WC& Lte (15ec81) Module - 2 Multicarrier Modulation

WC& LTE (15EC81) Module -2 MULTICARRIER MODULATION
Module -2 MULTICARRIER MODULATION

Multicarrier Modulation: OFDM basics, OFDM in LTE, Timing and Frequency Synchronization,
PAR, SC-FDE.
OFDMA and SC-FDMA: OFDM with FDMA,TDMA,CDMA, OFDMA, SC-FDMA, OFDMA and
SC-FDMA in LTE .
Multiple Antenna Transmission and Reception: Spatial Diversity overview, Receive Diversity,
Transmit Diversity, Interference cancellation and signal enhancement, Spatial Multiplexing, Choice
between Diversity, Interference suppression and Spatial Multiplexing.
Multicarrier Modulation: OFDM basics

OFDM employs an efficient computational technique known as DFT, which lends itself to highly
efficient implementation commonly known as FFT.
Block Transmission with Guard Intervals

Grouping of L data symbols into a block known as an OFDM symbol. An OFDM symbol lasts for a
duration of T seconds, where Y = LTs.To keep each OFDM symbol independent of others, guard time
is introduced between each OFDM symbol as shown below
After receiving a series of OFDM symbols , as long as the guard time Tg is larger than the delay spread
of the channel η each OFDM symbol will only interfere with itself.
OFDM transmissions allow ISI within an OFDM symbol, but by including larger guard band.
Circular Convolution and the DFT
After OFDM symbols have been rendered orthogonal with guard interval, the next task is to remove ISI
within each OFDM symbol.
Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 1

When an input data stream x[n] is sent through a linear invariant FIR channel h[n], the output is a linear
convolution of the input and the channel, i.e y[n] = x[n] * h[n].
y[n] in terms of circular convolution is given by y[n] = x[n] * h[n] = h[n] * x[n]
where x[n] * h[n] = h[n] * x[n] ∑ [ ] [ ]
and the circular function x[n]L = x[nmodL] is a periodic version of x[n] with period L.
Here it would be possible to get DFT of the channel output y[n]: DFT{y[n]} = DFT{h[n] * x[n]}
In frequency domain: Y[m] = H[m] X[m]
The L point DFT is defined as DFT {x[n]} = X[m] 1/√ ∑ [ ]
While its inverse IDFT is defined as IDFT {X[m]} = x[n] 1/√ ∑ [ ]

This formula describes an ISI free channel in the frequency domain, where each input symbol X[m] is
[ ]
scaled by complex value H[m]. X[m] =
[ ]
The Cyclic Prefix
OFDM utilizes FFT algorithm for computing DFT, and IFFT algorithm for computing the IDFT, that
reduces number of multiplications and addition from O( L2 ) to O(L log L). For FFT/IFFT to create an
ISI free channel, the channel must appear to provide a circular convolution as shown in below fig. If a
cyclic prefix is added to the transmitted signal as shown, then it creates a signal that appears to be x[n]L
and so y[n] = x[n] * h[n].
If maximum channel delay spread has a duration of x+1 samples, then by adding guard band of at least υ
samples between OFDM symbols. Representing such OFDM symbol in time domain as a length L
vector gives x = [x1+x2+x3+…xL].
After applying a cyclic prefix of length υ the actual transmitted signal is

The output of the channel is obtained by definition Ycp = h*Xcp, where is h is a length υ+1 vector
describing the impulse response of the channel during the OFDM symbol.
The output Ycp has (L+ υ) +( υ +1) – 1 = L+ 2υ samples.
Our claim is that these L samples of y will be equivalent to y = h* x
Consider for moment y0 i.e first element in y , as shown in below fig due to cyclic prefix y0 depends on
x0and the circularly wrapped values xL-x…xL-1 that is
A cyclic prefix that is at least as long as the channel duration allows the channel output y to be
decomposed into a simple multiplication of channel frequency response H = DFT{h} and the channel
frequency domain input X = DFT{x}.

Since the redundant symbols υ are sent, the required bandwidth for OFDM increases from B to .
Hence a cyclic prefix carries a power penalty of 10log dB in addition to bandwidth penalty .In brief
use of cyclic prefix entails data rateand power loss that are both
Rate loss = power Loss = .
An OFDM Block Diagram
1. First step in OFDM is to break a wideband signal of bandwidth B into L narrowband signals,
each of bandwidth B/L. Likewise the symbol rate is maintained, But each subcarrier experiences

flat fading as long as a cyclic prefix that exceeds the delay spread. The L subcarriers for a given
OFDM symbol are represented by a vector X, which contains L current symbols.
2. The subcarriers are created using IFFT operation, to use a single wideband radio instead of L
independent narrow band radios.
3. In order for the IFFT/FFT to decompose the ISI channel into orthogonal subcarriers, a cyclic
prefix of length υ must be appended after IFFT operation. The resulting L+ υ symbols are then
sent in serial through the wideband channel.
4. At the receiver, the cyclic prefix is discarded and the L received symbols are demodulated using
an FFT operation, which results in L data symbols, each of the form YL = HLXL + NL for
subcarrier L.
5. Each subcarrier can then be equalized via an FEQ by simply diving by the complex channel gain
H[i] for that subcarrier. This results XL = XL +NL / HL.
OFDM in LTE
The above figure shows an close-up view of passband OFDM modulation engine.
 The inputs are L independent QAM symbols and these L symbols are treated as separate
subcarriers.
 These L data-bearing symbols can be created from a bit stream by a symbol mapper and
serial to parallel converter.

 The L point IFFT then creates a time domain L-vector X that is cyclic extended to have
length L(1+G), where G is fractional overhead.
 In LTE G =0.07 for normal cyclic prefix and grows to G = 0.25 for the extended cyclic
prefix. This longer vector is then P/S converted into a wideband digital signal that can be
amplitude modulated with a single radio at a carrier frequency of fc = ωc/2Π.
 If 16QAM modulation was used (M=16) with the normal cyclic prefix the raw data rate
of this LTE system would be
The key OFDM parameters are summarized in the table below.
Timing and Frequency Synchronization

There are two important synchronization tasks for demodulate an OFDM signal, timing synchronization
and frequency synchronization. Timing synchronization requirements for OFDM are relaxed.

The timing offset of the symbol and the optimal timing instants need to be determined is referred to as
Timing synchronization. The receiver must align its carrier frequency as closely as possible with the
transmitted carrier frequency; this is referred to as frequency Synchronization. Frequency
synchronization requirements are more stringent, as the orthogonality of data symbols is reliant on their
being individually discernible in frequency domain. The below fig shows a representation of an OFDM
symbol in time and frequency.
Fig:
OFDM synchronization in time and frequency, here two subcarriers in time domain and eight subcarriers in the
frequency domain are shown, where fc = 10MHz and subcarrier spacing Δf = 1Hz.
In time domain, the IFFT effectively modulates each data symbol onto a unique carrier frequency. In above figure
only two of the carriers are shown, the actual transmitted signal is the superposition of all the individual carriers.
Frequency response of each subcarrier becomes a “sinc” function with zero crossings every 1/T = 1MHz. Using
Fourier trasnsform F{.}
f{cos(2Πfct).rec(t/T)} = F{cos(2Πfct)} * f{rec(2t/T)}
= sinc(T (f –fc))
Where rect (x) = 1, x ϵ (-0.5,0.5) and zero. The frequency response is shown for L = 8 subcarriers.
Timing Synchronization
Due to the presence of a cyclic prefix timing errors in symbol synchronization is relaxed in OFDM. Here we
assume that only L time domain samples after the cyclic prefix were utilized by the receiver. In this case even if
the cyclic prefix length Ng is equivalent to the length of the channel impulse response υ, OFDM symbols can be
decoded ISI free.
It can sustain a timing offset of η seconds without any degradation in performance as long as 0 ≤ η ≤ Tg – Tm ,
where Tg id guard time and Tm maximum channel delay spread .
η < 0 corresponds to sampling error ; η >0 is an ideal instant. As long as 0 ≤ η ≤ Tg – Tm, the timing offset results
in a phase shift per subcarrier of exp(- jΔfη) .As long as η remains constant, the channel estimator includes it as

part of fixed phase offset and it can be corrected by the FEQ without any loss in performance. This acceptable
range of η is referred to as timing synchronization margin shown in below fig
If the timing offset η is not within the window 0 ≤ η ≤ Tm – Tg, inter-symbol interference occurs.
For η > 0 the receiver losses some of the desired energy and also incorporates undesired from the
subsequent symbol.
Fir η < Tm – Tg the desired energy is lost while interference from the preceding symbol is included.
For both of these scenarios, the SNR loss can be given as
2
ΔSNR = - 2( . observations from this expressions are:
 SNR decreases quadratically with the timing offset.

 Longer OFDM symbols are more immune from timing offset.
 As η < <LTs, timing synchronization errors are not that critical as long as induced phase change
is corrected.
For LTE, primary and secondary synchronization channels are used for time and frequency
synchronization. The primary and secondary sync signals occupy the 31 tones on either side of the DC
subcarrier in the 5th and 6th OFDM symbols, respectively of the 0th and 10th slot in every subframe. The
five adjacent tones on “top” and “bottom” of the sync signals are not used. So 62 tones in all are used for
this purpose.
Frequency Synchronization
OFDM has a high degree of bandwidth efficiency compared to other wideband systems. The frequency
offset is not always zero, because of mismatched oscillators at the transmitter and receiver and Doppler
frequency shift due to mobility. Frequency offset is essential in a consumer OFDM system like LTE.

Let us analyze ICI in order to understand its effects on OFDM performance.
The matched filter receiver of subcarrier l can be simply expressed for the case of rectangular window as
where 1/LTs = Δf and again LTs is the duration of the data portion of OFDM symbol i.e T = Tg + LTs.
An interfering subcarrier m can be given as
If the signal is demodulated with a fractional frequency offset of δ , δ ≤ ½
The ICI between subcarriers l and l+m using a matched filter is given as
In the above expression as δ = 0 = Im = 0 and m= 0 = Im = 0. The total average ICI energy per symbol
on subcarrier l is then given as
Where C0 is a constant depends on various assumptions and εx is the average symbol energy. Since the
interference falls with m , this assumption is very accurate for subcarriers near the middle of the band,
and is pessimistic by a factor of 2 at either end. The SNR induced by frequency offset is given by

Observation from the above equation i.e. ICI expression and below fig are :
 SNR decreases quadratically with the frequency offset

 SNR decreases quadratically with the number of subcarriers.
 The loss in SNR is also proportional to the SNR itself.
 In order to keep loss negligible ,the relative frequency offset need to be about 1 to2% of the
subcarriers.
 This is a case where reducing CP overhead by increasing the number of subcarriers causes an
offsetting penalty.
The Peak to Average Ratio [PAR]

OFDM signals have higher PAR often called a peak-to-average power ratio [PAPR], because in time
domain, a multicarrier signal is the sum of many narrowband signals. Sometimes this might be large or
small, i.e peak value of the signal is larger than average value. This is one of the implementation
challenges in OFDM, as it reduces efficiency and increases the cost of RF power amplifier.
The PAR problem

When a high peak signal is transmitted through a non linear device such as high power amplifier(HPA)
or DAC, it generates out of band energy and in-band distortions these may affect the performance of the
system. The below figure shows HPA with the associated input backoff(IBO) and output backoff(OBO)
regions respectively.
A waveform with high-peak power must be transmitted in the linear region of HPA by decreasing the
average power of the input signal and results in proportional backoff.
IBO = 10log10 Pinsat /Pin! where Pinsat is the saturation power and Pin!
The power efficiency of an HPA can be increased by reducing the PAR of the transmitted signal. High
PAR signal requires high resolution for both the transmitters DAC and the receivers ADC, since the
dynamic range of the signal is proportional to PAR.
Quantifying the PAR
As multicarrier systems transmit data over number of parallel frequency channels, results in
superposition of L narrowband signals. Each of L output samples from an L-point IFFt operation
involves sum of L complex numbers. As per central limit theorem, the resulting output values can be
modeled as complex Gaussian random variables with zero means and variance ζ2 = εx/2.
The amplitude of the output signal is

Since x[n] is complex Gaussian for narrowband fading, the output power is therefore
Which is exponentially distributed with parameter ζ2. Here power is random, So PAR is not
deterministic .The PAR of the transmitted signal can be defined as
The PAR is considered for a single OFDM symbol, that consists of L + Ng ,T. Discrete PAR can be
defined for the IFFT output as
The average energy of IFFT outputs x[n] is the same as the average energy of inputs X[m] and equal to
ε, the analog PAR is not exactly the same as the PAR of the IFFT samples due to interpolation
performed by D/A converter. Analog PAR is higher than digital PAR, to bring them closer,
oversampling can be considered for digital signal.
Since the theoretical maximum PAR value seldom occurs and the PAR is a random variable. The
complementary cumulative distribution function (CCDF) of the PAR is most commonly used measure.
The CCDF of PAR for QPSK OFDM system is shown below.

Fig: The CCDF of PAR for QPSK OFDM system L = 16, 64, 256, 1024
Simple and accurate approximation of the CCDF for large L ( ≥ 64)
Where εmax is the peak power level and β pseudo approximation of the oversampling factor. The CDF of
the Nyquist sampled signal power can be obtained by
G(L, εmax) = P(max ‖ X(t) ‖ εmax ) = F (L, εmax)L
Clipping and other PAR techniques

To avoid operating the PA in the nonlinear region, the input power can be reduced to PAR.Two
important facts about IBO are:
1. Since the highest PAR values are uncommon, it might be slightly clip off the highest peaks.It can
be seen that even for a conservative choice of IBO .
2. There is a probability that a given OFDM symbol will have a PAR that exceeds IBO and causes
clipping. Clipping also called as “, “soft limiting” that exceeds the clipping level as

Where x[n] is the original signal and x[n] is the output after clipping and A is clipping level. The
Clipping ratio can be defined as
Clipping reduces the PAR at the expense of distortion of the desired signal .
Two important drawbacks of clipping are: (i) Spectral regrowth, which causes unacceptable interference
to users in neighbouring RF channels. (ii) Distortion of the desired signal.
Spectral Regrowth:
The clipping noise in frequency domain can be given as
Xk = Xk + Ck, K = 0,1,……L – 1,where Ck represents the clipped off signal in frequency domain.
The below figure shows the power spectral density of the original, clipped and clipped-off signals are
plotted for different clipping ratio of 3, 5 and 7dB.
The following effects observed :

1. The clipped-off signal Ck is strickingly increased as the clipping ratio is lowered from 7dB to
3dB, which increases correlation between Xk and Ck and lowers clipping ratio.
2. The out-band interference caused by the clipped signal X is determined by the shape of the
clipped-off signal Ck.
In-band Distortion
Even though desired signal and clipped signal are correlated, based on Bussgang Theorem to model in-
band distortion.
Attenuation of the desired signal is given as x[n] = ax[n] + d[n] for n = 0,1,….L-1. Now d[n] is
correlated with the signal x[n] and attenuation factor α is obtained by
The attenuation factor α is plotted in below fig, as a function of clipping ratio α.
The attenuation factor α is negligible when the clipping ratio γ > 8dB.
To estimate the impact of the clipped OFDM signals over an AWGN channel, assuming distortion d[n]
is Gaussian and correlated with the input and channel noise
The bit error probability (BEP) can be evaluated for different modulation types using the SNDR. In case
of M-QAM and the average power εx, the BEP can be

The below figure shows the BER for an OFDM system with L = 2048 subcarriers and 64QAM
modulation. As SNR increases, the clipping error dominates the additive noise and error floor is
observed.
Single-Carrier Frequency Domain Equalization (SC-FDE)

SC-FDE is an alternative approach to OFDM. SC-FDE maintains three benefits of OFDM
1. Low complexity even for severe multipath channels.

2. Excellent BER performance.
3. Decoupling of ISI from other types of interference.
System description of SC-FDE
Equalization is used in OFDM and SC-FDE to reduce time complexity. The block diagrams of both
are shown below for comparison.
The difference between two systems is that the IIFT is moved to the end of the receiver rather than
operating at the transmitter, to create a multicarrier waveform as in OFDM.
An SC-FDE system utilizes a cyclic prefix at least as long as the channel delay spread,but now since
the sequence is of QAM symbols that has low PAR.which cannot be lowered much below than that
of SC-FDE. At the receiving mode, FFT is applied in OFDM but in SC-FDE this operation moves
the received signal into the frequency domain.
Y[n] = x[n] *h[n] + ω[n] , where ω[n] is noise , therefore
FFT{y[n]} Y[m] = H[m] X[m] + W[m].

After FFT a simple 1-tap FEQ can be applied that inverts each virtual subcarrier.
[ ]
So that X[m] = the resulting signal can be converted back to time domain using IFFT operation
[ ]
to obtain x[n].
SC-FDE Performance Vs. OFDM
 The main difference in between them is the way way they treat noise.
 In both of the receivers, the FEQ typically inverts each frequency bin, i.e the FEQ consists of
L complex taps each of value 1/HL.
 In OFDM high SNR symbols remain at high SNR and low SNR symbol remains at low SNR.
The discrepancies between the SNR on each carrier can be handled by coding interleaving or
pre-subcarrier adaptive modulation.
 In LTE short SNR variations are addressed by coding and interleaving. In SC-FDE, the FEQ
does not operate on data symbols but rather on frequency domain dual of data symbols.
 Here low SNR parts of the spectrum have their power increased by a factor of (1/HL)2 and
also noise power is increased by a factor of (1/HL)2 .
 So total noise amplification factor is same in OFDM and SC-FDE.
Related to performance OFDM slightly outperforms SC-FDE. Primary determining factors are
strength of coding and the size of the constellation. OFDM does a bit better when the coding is
strong and and /or constellation is large, while the opposite is true for SC-FDE. Higher order
constellations are more sensitive to even modest amount of noise, which limits SC-FDE as it
subjects all its symbols to amplified noise.
Design Considerations for SC-FDE and OFDM

SC-FDE has a lower complexity transmitter but a higher complexity receiver compared to OFDM.IN
a cellular system, this asymmetry can be featured, since the uplink could utilize SC-FDE and the
downlink could utilize OFDM. In such situations the base station would perform 3IFFT/FFT
operations and the mobile is more power and cost sensitive.SC-FDE benefits of reduced PAR, cost
and power saving.
Channel estimation and synchronization are different for an SC-FDE systems and an OFDM system.
In OFDM channel estimation and synchronization are accomplished by preambles of known data
symbols and pilot tones.SC-FDE also use preamble, which is in time domain, so it not possible to
estimate frequency domain values HL.
Another disadvantage of SC-FDE is it has more dispersive spectrum compared to OFDM. OFDM „s
sharper spectrum results in less co-channel interference or less RF roll-off requirements.
Finally the combination of SC-FDE with MIMO is not as natural because detection cannot be done
in frequency domain. Hence it is possible to use maximum likelihood detection for MIMO with SC-
FDE.
The computational Complexity Advantage of OFDM and SC-FDE
Advantage of frequency domain equalization to time domain equalization is the FDE. A time
domain equalizer consists of a series of multiplications with several delayed versions of signal.
Number of delay taps depends on the symbol rate of the system and delay spread of channel. An
equalizer with η taps performs υ complexity multiple and accumulate (CMAC) operations. The
complexity of an equalizer is of the order O (υ.B) = O (B2Tm).
The IFFT and FFT are computational operations in an OFDM or SC-FDE.IFFT and FFT each have a
complexity of O(L log2 L), where L is FFT block size. In OFDM , L is number of subcarriers. The
number od subcarriers L must grow linearly with the bandwidth delay spread υ = BTm. Since L
BTm , means there are order O(1/Tm) OFDM symbols per second. So computational complexity in
terms of CMAC‟S for OFDM is O(BTmlog2BTm)O(/Tm) = O (B log2 BTm). The Complexity of a
time domain equalizer grows as the square of the data rate since both the symbol rate and the number
of taps increases linearly with the data rate.

OFDMA and SC-FDMA: OFDM with FDMA

Multiple Access for OFDM systems.
 OFDM is not a multiple access strategy, but a technique for mitigating frequency selectivity.
 OFDM creates many parallel streams of data, used by different users.
 In previous OFDM systems all subcarriers are used by single user at a time such as 802.11a/g
(WiFi) and 802.16/WiMax systems.
 In these single user OFDM systems, multiple users can be accommodated by simply time
sharing the channel.
Multiple Access Overview
 Multiple access strategies provide non interfering, orthogonal communication channels

for each active subscriber link.
 The available dimension can be divided into frequency, time and code division
multiplexing.
 In FDMA each user receives a unique carrier frequency and bandwidth.
 In TDMA each user is given a unique time slot.
 Orthogonal CDMA systems allow each user to share the bandwidth and time slot with
many other users.
 In an additive noise channel, TDMA, FDMA & orthogonal CDMA systems have same
theoretical capacity.
 Eg: assume it takes one unit of bandwidth to send a user‟s signal and eight units of
bandwidth are available.
 In FDMA, eight orthogonal frequency slots would be created, one for each user.
 In TDMA, each user would use all eight frequency slots, but only transmit one-eight of
the time.
 In CDMA systems each user would transmit all of the time over all of the frequency, but
would use one out of eight available orthogonal codes to ensure that there was no
interference with the other seven users.
Random Access Vs. Multiple Access
 Random access technique known as Carrier Sense Multiple Access (CSMA) commonly
used in packet-based communication systems such as Ethernet and wireless LANs.
 In Random access, user contend for the channel rather then being allocated a reserve time
frequency or code.
 Random access techniques include ALOHA and slotted ALOHA and also CSMA.
 In ALOHA, users simply transmit packets without regard to other users. After some time
period if packet is not acknowledge by the receiver, it is assumed lost and is
retransmitted.
 This technique is highly inefficient and delay prone as the traffic intensity increases.
 Slotted ALOHA has better aspects, by a factor of two since users transmit on specified
time boundaries hence collisions are nearly half.
 CSMA improves using ALOHA and slotted ALOHA through carrier sensing, where in
users “listen” to the channel before transmitting to avoid collisions whenever possible.
 A well known algorithm for CSMA is Distributed Coordination Function (DCF) of
802.11, where in user waits for a random time intervals before transmitting, in order to
reduce the probability of two stations transmitting immediately.
 Theoretical efficiency of CSMA is around 60-70% in wireless LANs.
 In FDMA and TDMA voice systems around half the bandwidth is wasted. This is the
major reason the CDMA has proven so successful for voice.
OFDM-FDMA
 FDMA can be implemented in OFDM systems by assigning different users with their
subcarriers.
 Simple method of allocation is a static allocation of subcarriers to each user as shown on
left of below fig.
 Eg: In a 64 sub carrier OFDM system, user 1 could take subcarriers 1-16 with users 2, 3,
and 4 using subcarriers 17-32, 33-48, and 49-64 respectively.
 The allocations are done using multiplexer before IFFT operation. Such a system can be
referred to as OFDMA as it allows multiple users to share the OFDM subcarriers.
 OFDMA in LTE has explicit time-sharing and procedures to allow for the dynamic
allocation of subcarriers.

OFDM-TDMA
 TDMA systems employ both FDMA and TDMA, since the electromagnetic spectrum must
be shared with many other users.
 Static TDMA methodology shown in right of above fig is appropriate for constant data rate
applications like voice and video streaming.
 A packet based system like LTE can employ more scheduling algorithms based on queue-
lengths, channel conditions and delay constraints to obtain better performance than static
TDMA.
OFMD –CDMA or MC-CDMA
In CDMA, bandwidth is much larger than the data rate used to suppress the interference.
CDMA broadband standards such as HSDPA and 1XEVDO3 have very small spreading
factor and allow each user to employ multiple codes at the same time.
OFDM and CDMA can be combined to create a Multicarrier CDMA (MC-CDMA)
waveform.
We can use spread spectrum signaling to separate users by codes in OFDM by spreading in
either time or frequency domain.
Time domain spreading entails each subcarrier transmitting the same data symbol on several
consecutive OFDM symbols.
Frequency domain spreading, has better performance than time domain, entails each data symbol
being sent simultaneously on N different subcarriers.

Orthogonal Frequency Division Multiple Access (OFDMA)
 OFDMA systems allocate subscribers time- frequency slices consisting of M subcarriers over
number of consecutive OFDM symbols in time.
 The M subscribers can either be
1) Spread out over the band, often called a “distributed” or “diversity” allocation
2) Bunched together in M contiguous subcarriers, often called a “band AMC”.
 The band AMC mode, attempts to use subcarriers where the SINR is roughly equal.
 If accurate SINR information can be obtained at the receiver about each band‟s SINR, then
band AMC outperforms distributed subcarrier allocation.
Below table summarizes the notation used here after

How OFDMA Works:
The basic concept is similar to an OFDM system except for K users share the L subcarriers, with
each user being allocated Mk subcarriers. The method for how users are mapped to subcarriers is not
specified by the LTE standard. In theory it is possible to have users share subcarriers, but this never
happen practically. So k Mk = L and each subcarrier has one user assigned to it.
At each receiver, the user cares only about its own Mk subcarriers, but still has to apply an L point
FFT to the received digital waveform. To extract the correct subcarriers, the receiver has to know
which time- frequency resources have been allocated. Thus OFDMA downlink receivers demodulate
the entire waveform. Here power is wasted, but digital separation of users is simple to enforce at the
receiver and also residual interference is very low compared to either CDMA or conventional
FDMA.
Even though OFDMA is not used in the LTE uplink, The below figures shows OFDMA uplink
block diagrams, to show the differences and several similarities between OFDMA & SC-FDMA.

The transmitter modulates user k‟s bit over the Mk subcarriers. Here we have taken Mk = M for all
users. All the users signals collide at the receiver‟s antenna and gets demodulated using receivers
FFT. Here we should be noted that uplink OFDMA is more challenging than downlink OFDMA as
the uplink is naturally synchronous, but this is not the case in downlink, since the transmitter is
common for all users.
In LTE the up-link multi-access scheme uses only the localized subcarrier mode due to SC-FDMA
uplink nature. The lack of frequency and time synchronization between multiple users leads to
inter-carrier interference but this is limited to subcarriers at the edge of the transmission band of each
user. Higher level view of OFDMA is shown below: where a base station is transmitting a band
AMC-type OFDMA waveform to four devices simultaneously. The three arrows for each user
indicate the signaling.

 First, the mobiles measure and feed back the quality of their channel state information (CSI)
to the base station.
 Usually, the CSI would be measurement corresponding to SINR.
 The base station would then allocate subcarriers to the four users and send that subcarrier
allocation information to the four users in an overhead message.
 Finally the actual data is transmitted over the subcarriers assigned to each user.
OFDMA Advantages and Disadvantages
The advantages of OFDMA start with the advantages of single user OFDM in terms of
multipath suppression, low complexity and the creation of frequency diversity.
[Frequency diversity means using two or more widely spaced frequency channels to send
the same message, usually at the same time. The idea is that channel propagation and
interference issues will not affect all frequencies to the same extent, so at least one signal
will be received with acceptable SNR].
OFDMA is a flexible multiple access technique that can accommodate many users with
widely varying applications, data rates and QoS requirements.
Multiple access is performed in digital domain allows time and frequency domain
scheduling algorithms.
Lower data rates and burst data are handled much more efficiently in OFDMA than in
single user OFDM.
In case of voice example, If OFDMA was not used, each downlink user would receive
very high rate signal for a very short period of time.

Disadvantages : This could require the receiver to process a large amount of data and have bad
latency and jitter properties as voice decompressor has to wait before decoded bits were
available.
Switching between users has to be very rapid, more signaling would be needed that
reduces overall system throughput.
In uplink OFDM-TDMA would be more toxic, the subscribers have to transmit a
wideband signal at very high power for a short time.
OFDMA doesn‟t suffer from these issues since de allocation of time-frequency resources
are extremely flexible and can be employed to meet throughput, delay and other QoS
requirements.
Single Carrier Frequency Division Multiple Access (SC-FDMA)

SC-FDMA is employed in LTE uplink; concept evolves from SC-FDE.
Goal of SC-FDMA is to take low peak-to-average properties of SC-FDE and achieve
them in an OFDMA type system that allows partial usage of frequency band.
SC-FDMA: How it Works
An SC-FDMA uplink transmitter is shown below, which is similar to OFDMA uplink

transmitter. The only difference is users Mk symbols are pre-processed with an FFT of size Mk.
In LTE, Mk is related to the number of resource blocks allocated to the user K for its uplink
transmission. The FFT operation creates a frequency domain version of the signal
X[m] = FFT(x[n]), so that when an L point IFFT is applied later. The time domain outputs of the
IFFT correspond to an oversampled and phase-shifted version of original time domain signal
x[n].

The SC-FDMA uplink receiver is shown in above figure, which is similar to OFDMA uplink
receiver. The difference is that for each user‟s Mk subcarriers, an additional small IFFT must be
applied before detection, to get receiver data back into time domain. Here also frequency domain
equalization is applied to each user‟s signal independently after FFT, and signals are demapped
based on current subcarrier allocation.
SC-FDMA Advantages and Disadvantages: The rationale for SC-FDMA is twofold.
First, the key advantage of OFDMA is preserved: only part of frequency spectrum is used
by one user at a time. This allows the band used to be chosen adaptively for higher
throughput.
The second rationale is that the PAR of SC-FDMA is lower than OFDMA. Because the
transmitted SC-FDMA signal for each user is an oversampled single-carrier signal.
The tradeoffs between SC-FDMA and OFDMA are closely related to the tradeoffs SC-
FDE faces versus OFDM.
SC-FDMA can experience more spectral leakage then OFDMA and obtain frequency
diversity.
SC-FDMA has a complexity disadvantage versus OFDMA in both the transmitter and
receiver as an additional FFT of size Mk has to be performed for each user at the
transmitter and receiver.
OFDMA and SC-FDMA in LTE
 OFDMA systems must specify things in order for the system to work.
 First, it must specify the “quanta” or units.
 Second, it must specify messaging protocols for both transmission and reception.
 Third, ranging procedures must be specified so that simultaneous uplink transmissions
from several different mobile units can be decoded at the base station.
Let‟s have an overview how LTE accomplishes these 3 tasks.

The LTE Time-Frequency Grid

In LTE, mobile units are allocated group of subcarriers over time and frequency known as
resource block. Any subcarrier can be assigned to any user at any time slot, but it may take
large overhead to specify current allocation to all mobile units. Lower overhead can be
achieved by an OFDM-TDMA system, where one user has access to all the subcarriers and
uses same AMC level. A typical resource block consists of 12 subcarriers over 7 OFDM
symbols.
Subcarriers of resource block are allocated in two ways.
 Distributed subcarrier allocation & adjacent subcarrier allocation.
Distributed subcarrier allocation: Here resource block hop across the entire channel
bandwidth. This is obtained by using a “comb” pattern at any given point of time for a given
user. This approach is used in downlink, when distributed subcarriers allocation is used.
Frequency diversity can be achieved by hopping a contiguous block of subcarriers in time.
Example, 12 subcarriers in a resource block could hop to a different part of the spectrum over
each OFDM symbol utilized. Frequency diversity is achieved as long as sufficient
interleaving is employed.
Adjacent subcarrier allocation: This approach depends on a channel-aware allocation of

resources, so that each user can be allocated to resource block that have strong channel. Since
a block of 12 subcarriers is smaller than the coherence bandwidth of the channel. Frequency
diversity is not achieved as long as the scheduler is able to assign “good” blocks to each user.
Allocation Notification and Uplink Feedback

Similar to UMTS standards overhead signaling is done on a logical control channel. Here the
Physical downlink control channel (PDCCH) specifies the following:
 Downlink resource block allocation
 Uplink resource block allocation
 QAM constellation to use per resource block
 Type and rate of coding to use per resource block.
 The PDCCH is sent over first 2 or 3 OFDM symbols of each sub frame across all the
subcarriers.
 Resource block frame consists of 168 subcarriers over 14 OFDM symbols.
 Here about 14-21% of total downlink capacity is used by the PDCCH.
 In the downlink, the BS has knowledge of buffered data for each user, while in the uplink
it can estimate the channel from each user.
 Hence BSR feedback is only used for uplink scheduling while channel quality
information (CQI) feedback is only used for downlink scheduling.

Power Control
OFDMA systems suffer from two forms of self-interference..
 The first is inter cell interference, where neighboring cells allocate the same time-frequency
resource blocks and hence cause interference. This can occur in uplink or downlink. This inter
cell interference can be problematic for cell-edge users in the downlink.
 The second form of inter cell interference is related to imperfect time-frequency power
synchronization between multiple uplink users. This is not a problem in downlink, as each
base station receives a single waveform from the base station. In the uplink, the received
waveform is aggregated in the air.
 In LTE, closed loop power control is possible in the uplink; where the base station can
explicitly indicate the maximum transmit power density.
 Fractional power control, where the channel is partially inverted, i.e the transmit power is
proportional to h-8 where s is a fractional value between 0 and 1. Fractional power control is
the open-loop power control scheme in LTE.
Multiple Antenna Transmission and Reception

Multicarrier modulation enables richer, more efficient use of multiple antennas and receivers in
wideband channels. Multiple antennas can be grouped into three categories: diversity,
interference suppression and spatial multiplexing.
Spatial diversity allows a number of different versions of the signal to be transmitted/ received
and provides resilience against fading.
Interference suppression uses the spatial dimensions to reject interference from other users either
through array processing such as linear precoding, post coding or interference cancellation.
Spatial multiplexing allows two or more independent streams of a data to be sent simultaneously
in the same bandwidth. All these three approaches are collectively referred to as multiple input-
multiple output (MIMO) communication.
Spatial diversity : Primary advantage of spatial diversity is that no additional bandwidth or power.
Instead, spatial diversity is exploited through two or more antennas. The cost and space consumed by
each additional antenna, its RF transmit and its associated signal processing operation for modulate or
demodulate multiple spatial streams may not be negligible.
Array gain
When multiple antennas are used, there are two forms of gain available: Diversity gain and Array gain.
Array gain does not rely on statistical diversity between different channels, but it achieves performance
enhancement by coherently combining energy of each antenna. Due to array gain, even if the channels
are completely correlated the received SNR increases linearly with number of receive antennas, Nr.

For a Nt x Nr system the array gain is Nr. In correlated flat fading, each antenna i ϵ (1, Nr) receives a
signal, characterized as yi = hix + ni = hx + ni where hi = h for all antennas as they are perfectly
correlated. Hence γi = ∣h∣ /  where noise power is  and we assume that unit signal energy E∣x∣ .
2 2 2 2
If all the receive antenna paths are added, the resulting signal is y = ∑ = Nr hx + ∑
and the combined SNR is γ = ∣Nrh∣ / Nr  Nr∣h∣ /  . Hence the received SNR also increases
2 2= 2 2
linearly with the number of receive antennas.
Diversity Gain
Objective of spatial diversity is to improve the communication reliability by decreasing the sensitivity to
–C γ
fading. In an additive noise, the bit error probability (BEP) can be Pb ≈ C1e 2 where C1 & C2 are
constants and γ is the received SNR. The error probability is exponentially decreasing with SNR.
-1
Without diversity, the average BEP decreases very slowly in a fading channel can be be Pb ≈ C3 γ . If
Nt transmit antennas and Nr receive antennas are added to the system, the diversity order is Nd = NrNt.
As the probability of all Nd uncorrelated channels having low SNR is very small, the diversity order has
a effect on system reliability.
– γ Nd
With diversity, BEP improves to be Pb ≈ C4e . Example, if the BEP without any diversity was
about 1 in 10, the BEP with two antennas at both the transmitter and receiver would be closer to 1 in
10,000. Diversity gain is powerful.
Increasing the data rate with Spatial Diversity: The system reliability increases as the diversity
techniques are very effective at averaging out fading. Diversity techniques also increase received SNR
due to array gain. The Shannon capacity formula gives maximum achievable data rate of a single
communication link in AWGN as : C = B log2 (1 + γ). As antenna diversity increases the SNR linearly,
diversity techniques increase the capacity on logarithmically w.r.t number of antennas.
Increased Coverage or reduced Transmit Power: Diversity benefits can also be used to increase the
coverage area and to reduce required transmit power. Due to the array gain, the average SNR is
approximately Nr γ, where γ is average SNR per branch.
Receive Diversity : The most prevalent form of spatial diversity is receive diversity, with just two
antennas. Nr = 2 for most of the cellular base stations and wireless LAN access points and is mandatory
for LTE handsets. The widely used combining algorithms are selection combining (SC) and maximal
ratio combining (MRC). We shall focus on flat fading scenario, where signal received by each of the
Nr antennas is uncorrelated and each has same average power.
Selection combining (SC) : Simplest form of “combiner” which estimates strengths of each of the Nr
streams and selects the highest one. SC‟c simplicity and reduced hardware and power requirements
make it attractive for narrow band channels.

In a wideband channel, different coherent bands will have different SNR‟s and selection diversity can be
used on each band.
Fig: Receive diversity – Selection combining
The diversity gain from employing selection combining can be confirmed by considering the outage
probability, defined as the probability that the received SNR drops below threshold, Pout = P [γ  γ0] = p.
Assuming Nr uncorrelated receptions of the signal.
-γ /γ
For Rayleigh fading channel, p = 1 – e 0 where γ is the average received SNR at that
location. Thus selection combining decreases the output probability by
The average received SNR for Nr branch SC can be derived in Rayleigh fading as
Hence, each added antenna does increase the average SNR. Plot of the BEP with different amounts of
selection diversity are shown in below fig even though the performance improvement with Nr reduces,
the improvement for first few antennas is substantial.

Maximal ratio combining (MRC): MRC combines the information from all received branches to
maximize the ratio of SNR. MRC works by weighting each branch with a complex factor qi = ∣qi∣e jɸi
and then adding upto Nr branches as shown in below fig.
Fig: Receive diversity – Maximal ratio combining
The received signal on each branch can be given as x(t)hi, assuming fading is flat with a complex value
jθ
of hi = ∣hi∣ e i on ith branch. The combined signal then be given as
If ɸi = -θi for all branches, then SNR of y(t) is given as
Where εx is the transmit signal energy. Branches with better signal energy should be enhanced, whereas
branches with lower SNRs should be less weight. The resulting SNR can be

Even though MRC maximize SNR and performs well, it may not be optimal in many cases as it ignores
interference power. Equal gain combining (EGC) only corrects the phase, ∣qi∣ = 1 and ɸi = -θi ,
achieves an SNR of
The difference between above two equations is that EGC occurs a noise penality of Nr for not requiring
channel gain estimation.The BEP performance of SC & MRC is shown in below fig
Even though the BEP slopes are similar to selection combining as the techniques have same diversity
order. The SNR gain is several dB due to its array gain.
Transmit Diversity
Transmit diversity is a recent development in early 2000s. Because the signals sent from different
transmit antennas interfere with one another at the receiver. Additional signal processing is needed at
both transmitter and receiver to achieve diversity while attenuating spatial diversity. Transmit diversity
is useful in the downlink as the base station can accommodate more antennas than mobile station.
Multiple antennas transmit schemes are categorized into two classes: Open-loop and Closed –loop.
Open-loop system that do not require knowledge of the channel at the transmitter as shown in below fig
Open-loop systems require channel knowledge at the transmitter.

Open-Loop Transmit Diversity: 2 x 1 Space- Frequency Block Coding [SFBC’s]
The most popular open-loop transmit diversity scheme is space-time coding, where a particular code
known to receiver is applied at the transmitter. Space coding was first suggested in early 1990s. There
are many types of space-time codes. Here we discuss on Space- Frequency Block Coding [SFBC’s]. In
1990s the SFBC is referred to as either Alamouti code or orthogonal space-time block code (OSTBC).
This code has become popular means of transmit diversity due to its easy implementation, Optimality ,
adaptability to wideband fading channel by using adjacent subcarriers rather than consecutive symbols.
SFBC are preferred to STBCs as they have less delay. The simplest SFBC code consists of two transmit
antennas and a single receive antenna. If two symbols s1 and s2 are to be transmitted, the Alamouti code
sends the following two subcarriers f1 and f2.
Antenna 1 2
Subcarrier f1 s1 s2
f2 -s2* s1 *
The 2 x 1 Alamouti SFBC is referred to as a rate 1 code, as data rate is neither increased or decreased:
two symbols are sent over two adjacent subcarriers. Goal of SFBC is to harness the spatial diversity of
the channel.
Assume a flat fading channel on each subcarrier, then h1(f1) is the complex channel gain from transmit
antenna 1 to the receive antenna and h2(f2) is from transmit antenna 2.
Additional assumption is that channel is constant over two adjacent subcarriers that is h1(f1) = h2(f2) = h1.
The received signal r(f) can be given as r(f1) = h1s1 + h2s2 + n(f1) and r(f2) = - h1s2* + h2s1* + n(f2).
Where n(.) is a sample of white Gaussian noise.
Diversity combining scheme can be used assuming channel is known at the receiver :

y1 = h1*r(f1) + h2r* (f2)
y2 = - h2*r(f1) – h1r* (f2)
Hence it can be seen that
y1 = h1*(h1s1 + h2s2 +n(f1)) + h2 (- h1*s2 +h2s1 + n *(f2))
= (∣h1∣2 + ∣h2∣2)s1 + h1*n(f1) + h2n*(f2)
Similarly y2 = (∣h1∣2 + ∣h2∣2)s2 + h2*n(f1) – h1n*(f2)

Hence a simple decoder combines two received samples r(f1) and r*(f2). The SNR can be given as
To have same transmit power as in MRC, each transmit antenna must halve its transmit power so that
total energy per actual data symbol is εx for both cases, i.e for SFBS E∣s1∣ = E∣s2∣ = εx / 2.
2 2
Open- Loop Transmit Diversity with More Antennas
Orthogonal SFBCs like 2x1 Alamouti code do not exist for most combinations of transmit and receive
antennas. Two different techniques proposed are discussed below
2 x 2 SFBC : It uses the same transmit encoding scheme as 2 x 1 transmit diversity, here channel
description is represented as 2 x 2 matrix rather than a 2 x 1 vector.
The resulting signals at subcarriers f1 and f2 on antennas 1 and 2 can be given as
r1(f1) = = h11s1 + h21s2 + n1(f1)

r1(f2) = - h11S2* + h21 s1* + n1(f2)
r2(f1) = h12s1 + h22s2 + n2(f1)
r2(f2) = - h12s2* + h22s1* + n2(f2)

Using the following combining scheme
Yields the following decision statistics
And results in the following SNR
An orthogonal, full-rate, full diversity SFBC over an Nt x Nr channel will provide a diversity gain
equivalent to that of an MRC system with NtNr antennas, with a 10log10 Nt dB transmit power penalty
due to Nt transmit antennas.
4 x 2 stacked STBC
In LTE, it is common to have four transmit antennas at the base station. Here two data streams can be
sent using a double space-time transmit diversity (DSTTD) which consists of 2x 1 Alamouti code in
parallel. DSTTD is also called as stacked STBSs that combines transmit diversity and maximum ratio
combining techniques along with spatial multiplexing as shown in figure.

The received signals at subcarriers f1 and f2 on antennas 1 and 2 can be represented with equivalent
channel model as :
Then, the equivalent matrix channel model of DSTTD can be given as
As shown in above equation, each Hij channel matrix is equivalent channel of Alamouti code. Thus,
DSTTD can achieve a diversity order of Nd = 2Nr due yo 2 x 1 Alamouti code.
If The same linear combining scheme is used as in the 2 x 2 STBC case, then statistics obtained as
Where Ii is the interference from the ith transmit antenna due to transmitting two simultaneous data
streams.

4 x 2 in LTE: When 4 transmit antennas are available, a combination of SFBC and frequency switched
transmit diversity (FSTD) is employed. This combination of SFBC and FSTD is a rate 1 diversity
scheme. Here 4 modulation symbols are sent over four OFDM symbols using the following space-
frequency encoder, where column corresponds to subcarrier index and the rows to the transmit antenna.
The first and second symbols are sent over antenna ports 0 and 2 on the first two OFDM subcarriers in
the block. Likewise the third and fourth symbols are sent using antenna port 1 and 3.This encoder is rate
1 and can be detected using a simple linear ML receiver.
Transmit Diversity Vs Receive Diversity
The Above three examples of STBC showed that transmit and receive diversity can provide enhanced
diversity that increases the robustness of communication over wireless fading channels.
Receive diversity for maximal ratio combining with Nr antennas and only one transmit antenna, the
received SNR grows as antennas are added, and the growth is linear.
The average combined SNR can be obtained as
SNR growth is linear with the number of receive antennas.
Transmit Diversity : Due to the transmit power penalty inherent to transmit diversity techniques, the
received SNR doesn‟t always grow as transmit antennas are added. SNR in an orthogonal STBC scheme
is generally of the form.
as the number of antennas grows, expression becomes

Thus open-loop transmit diversity causes the received SNR to “harden” to the average SNR.
Closed-Loop Transmit Diversity
The basic configuration for closed-loop transmit diversity is shown below. Here receiver could also have
multiple antennas. An encoding algorithm is responsible for using the CSI to effectively use its Nt
available channels. Here we assume that the transmitter has fully accurate CSI available to it due to the
feedback channel.
Transmit Selection Diversity: TSD is the simplest form of transmit diversity. Here only a subset
N* < Nt of available Nt antennas is used.
Advantages of transmit antenna selection are
1) Hardware cost and complexity is reduced
2) Spatial interference is reduced
3) Diversity order is still NtNr even though only N* of Nt antennas are used.
4) Despite its optimal diversity order, transmit selection diversity is not optimal in terms of diversity
gain.
A single transmit antenna is selected, which results in highest between transmit and receive antenna.
Optimum antenna choice is i*: i* = arg max iϵ(1,Nt) ∣h∣ . Hence transmit diversity does not incur the
2
power penalty relative to receive selection diversity.

The average SNR with single transmit antenna selection in Nt x 1 system is
 Lower average SNR can be achieved with beam forming techniques that use all transmit antenna.
 The feedback required for antenna transmits selection.
 Only log2 Nt bits of feedback are needed for each channel realization.
 Example: if there were Nt = 4 transmit antenna and the channel coherence time was
Tc = 10msec, only 1Kbps of channel feedback is needed.
 The required feedback for transmit antenna selection is about N* log2Nt bits per channel
coherence time.
Linear Diversity Precoding
It is a technique for improving the data rate or the link reliability by exploiting the CSI at the transmitter.
The linear precoder at the transmitter and a linear postcoder at the receiver are applied only to improve
the link reliability. For a linear precoding, the received data vector Z can be given as :
z = Gy = G(HFx +n), where the size of transmitted vector(x) is M x 1 and received vector y is Nr x 1.
The postcoder matrix G is M x Nr to give z dimensions of M x 1, while the channel matrix H is Nr x Nt,
the precoder matrix F is Nt x M, The noise vector n is Nr x 1and M is the number of spatial data streams
sent.
For a pure diversity precoding only one data symbol is sent at a time, so M = 1 and the SNR
maximizing precoder G are the right and lefy singular vectors of H corresponding to its largest singular
matrix max. The equivalent channel model after precoding and postcoding for a transmitted data symbol
x becomes y = hx + n which is Nr x 1 & when multiplying by 1 x Nr postcoding vector g gives
z = max. x +nz . The received SNR is
where 2 is the noise variance of nz. The SNR can only be bounded as

Where ∣∣ . ∣∣F denotes the frobenius form, defined as
By generalizing the SNR expression for 2 x 2 STBCs, is given as
By comparing above equations, it is clear that linear precoding achieves a higher SNR than the open-
loop STBCs. When Nr = 1, the full SNR gain of 10log10Nt dB is achieved. To employ linear diversity
precoding, CSI is required at the transmitter.
Interference Cancellation Suppression and Signal Enhancement

In a multichannel system, the channel is multidimensional and channel dimensions can be applied to null
interference in certain direction, while amplifying signals in other direction. The dimensions can be used
to create linear transmitters and receivers that project desired signals into strong dimensions while
attenuating the dimensions where interferes exist.
DOA- Based Beam steering
Here we discuss about beamsteering approaches, where beam patterns gets adjusted to attenuate
undesired signal. The incoming signals to a receiver consist of desired energy and interference energy.
Signals are characterized in terms of direction of arrival (DOA) and angle of arrival (AOA) of each
received signal. Each DOA can be estimated using signal processing techniques such as the MUSIC,
ESPRIT, and MLE algorithms. From these acquired DOAs, a beamformer extracts a weighing factor for
the antenna elements.
When a plane wave arrives at the d-spaced uniform linear array (ULA) with AOA θ, the wave at first
antenna element travels an distance of d sin θ to arrive at the second element. Arrival time delay is given
as η = d/c sinθ. So, the signal arriving at the second antenna can be given in terms of signal at the first
antenna element y2(t) = y1(t) exp(- 2fcη)
= y1(t) exp (- j2 )
For an antenna array with Nr elements all spaced by d, the resulting received signal is given as

Where a(θ) is the array response vector.
Principle of DOA –based beamforming with an example is discussed below:
Consider a three element ULA with d = / 2 spacing between the antenna elements. Assume that user‟s
signal is received with an AOA of θi = 0 and two interfering signals are received with AOAs of θ 2 = /3
and θ3 = - /6. The array response vectors are given by :
 The beam forming weight vector w = [ w1 w2 w3]T should increase the antenna gain in the
direction of desired user, while minimizing the gain in the direction of interferers.
 Thus weight should satisfy w* [ a(θ1) a(θ2) a(θ3)] = [ 1 0 0]T and a unique solution for the
weight vector is readily obtained as w = [0.3034 +j0.1966 0.3932 0.3034 – j0.1966 ]T .
 Below figure shows the beam pattern using weight vector.
 A beam former has unity gain for the desired user and two nulls in the direction of two
interferers.
 As the beam former can place nulls in the directions of interferes, the DOA based beam former
here in this example is called as null-strering beamformer.
 If the number of receive antennas is Nr, then Nr – 1 independent interferers can be cancelled.
 The disadvantage of this approach is that a null is placed in the direction of the interferers, so
that antenna gain is not maximized at the direction of desired user.

Linear Interference Suppression: Complete Knowledge of Interference Channels
Consider a single transmitter with Nt antennas, communicating to a receiver Nr > Nt in presence of one
or more LI, interfering transmitter each with Nt,i antennas Thus L = ∑ .
Let‟s assume L = 1 and Nt = 1 for both transmitter and interferer and Nr = 2. So we have two
transmitted streams, to a two-antenna receiver as shown below.
Fig: Simple two-user interference cancellation
The received signal model is y = Hx + n, where H is a 2 x 2 matrix of both desired and interfering
channels. If we assume that receiver knows its own channel vector and also interfering channel, then the
direction of its desired signal x1 is straightforward.

For example, a zero-forcing receiver G = H-1 would give z = x + H-1 n. This is a example of spatial
multiplexing where transmitters are independent, which is also known as uplink multiuser MIMO or
uplink SDMA.
Linear Interference Suppression: Statistical Knowledge of Interference Channels
 Consider a desired transmitter with Nt antennas for transmission and the desired receiver Nr
antennas for reception in a flat fading channel.
 There are LI distinct cochannel interferers each with Nt,i antennas Thus L = ∑ .
 We allow transmitter to precode its signal with Nt x 1 beamforming vector wt , Nr dimensional
received signal is given as y = Hwtx + HIxI + n where x is desired symbol with energy εx,
 xI =[x1 x2….xL]T is the interference vector and n is the noise vector with covariance matrix 2I,H
is the Nr x Nt channel gain matrix for the desired user.
 HI is the Nr x L channel gain matrix for the interferers. To maximize the output SINR at the
receiver,weighing vectors at both transmitter and receiver can be given as wW t = Eigenvector
corresponding to the largest eigen value max (H*R-1H).
 Wr = R-1Hwt where  is arbitrary constant that does not affect the SNR.
 R = 2I + E [HIXIXI*HI*] is the interference-plus-noise covariance matrix and max(A) is the
largest eigenvalue of A.
-1
 The maximum output SINR: γ = εx max (H*R H). This shows that transmit power is focused
on largest eigenchannel among min (Nt, Nr) eigenchannel to maximize post beamforming SINR.
 This approach is called as optimum eigen-beamformer, interference-aware beam forming,
and/or optimum combiner (OC).
 Interference-aware beamforming is similar to linear diversity precoding, difference is that eigen-
beamformer takes interfering signals into account.
 The below shows a performance comparision between the eigen beamformer and other
transmit/receive diversity schemes.
Here optimal beamformer null a strong interferer by sacrificing a degree of freedom at the receiver.i.e
2 x 2 optimumbeamformer with one strong interferer is equivalent to 2 x 1 MRTwith no interference.
Exploiting channel knowledge at transmitter provides array gain in case of single receive antenna, the
transmit diversity using MRT has same array gain and diversity order of receive diversity MRC.

Spatial Multiplexing
Spatial multiplexing refers to breaking the incoming high data stream into M parallel data streams as
shown in above fig, for M = Nt and Nt < Nr. Here spectral efficiency is increased by a factor of M.
Addition of antenna elements can data rate without any increase in bandwidth.
Introduction: The standard mathematical model for spatial multiplexing is similar to linear precoding
and interference suppression i.e, y = Hx +n. where the size of vector x is Nt x 1, the channel matrix H is
Nr x Nt, the transmit vector x Nt x 1 . So that each symbol in x has average energy εx/Nt. The channel
matrix is
All spatial channels experience uncorellated Rayleigh fading and Gaussian noise, if there is sufficient
scattering, antenna spacing, and no dominant interferers. This models enables a rich framework for
mathematical analysis for MIMO systems based on random matrix theory, information theory and linear
algebra.
The keys points of single-user MIMO system model are:

The maximum data rate grows as minimum (Nt,Nr) log(1 + SNR) when the SNR is large . When
SNR is high spatial multiplexing is optimal.
When the SNR is low, the capacity-maximizing strategy is to send a single stream of data using
diversity precoding.
Both of these cases are superior in terms of capacity to space-time coding, where the data rate
grows logarithmically with Nr.
The average SNR of all Nr streams can be estimated without increasing the total transmit power
relative to a SISO system. As each transmitted stream is received at Nr> Nt antennas and hence
recovers the transmit power penalty of Nt due to array gain.
Open-Loop MIMO: spatial Multiplexing without channel Feedback
Here we assume that channel is known at receiver through pilot symbols or other channel estimation
techniques. The open-loop techniques for spatial multiplexing attempt to suppress the interference that
results from all Nt streams being received by each of the Nr antennas.
Optimum Decoding: Maximum Likelihood Detection
If the channel is unknown at the transmitter the optimum decoder is the maximum likelihood decoder,
input vector x via a minimum distance criterion : x arg min ∣∣y - Hx∣∣2. An search must be done over all
MNt input vectors. Where M is the order of modulation. Lower complexity approximations of the ML
detector can be used to achieve the performance of the ML detector in many cases and these have
potential for high performance, open-loop MIMO systems.

Linear Detectors : Linear detectors are capable of recovering the transmitted vector x.
This detector is zero forcing detector that sets the receiver equal to inverse of the channel Gzf = H-1 when
Nt = Nr: Gzf = (H*H)-1H*. The zero forcing detector completely removes the spatial interference from
the transmitted signal. X = Gzf y = Gzf Hx + Gzf n = x + (H*H)-1 H*n. As Gzf inverts the
eigenvalues of H, subchannels can amplify the noise. This is problematic in MIMO systems and results
in poor performance. The zero-forcing detector is not practical for LTE.
Alternative to zero-forcing receiver is MMSE receiver, which balances between spatial interference
suppression and noise enhancement by minimizing distortion.
That can be obtained using well-known orthogonality principle as
Where pt is the transmitted power. As SNR grows large, the MMSE detector converges to the ZF
detector, at low SNR it prevents eigenvalues being inverted.
BLAST
The spatial multiplexing receiver was invented and prototyped in Bell Labs and is called Bell
labs Layered space-Time(BLAST).
BLAST consists of parallel “layers” supporting multiple simultaneous data streams.
The layers in BLAST are separated by interference cancellation techniques that decouple the
overlapping data streams.

The two most important techniques are the original diagonal BLAST (D-BLAST) and vertical
BLAST(V-BLAST).
D-BLAST groups the transmitted symbols into “layers” which are then coded in time
independent of the other layers.
These layers are then cycled to the different transmit antennas in a cyclical manner. So each
layer being transmitted in a diagonal of space and time.
Each symbol stream achieves diversity in time via coding, and in space since it rotates among all
the different antennas. Therefore, the Nt transmitted streams will equally share the good and bad
spatial channels.
BLAST techniques lies in the detection of the overlapping and manually interfering spatial
streams.
The diagonal layered structure of D-BLAST can be detected shown in the left side of figure
shown below.
Here each layer is detected by nulling the layers that have not yet been detected, and cancelling the
layers that have already been detected. In the figure, the layer to the left of the layer-2 block has already
been detected and hence subtracted from the received signal while those to the right remain as
interference but can be nulled using knowledge of the channel. The time-domain coding helps
compensate for errors or imperfections in the cancellation and nulling process.
Two drawbacks of D-BLAST are that the decoding process is iterative and somewhat complex, and the
diagonal layering structure wastes space-time slots at the beginning and end of a D-BLAST block.
V-BLAST was subsequently addressed in order to reduce the inefficiency and complexity of D-BLAST.
V-BLAST is simpler than D-BLAST. In V-BLAST, each antenna simply transmits an independent
symbol stream. A variety of techniques can be used at the receiver to separate the various symbol
streams from each other.

These include linear receivers such as the ZF and MMSE, which take the form at each receive antenna
of a length Nr vector that can be used to null out the contributions from the Nt – 1 interfering data
streams. Here, the post-detection SNR for the ith stream is
Where wr,i is the ith row of the zero-forcing or MMSE receiver G.

The essence of V-BLAST is to combine a linear receiver with ordered successive interference
cancellation. Instead of detecting all Nt streams in parallel, they are detected iteratively.
First, the strongest symbols stream is detected then they can be subtracted out from the
composite received signal.
Then, the second strongest signal is detected. Which now sees effectively Nt – 2 interfering
streams. In general, the ἰth detected stream experiences interference from only Nt – i of the
transmit antennas.
So that by the time the weakest symbol stream is detected, the vast majority of spatial
interference has been removed. Employing the ordered successive interference cancellation
lowers the block error rate by about a factor of 10 relative to a purely linear receiver. Decreases
the required SNR by about 4dB. V-BLAST prototypes have shown spectral efficiencies above
20bps/Hz.
In both BLAST schemes, these imperfections can quickly lead to catastrophic error propagation
when the layers are detected incorrectly.
Closed-Loop MIMO
The potential gain from transmitter channel knowledge is quite significant in spatial multiplexing
systems. Initially lets consider a simple theoretical example using singular value decomposition (SVD)
that shows the potential gain of closed-loop spatial multiplexing methods.
SVD precoding and postcoding : The gain of transmitter channel knowledge is obtained by
considering the singular value decomposition (SVD) of the channel matrix H. which as noted previously
can be written as: H = U∑V*.Where U and V are unitary and ∑ is a diagonal matrix of singular values.

As shown in above figure, linear operations at the transmitter and receiver. i.e., multiplying by V and
U*. respectively, the channel can be diagonalized. This is confirmed by considering a decision vector d
that should be close to the input symbol vector b. the decision vector can be written systematically as
d = U*y.
= U*(Hx + n).
= U*(U∑V*Vb + n).
= U*U∑V*Vb + U*n.
= ∑b + U*n.
Which has diagonalized the channel and removed all the spatial interference without any matrix
inversions or non-linear processing. Because U is unitary. U*n still has the same variance as n.
The singular value approach does not result in noise enhancement. SVD-MIMO is not particularly
practical since the complexity of finding the SVD of an Nt x Nr matrix is on the order of O( ) if
and requires a substantial amount of feedback.
Linear precoding and postcoding
The SVD was an illustrative example of how linear precoding and postcoding can diagonalize the
MIMO channel matrix to provide up to min ) dimensions to communicate data symbols.
The precoder and postcoder can be jointly designed based on criteria such as the information capacity,
the error probability, the detection MSE, or the received SNR]. The general precoding formulation is

y = G(HFx + n),
where x and y are M x 1, the postcoder matrix G is M x Nr. The channel matrix H is Nr X Nt, the
procedure matrix F is Nt X M, and n is Nr X 1. For the SVD example M = min(Nr,Nt), G = U*, and F =
V.
The linear precoder and postcoder decompose the MIMO channel into a set of parallel subchannels as
shown in below figure. Therefore, the received symbol for the ith subchannel can be expressed as:
Where xi and yi are the transmitted and received symbols, respectively, with E| x i|2 = ϵx as usual, ζi are
the singular values of H.
αi and βi are the precoder and the postcoder weights. And ni is the noise per subchannel. Which in
addition to the usual noise also includes any residual spatial interference from the imperfect separation
of the streams. The number of subchannels is bounded by: 1 ≤ M ≤ min(Nr,Nt).
Where M = 1corresponds to the maximum diversity order and M = min(Nt,Nr) achieves the maximum
number of parallel spatial streams.
How to Choose Between Diversity, Interference Suppression, and Spatial Multiplexing
Diversity provides robustness to fades and interference suppression provides robustness to interference.
They do increase the possible throughput on the stream that is sent by increasing the SINR = S/(I+N). In
particular, diversity increases and steadies S. While interference suppression reduces I. Spatial
multiplexing creates more parallel streams but does not necessarily increase the per-stream SINR.

Interference suppression and nulling is often considered impractical in a cellular system. The main
reasons for this are
(1) The interfering transmitters are numerous and fairly far away from the receiver, so the gain from
cancelling just the few strongest ones is not always large.
(2) Acquiring the needed channel state information from the interferers can be quite difficult. So
accurate suppression is not usually possible. Therefore, most research has focused on diversity and
multiplexing.
 When the transmitter has full CSI. The optimal precoder with link adaptation will select the
optimum number of streams and the power sent on each. So with accurate CSIT.
 There is an inherently optimal tradeoff point between diversity and multiplexing. When the
transmitter does not have exact CSIT, but instead partial CSIT, for example, knowledge of the
average link SINR.
 The notion of switching or balancing between diversity and multiplexing based on a proxy for
SINR was to our knowledge first introduced by heath.
 It was then developed into an elegant theory known as the diversity-multiplexing tradeoff
(DMT), which has gained many adherents and spawned significant follow-up research.
 The DMT stipulates that both diversity gain and multiplexing gain can be achieved in a multiple
antenna channel but that there is a fundamental tradeoff between how much of each gain can be
achieved. Hence, diversity gain the reliability exponent Nd and multiplexing gain in essence, the
number of streams sent are obtained at the expense of each other.
 In the context of LTE, a more relevant perspective on DMT was recently developed by Lozano
and jindal. This results in several conclusions at odds with those of the traditional model.
 Lozano and jindal use a model that includes frequency-selective fading, ARQ and coding and
focuses on achieved rate at fixed error probability all of which are appropriate adjustments for
LTE.
 The principle is that all the spatial degrees of freedom should be used for multiplexing and none
for spatial diversity.
 In short, there is no tradeoff observed in below figure. on the left, with setting corresponding to
the simplistic DMT model (and Nt = Nr = 4), we see that for all but the highest SNR values,
transmit diversity indeed outperforms spatial multiplexing.

 In fact, spatial multiplexing even does worse than no transmit diversity, because so many errors
are made on the weakest streams. But with the wideband channel model on the right of figure,
specifically, the UMTS Typical Urban (TU) channel with Tmax = 1µsec, and hybrid ARQ, the
relation between them flips.
 First, modern wireless systems have many forms of diversity, mostly time and frequency
diversity. Which are exploited using coding, interleaving, retransmission (ARQ), OFDMA, and
adaptive modulation. There is very little diversity left in the channel to exploit with spatial
diversity when these are considered.
 Second, link adaptation is used to maintain a target block error probability, and there is very little
benefit to beat this target.

WC& Lte (15ec81) Module - 2 Multicarrier Modulation

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

WC& Lte (15ec81) Module - 2 Multicarrier Modulation

Încărcat de

Drepturi de autor:

Formate disponibile

WC& LTE (15EC81) Module -2 MULTICARRIER MODULATION

Module -2 MULTICARRIER MODULATION

Multicarrier Modulation: OFDM basics

Block Transmission with Guard Intervals

Circular Convolution and the DFT

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 1

where x[n] * h[n] = h[n] * x[n] ∑ [ ] [ ]

In frequency domain: Y[m] = H[m] X[m]

The L point DFT is defined as DFT {x[n]} = X[m] 1/√ ∑ [ ]

While its inverse IDFT is defined as IDFT {X[m]} = x[n] 1/√ ∑ [ ]

The Cyclic Prefix

After applying a cyclic prefix of length υ the actual transmitted signal is

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 2

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 3

Rate loss = power Loss = .

An OFDM Block Diagram

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 4

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 5

The key OFDM parameters are summarized in the table below.

Timing and Frequency Synchronization

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 6

f{cos(2Πfct).rec(t/T)} = F{cos(2Πfct)} * f{rec(2t/T)}

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 7

For both of these scenarios, the SNR loss can be given as

 SNR decreases quadratically with the timing offset.

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 8

Let us analyze ICI in order to understand its effects on OFDM performance.

An interfering subcarrier m can be given as

If the signal is demodulated with a fractional frequency offset of δ , δ ≤ ½

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 9

 SNR decreases quadratically with the frequency offset

The Peak to Average Ratio [PAR]

The PAR problem

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 10

Quantifying the PAR

The amplitude of the output signal is

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 11

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 12

Simple and accurate approximation of the CCDF for large L ( ≥ 64)

G(L, εmax) = P(max ‖ X(t) ‖ εmax ) = F (L, εmax)L

Clipping and other PAR techniques

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 13

The clipping noise in frequency domain can be given as

The following effects observed :

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 14

The attenuation factor α is plotted in below fig, as a function of clipping ratio α.

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 15

Single-Carrier Frequency Domain Equalization (SC-FDE)

1. Low complexity even for severe multipath channels.

System description of SC-FDE

Y[n] = x[n] *h[n] + ω[n] , where ω[n] is noise , therefore

FFT{y[n]} Y[m] = H[m] X[m] + W[m].

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 16

SC-FDE Performance Vs. OFDM

Design Considerations for SC-FDE and OFDM

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 17

The computational Complexity Advantage of OFDM and SC-FDE

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 18

OFDMA and SC-FDMA: OFDM with FDMA

Multiple Access Overview

 Multiple access strategies provide non interfering, orthogonal communication channels

Random Access Vs. Multiple Access

Prof. Ravi Kumar.M, Dept of ECE CiTech, Bengaluru Page 20

y1 = h1r(f1) + h2r (f2)

y2 = - h2r(f1) – h1r (f2)

y1 = h1(h1s1 + h2s2 +n(f1)) + h2 (- h1s2 +h2s1 + n *(f2))

= (∣h1∣2 + ∣h2∣2)s1 + h1n(f1) + h2n(f2)

Similarly y2 = (∣h1∣2 + ∣h2∣2)s2 + h2n(f1) – h1n(f2)