The Rayleigh fading model uses a statistical approach to analyse the propagation, and can be
used in a number of environments. The Rayleigh fading model is ideally suited to situations
where there are large numbers of signal paths and reflections.

“Rayleigh fading models assume that the magnitude of a signal that has passed through such a
transmission medium (also called a communications channel) will vary randomly, or fade,
according to a Rayleigh distribution — the radial component of the sum of two uncorrelated
Gaussian random variables.”

Let us consider flat fading.

- The results for frequency selective channels are very similar since the amplitudes of
different delta functions fade independently.
- We also assume that there is no LOS (Line of Sight) path between the transmitter and
the receiver.
- In a multipath channel with I multiple paths, transmitting a signal over the carrier
frequency fc results in receiving the sum of I components from different paths plus a
Gaussian noise as follows:

- where ai and φi are the amplitude and phase of the i th component, respectively, and η(t)
is the Gaussian noise. Expanding the cos (·) term, we get,

- The terms and are the summation of I random

variables since the objects in the environment are randomly located.
- For a large value of I, using the central limit theorem, the random variables A and B
are independent identically distributed (iid) Gaussian random variables.

- The envelope of the received signal is .

- Since A and B are iid zero-mean Gaussian random variables, the envelope follows a
Rayleigh distribution.

Dr.HMV
 - The probability density function (PDF) of a Rayleigh random variable is

where σ2 is the variance of the random variables A and B.

- The received power, is an exponential random variable with a PDF:

Alamouti Space Time Code

Let us assume a system with N = 2 transmit antennas and one receive antenna, employing
Alamouti code as shown in below figure.

Fig. Transmitter block diagram for Alamouti code.

- To transmit b bits/cycle, we use a modulation scheme that maps every b bits to one
symbol from a constellation with 2b symbols.
- The constellation can be any real or complex constellation.
- First, the transmitter picks two symbols from the constellation using a block of 2b bits.
- If s1 and s2 are the selected symbols for a block of 2b bits, the transmitter sends s1 from
antenna one and s2 from antenna two at time one.
- Then at time two, it transmits −s2∗ and s1∗ from antennas one and two, respectively.
- Therefore, the transmitted codeword is

- To check if the code provides full diversity, we need to calculate the rank of all possible
difference matrices D(C,C') and show that it is equal to two for every C'=C.
- Let us consider a different pair of symbols (s'1,s2) and the corresponding codeword

Dr.HMV
 - The difference matrix D(C,C') is given by

- The determinant of the difference matrix det[D(C,C')] is zero if and only if s'1= s1 and
s'2= s2.
- Therefore, D(C,C') is always full rank when C'=C and the Alamouti code satisfies the
determinant criterion.
- It provides a diversity of 2M for M receive antennas and therefore is a full diversity
code.
- Let us assume that the path gains from transmit antennas one and two to the receive
antenna are α1 and α2, respectively.
- The decoder receives signals r1 and r2 at times one and two, respectively, such that

- For a coherent detection scheme where the receiver knows the channel path gains α1
and α2, the maximum-likelihood detection amounts to minimizing the decision metric

over all possible values of s1 and s2.

- Such a decoding scheme requires a full search over all possible pairs (s1, s2) and in
general its complexity grows exponentially by the number of transmit antennas.

- The receiver should minimize to decode s1 and minimize

to decode s2.
- Therefore, the decoding consists of first calculating

- Then, to decode s1, the receiver finds the closest symbol to in the constellation,
similarly for s2.

Fig. Receiver block diagram for Alamouti code.

Dr.HMV
The Alamouti code provides two important properties.

- Simple decoding: Each symbol is decoded separately using only linear processing.
- Maximum diversity: The code satisfies the rank criterion and therefore provides the
maximum possible diversity.

MIMO Spatial Multiplexing

One of the key advantages of MIMO spatial multiplexing is the fact that it is able to provide
additional data capacity. MIMO spatial multiplexing achieves this by utilising the multiple
paths and effectively using them as additional "channels" to carry data.

- To take advantage of the additional throughput capability, MIMO utilises several sets
of antennas.
- In any case for MIMO spatial multiplexing the number of receive antennas must be
equal to or greater than the number of transmit antennas.
- To take advantage of the additional throughput offered, MIMO wireless systems utilise
a matrix mathematical approach.
- Data streams t1, t2, . . . tn can be transmitted from antennas 1, 2, . . . n.
- Then there are a variety of paths that can be used with each path having different
channel properties.
- These can be represented by the properties h12, travelling from transmit antenna one to
receive antenna two and so forth. In this way for a three transmit, three receive antenna
system a matrix can be set up as:
r1 = h11* t1 + h21* t2 + h31* t3
r2 = h12 * t1 + h22 * t2 + h32 * t3
r3 = h13 * t1 + h23 * t2 + h33 * t3
Where r1 = signal received at antenna 1, r2 is the signal received at antenna 2 and so
forth.
- In matrix format this can be represented as:
[R] = [H] x [T]
- To recover the transmitted data-stream at the receiver it is necessary to perform a
considerable amount of signal processing.
- First the MIMO system decoder must estimate the individual channel transfer
characteristic hij to determine the channel transfer matrix.

Dr.HMV
 - Once all of this has been estimated, then the matrix [H] has been produced and the
transmitted data streams can be reconstructed by multiplying the received vector with
the inverse of the transfer matrix.
- [T] = [H]-1 x [R]
- This process can be likened to the solving of a set of N linear simultaneous equations
to reveal the values of N variables.
- In reality the situation is a little more difficult than this as propagation is never quite
this straightforward, and in addition to this each variable consists of an ongoing data
stream, this nevertheless demonstrates the basic principle behind MIMO wireless
systems.

Layered Space Time Receivers

The three most common receivers for uncoded spatial multiplexing (SM) are ML ZF, and
MMSE.

Maximum Likelihood (ML) Receiver

- This is an optimum receiver. If the data stream is temporally uncoded, the ML receiver
solves

where 𝑠̂ is the estimated symbol vector. The ML receiver searches through all the vector
constellation for the most probable transmitted signal vector.

- This implies investigating the exponent of transmit antenna combinations, a very

difficult task. Hence, these receivers are difficult to implement, but provide full receiver
diversity and zero power losses as a consequence of the detection process. In this sense
it is optimal.

Zero-Forcing Receiver

- The ZF receiver is a linear receiver. It behaves like a linear filter and separates the data
streams and thereafter independently decodes each stream.
- Assume that the channel matrix H is invertible and estimate the transmitted data symbol
vector as

Dr.HMV
 where † represents pseudoinverse.
- Since an inverse of H can only exist if the columns of H are independent, it is assumed
that H = 𝐻𝑤
- The ZF receiver decomposes the link into MT parallel streams, each with diversity gain
and array gain proportional to MR - MT + 1. Hence, it is suboptimum

Minimum Mean Square Error (MMSE) Receiver

- Minimum Mean Square Error (MMSE) approach alleviates the noise enhancement
problem by taking into consideration the noise power when constructing the filtering
matrix.
- The MMSE is used to minimize the mean square error (MSE).
- The goal of the MMSE equalizer is to multiply by a matrix such that the resulting
effective noise is minimized.
- Equivalently, MMSE equalizer maximizes the effective SNR.
- Let us assume that x be an unknown random variable and R be a known random
variable, then
R=HX+n
- An estimator X(R) is any function of the measurement y, and its mean square error is
given by
MSE = E{( 𝑋̂- X)2}
- The MMSE always performs better than the ZF equalizer and is of the same
complications of implementation.

Sphere Decoding

The main idea behind sphere decoding is to limit the number of possible codewords by
considering only those codewords that are within a sphere cantered at the received signal
vector.

- The main idea in sphere decoder is to reduce the number of candidate vector symbols
to be considered in the search that solves without accidentally excluding the ML
solution.
- However, it considers only a small set of vectors within a given sphere rather than all
possible transmitted signal vectors.
- Sphere decoding adjusts the sphere radius until there exists a single vector (ML solution
vector) within the sphere.

Dr.HMV
 - This goal is achieved by constraining the search to only those points of Hx that lies
inside a hyper sphere with radius R around the received point.
- The search can easily be restricted by drawing a circle around the received signal.
- So, the search allows only those codeword to be checked that happen to fall within the
sphere.
- All the remaining codeword outside the sphere are not taken into consideration for
decoding. The corresponding inequality is referred to as the sphere constraint (SC).

The next step is to find the proper radius and radius reduction.

- The initial radius selected plays a critical role in identifying the correct point in the
lattice.
- Ideally, the noise variance of the system is found and the initial radius of the sphere is
adjusted according to the Signal to Noise Ratio. This entails the sphere decoder to find
at least a single point inside the sphere.
- The radius is increased when there exists no vector within a sphere, and decreases the
radius when there exist multiple vectors within the sphere.

- In the above depicted figure, the sphere includes four candidate vectors, one of which
is the ML solution vector.
- We note that no vector outside the sphere can be the ML solution vector because their
ML metric values are bigger than the ones inside the sphere.
- If we were fortunate to choose the closest one among the four candidate vectors, we
can reduce the radius, so that we may have a sphere within which a single vector
remains. In other words, the ML solution vector is now contained in this sphere with a
reduced radius, as illustrated in Figure (b)

If the radius is too large, average processing cycle becomes extremely high, making real time
operation impossible. On the other hand, if the radius is too small, even the ML solution cannot

Dr.HMV
satisfy the sphere constraint. Thus, setting the appropriate radius is very critical in successful
implementation of the sphere decoding.

BLAST architectures

Depending on the type of coding employed, BLAST architectures are classified into:

1. Diagonal-BLAST (DBLAST)
2. Vertical-BLAST (V-BLAST)
3. Turbo-BLAST

Unlike other spread-spectrum multiple access techniques, the total channel bandwidth utilizes
in a BLAST system is only a small fraction in excess of the symbol rate.

V-BLAST

The encoder of V-BLAST is as shown below.

Figure. Encoder of V-BLAST.

- The input bitstream is demultiplexed into N parallel substreams.

- Then, each substream is modulated and transmitted from the corresponding transmit
antenna.
- At the cost of bandwidth, it is possible to improve the performance by coding each
substream.
- Representing the nth symbol of the codeword C by cn and the nth row of H by Hn, the
output vector can be written as

- First, the algorithm decodes the strongest symbol.

- Then, cancelling the effects of this strongest symbol from all received signals, the
algorithm detects the next strongest symbol.

Dr.HMV
 - The algorithm continues by cancelling the effects of the detected symbol and the
decoding of the next strongest symbol until all symbols are detected.
- The optimal detection order is from the strongest symbol to the weakest one.
- The algorithm works only in the number of receive antennas is more than the number
of transmit antennas.
- The V-BLAST decoder is as shown in the below figure.

Figure. V-BLAST decoder block diagram.

- The mechanism of the algorithm includes three steps, namely:
i. Ordering
ii. Interference cancellation
iii. Interference nulling.
- The purpose of the ordering step is to decide which transmitted symbol to detect at each
stage of the decoding. The symbol with highest SNR is the best pick in this step.
- The goal of the interference cancelation is to remove the interference from the already
detected symbols in decoding the next symbol.
- Finally, interference nulling finds the best estimate of a symbol from the updated
equations.

D-BLAST

- The encoder of the D-BLAST is shown below, which is similar to that of V-BLAST.

Figure. D-BLAST encoder.

Dr.HMV
 - The main difference from V-BLAST is in the way that the signals are transmitted from
different antennas.
- In V-BLAST, all signals in each layer are transmitted from the same antenna.
- However, in D-BLAST, the signals are shifted before transmission.
- Let us enumerate the output of the demultiplexing as layers 1 to N.
- For the first block of T time slots, Layer n is sent by the nth transmit antenna. (Note
that the block size T can be as small as T = 1)
- In the second block, Layer n is transmitted by the (n + 1)th antenna, for n = 1, 2, . . . ,
N − 1, while Layer N is transmitted by the first antenna.
- The shift continues as a cyclic shift after every block of T time slots.
- Table demonstrates how different layers are transmitted diagonally because of the
cyclic shift for a system with N = 4 transmit antennas.
Table. Layering in D-BLAST

- The receiver of a D-BLAST architecture is similar to that of a V-BLAST system

- although the shifting creates a higher complexity.
- To decode a symbol, the main steps are interference nulling and interference
cancelation to detect the best estimate of the symbol.
- Layers are detected one by one following the diagonal pattern of the transmitter.
- Let us assume N = 4 transmit antennas as in Table.
- First, detect the best signals to decode the symbols in Layer 1 during the first N = 4
blocks.
- Since we only transmit Layer 1 during the first block, there is no interference and a
maximum ratio combining provides the best candidate for decoding.

Dr.HMV
 - During the second block for Layer 1, there is interference from Layer 4, considered as
noise. Then it is nullified using ZF or MMSE.
- Similarly, the interference of Layers 3 and 4 to Layer 1 are nulled during the third block.
- Finally, during the fourth block, the interference of Layers 2, 3, and 4 is removed from
Layer 1 and called as detected Layer 1.
- The decoded symbols of Layer 1 are used to cancel their interference on the undetected
layers.
- After cancelling the interference of the decoded Layer 1, a similar method is applied to
detect Layer 4 in the second diagonal starting from Block 2.
- A combination of interference nulling and cancelation is sequentially applied to every
diagonal to detect all layers.

Turbo-BLAST

Figure. Turbo-BLAST encoder.

- First, the input bit stream is demultiplexed into N equal rate substreams.
- Then, each substream is independently encoded using a linear block code.
- The output bits of the block codes are bit-interleaved using a random interleaver. The
interleaver is independent of the input bitstream and is designed offline.
- It is important that each substream uses different paths equally.
- The diagonal space interleaver is shown below.

Dr.HMV
 Table. Diagonal interleaving.

- The original diagonal layering of D-BLAST suffers from boundary effects that result
in some throughput waste.
- The output of the space-time interleaver is modulated separately for each antenna and
is transmitted.
- The effects of the diagonal interleaving on the quasi-static fading channel is to create
an “effective” time-varying channel. Using this effective time-varying channel, the
encoder can be represented as serially concatenated turbo code as shown in figure.

Figure. Turbo-BLAST encoder represented by serially concatenated codes.

- The “outer code” consists of block encoders for different substreams.
- The “inner code” is the effective channel resulting from diagonal interleaving.
- The complexity of an optimal decoder for turbo codes grows exponentially with N,
constellation size, and the block size. Therefore, iterative decoding algorithm is
suggested.

Dr.HMV
 - The main idea behind an iterative decoding algorithm is first to divide the problem into
two stages. Then, we solve the problem for each stage optimally and iteratively while
exchanging the information between the two stages.
- However, at the cost optimality, it provides a very good performance with a manageable
complexity.
- The two stages are decoding of the inner code and decoding of the outer block code.
- The inner decoder and outer decoder are separated by space-time interleavers and de-
interleavers similar to a regular turbo decoder.
- The role of the space-time interleavers and de-interleavers is to reverse the effects of
the spacetime interleaving at the encoder.
- The block diagram of the iterative decoder is as shown below.

Figure. Turbo-BLAST decoder block diagram.

Dr.HMV