LTE-ADVANCED: NEXT-

GENERATION
WIRELESS BROADBAND
TECHNOLOGY











SUBMITTED BY: NITHU V.C.
S8 ECE
ROLL NUM:58
ABSTRACT




LTE Release 8 is one of the primary broadband technologies based on OFDM,
which is currently being commercialized. LTE Release 8, which is mainly
deployed in a macro/microcell layout, provides improved system capacity and
coverage, high peak data rates, low latency, reduced operating costs, multi-antenna
support, flexible bandwidth operation and seamless integration with existing
systems. LTE-Advanced (also known as LTE Release 10) significantly enhances
the existing LTE Release 8 and supports much higher peak rates, higher throughput
and coverage, and lower latencies, resulting in a better user experience.
Additionally, LTE Release 10 will support heterogeneous deployments where low-
power nodes comprising picocells, femtocells, relays, remote radio heads, and so
on are placed in a macrocell layout. The LTE-Advanced features enable one to
meet or exceed IMT-Advanced requirements. It may also be noted that LTE
Release 9 provides some minor enhancement to LTE Release 8 with respect to the
air interface, and includes features like dual-layer beamforming and time -
difference-of-arrival-based location techniques. In this article an overview of the
techniques being considered for LTE Release 10 (aka LTEAdvanced) is discussed.
This includes bandwidth extension via carrier aggregation to support deployment
bandwidths up to 100 MHz, downlink spatial multiplexing including single-cell
multi-user multiple-input multiple-output transmission and coordinated multi point
transmission, uplink spatial multiplexing including extension to four-layer MIMO,
and heterogeneous networks with emphasis on Type 1 and Type 2 relays. Finally,
the performance of LTEAdvanced using IMT-A scenarios is presented and
compared against IMT-A targets for full buffer and bursty traffic model.








INTRODUCTION



Universal Mobile Telecommunications System (UMTS) Long Term Evolution
(LTE) Release 8 provides high peak data rates of 300 Mb/s on the downlink and 75
Mb/s on the uplink for a 20 MHz bandwidth, and allows flexible bandwidth
operation of up to 20 MHz. Currently, enhancements are being studied to provide
substantial improvements to LTE Release 8, allowing it to meet or exceed
International Mobile Telecommunications- Advanced (IMT-A) requirements [1].
These enhancements are being considered as part of LTE-Advanced (LTE-A, also
known as LTE Release 10), which includes carrier aggregation, advanced uplink
(UL) and downlink (DL) spatial multiplexing, DL coordinated multipoint (CoMP)
transmission, and heterogeneous networks with special emphasis on Type 1 and
Type 2 relays.
This article provides an overview of the technologies being considered for LTE-
A. This article is organized as follows. In the next section an overview of the LTE
Release 8 physical layer (PHY) is provided. This is followed by an overview of
evolved UMTS terrestrial radio access (E-UTRA) LTE-A requirements. In the
following section a discussion on carrier aggregation is provided. We then provide
an overview of DL and UL spatial multiplexing and fundamentals of DL CoMP
design. We introduce the concept of heterogeneous networks, withy an emphasis
on LTE relays. We compare the performance of LTE Release 8 and LTE-A in the
context of IMT-A requirements. Finally, conclusions are drawn in the last section.









New features of LTE-Advanced: from Release 8 to Release 10

LTE is the result of the standardization work done by the 3GPP to achieve a new
high speed radio access in the mobile communications frame. It was introduced in
the Release 8 in 2008. In 2010 the Release 9 has come to provide some
enhancements to LTE and in 2011 Release 10 will bring LTE-Advanced, which
will expand the limits and features of Release 8 to meet the requirements of the
IMT-Advanced of ITU-R for the fourth generation of mobile technologies (4G),
and the future operator and end users requirements. LTE-Advanced terminals
have to be compatible with LTE-Release8 networks and vice versa, LTE-Release 8
terminals have to be compatible with LTE-Advanced networks. The key measure
of LTE is the ability to provide very high bit rates. In addition, it provides high
spectral efficiency, very low latency and support of variable bandwidth. Some of
the LTE new features are:
OFDMA, a multi-user version of the modulation scheme called (Orthogonal
Frequency-Division Multiplexing) in the downlink. This gives robustness
against multipath interference and connects with some advanced techniques
also used like MIMO and frequency domain channeldependent scheduling.

Single-Carrier FDMA with Dynamic Bandwidth in the uplink. SC-FDMA
has lower peak-to-average power ratio (PAPR) which is a major
improvement for the user equipment (UE), as it improves the transmission
power efficiency.

Multiple antenna solutions. MIMO (Multiple Input Multiple Output) is
probably the most important feature of LTE for improving the data bit rates
and the spectral efficiency. It consists on the use of multiple antennas in both
the receiver and the transmitter in order to use the multipath effects, which
reduces the interference and leads to high transmission rates. MIMO works
by dividing the data flow into multiple unique flows, and transmits them in
the same radio channel at the same time. They will be merged using an
algorithm or special signal processing.

Figure:1 MIMO block

Very low latency due to a short setup time and small transfer delays. This is
a basic feature as many applications, especially the ones related to voice and
video transfer, rely on low latency times.

LTE can support variable bandwidths, in the range between 1.4 and 20MHz.

LTE-Advanced extends the features of LTE in order to exceed or at least meet the
IMT-Advanced requirements. It should be a real broadband wireless network that
behaves as an advanced fixed network like FTTH (Fiber To The Home) but with
better quality of service. It also must fulfil operators demands like a reduced cost
(per Mbit transmitted), compatibility with all 3GPP previous systems and a better
service providing in terms of homogeneity, constant quality of the connection and
smaller latency. Here we list some of the LTE-Advanced proposals to achieve the
goals of this standard:

Support of asymmetrical bandwidths and larger bandwidth (maximum of
100MHz). In LTE (release 8), the bandwidth could have different sizes but
had to be the same in the downlink and in the uplink. In LTE-Advanced
(Release 10) bandwidths can be different because due to actual demand in
mobile networks, the traffic from the station to the user is bigger than the
one from the user to the station. And they can be as asymmetric as they want
within the limit of the 100 MHz LTE-Advanced provides. The sum of both
bandwidths (downlink + uplink) cannot exceed 100 MHz. Carrier
aggregation to achieve wider bandwidth is a key factor as well as the support
of spectrum aggregation, to achieve higher bandwidth transmissions.

Enhanced multi-antenna transmission techniques. LTE introduced MIMO in
the data transmission and LTE-Advanced the MIMO scheme has to be
extended to gain spectrum efficiency (which is proportional to the number of
antennas used), cell edge performance and average data rates. LTE-
Advanced considers a configuration 8x8 in the downlink and 4x4 in the
uplink.

Figure 2: MIMO scheme(8x4)


Coordinated multipoint transmission and reception (CoMP) which improves
the received signal of the user terminal. Both the serving and the neighbour
cells are used in a way that the co-channel interference from neighbouring
cells is reduced. It implies dynamic coordination between geographically
separated transmission points in the downlink and reception at separated
points in the uplink. This mechanism will improve the coverage of high data
rates and will increase the system bit rate.

Relaying. Relaying increases the area covered and the capacity of the
network. Users mobile devices communicate with the relay node, which
communicates with a donor eNB (enhanced Node B, thats 3GPPs term for
base stations). Relay nodes can also support higher layer functionality like
decoding user data from the donor eNB and re-encoding the data before
transmitting it to the user terminal.


Type 1 relay nodes control their cells with their own cell identity, and among their
tasks we can find the transmission of synchronization channels and reference
symbols. This type of relays guarantee compatibility with previous releases as it
appeared in Release 8 to provide service to Release 8 mobile terminals.


Figure3: Type1 relay

Type 2 relay nodes dont own an identity, so the mobile user wont be able to
distinguish if a transmission comes from the donor eNB or from the relay. eNBs
transmit control data and relays transmit user data.


Figure 4: Type 2 rela

LTE-A REQUIREMENTS


LTE-A is the evolutionary path from LTE Release 8. The LTE-A technology
should be able to satisfy the IMT-A system requirements specified in : IMT-A
systems are mobile systems that include the new capabilities of IMT that go
beyond those of IMT-2000. Such systems provide access to a wide range of
telecommunication services including advanced mobile services, supported by
mobile and fixed networks, which are increasingly packet-based. All of the IMT-
A requirements cannot be fulfilled by LTE Release 8 and require technology
beyond Release 8. As such, LTE-A is
designed to meet and exceed these IMT-A requirements as summarized in Table 1.
These requirements are met using a variety of techniques,
including:
Carrier aggregation

DL spatial multiplexing using up to eight-layer multiple-input multiple-output
(MIMO)

DL intracell CoMP transmission and reception

UL Spatial Multiplexing using four-layer MIMO

In addition to the improvement in spectral efficiency, substantial reduction in
latency is also targeted. The goals are to reduce the transition time from idle to
connected mode from 100 ms in LTE to less than 50 ms in LTE-A. Similarly, the
transition from dormant to active should be reduced from 50 ms in LTE to less
than 10 ms in LTE-A. In LTE-A capacity and coverage enhancement can also be
achieved using a heterogeneous network, which is a collection of low-power nodes
distributed across a macrocell (homogeneous) network. There are various types of
low-power nodes including microcells, picocells, femtocells, and relays. These
low-power nodes are deployed in various environments including hot spots,
homes, enterprise environments, and low geometry locations to improve the
overall capacity and coverage of the system.


Table 1. LTE-A targets for full buffer traffic.


CARRIER AGGREGATION


Bandwidth extension in LTE-A is supported via carrier aggregation. Carrier
aggregation allows deployment bandwidths of up to 100 MHz, enabling peak
target data rates in excess of 1 Gb/s in the DL and 500 Mb/s in the UL to be
achieved. In addition, it can be used to effectively support different component
carrier types that may be deployed in heterogeneous networks. Carrier aggregation
is attractive because it allows operators to deploy a system with extended
bandwidth by aggregating several smaller component carriers while providing
backward compatibility to legacy users. Three different types of component
carriers are envisioned:
Backward-compatible carrier: All LTE UE can access this type of carrier
regardless of the supported release. In this case all of the current LTE features must
be supported.
Non-backward-compatible carrier: Only LTEA UEs can access this type of
carrier. This carrier may support advanced features such as control-less operations
or the anchor-carrier concept not available to LTE UE.
Extension carrier: This type of carrier operates as an extension of another
carrier. It may be used, for example, to provide services to home eNBs in the
presence of high interference from the macrocell. UE can only access this type of
carrier as part of a carrier aggregation set. The extension carrier is addressed
using a separate PDCCH and uses a separate HARQ process. Three possible
aggregation scenarios are possible: contiguous aggregation of component carriers
in a single band, non-contiguous aggregation of component carriers in a single
band, and non-contiguous aggregation of component carriers over multiple bands.
Up to five component carriers may be aggregated together, providing a maximum
bandwidth of 100 MHz. Under the current PHY specifications, the peak DL data
rate for a 20 MHz system bandwidth is 150.752 Mb/s with two-layer spatial
multiplexing and 299.552 Mb/s with four-layer spatial multiplexing. With carrier
aggregation, these peak rates will scale linearly with the number of carriers. For
instance, with carrier aggregation of 5 20 MHz carriers in the DL, a peak
theoretical DL data rate of 1.5 Gb/s can be achieved without any additional
features. In the UL a peak data rate of 75.376 Mb/s is currently possible. With
carrier aggregation, a peak UL rate of 376.88 Mb/s is theoretically possible. A
peak UL data rate in excess of 500 Mb/s can then be achieved with additional
enhancements such as UL spatial multiplexing. Prioritized deployment scenarios
have been defined in [4] based on preferences provided by leading network
operators. Illustrative examples include:

UL: 40 MHz (20 + 20 MHz, 3.5 GHz), DL: 80 MHz (2 20 + 2 20 MHz, 3.5
GHz), FDD
UL: 100 MHz, DL: 100 MHz (5 20 MHz, 3.5 GHz), TDD

Note that since non-contiguous multiband aggregation is possible, coverage and
performance can be different across the aggregated component carriers. In
addition, UE may be assigned different UL and DL component carrier sets that are
subsets of the system configuration. For example, the cell may be configured to
support UL: 40 MHz, DL: 80 MHz, while UE is configured to support on UL: 20
MHz, DL: 40 MHz. Different users within the cell may be configured differently,
providing great implementation flexibility. This flexibility allows for efficient
support of techniques such as load balancing, interference coordination, quality of
service (QoS) management, and power control. In addition, different UE
bandwidth capability classes and heterogeneous networks can also be supported.
This flexibility, however, requires robust control channel design to support all
possible configurations. For instance, with asymmetric carrier aggregation, some
carriers are missing their counterparts which presents a challenge because pair-
wise operational support (e.g., ACKs) are necessary for FDD. As a result, some
carriers must be configured to support more than one counterpart.

UE IMPLEMENTATION

SC-FDMA using DFT-S-OFDM is the PHY access scheme in the UL. With carrier
aggregation, N SC-FDMA, which is conceptually analogous to parallel SC -
FDMA transmitters is used. In this case N DFT-inverse fast Fourier transform
(IFFT) pairs are required to carrier aggregation in the UL [5]. A block diagram
of the UL demonstrating N SC-FDMA for N = 2 is shown in Fig. 4. One potential
issue with carrier aggregation is the limitation on the UEs transmission power. As
a result, for FDD it is expected that two component carriers supporting aggregated
bandwidth of 40 MHz may be the most practical scenario. For TDD, however, up
to 100 MHz must be supported based on the DL bandwidth. With carrier
aggregation, the single carrier property in the UL is no longer preserved when
transmitting on multiple carriers. As a result, the cubic metric (defined in [6] as the
cubic power of the signal of interest compared to a reference signal) increases,
which requires a larger backoff in the power amplifier, thereby reducing the
maximum transmit power at the UE. An illustrative comparison is shown in Table
2 for different numbers of component carriers. For example, the cubic metric
increases from 2.2 to 3.0 dB when the UE is transmitting using 16-quadrature
amplitude modulation (QAM) on two simultaneous component carriers. This
substantial increase in cubic metric has the effect of reducing the coverage when
carrier aggregation is used, although it can be seen that the cubic metric is
generally still less than that for orthogonal frequency- division multiple access
(OFDMA, note that results for OFDMA are independent of modulation). In
addition, the increase is most significant in going from one to two simultaneous
component carriers, and only increases gradually as more carriers are used. It
should be noted that since multicarrier transmission will usually be used for UE in
good channel conditions, there should be no loss of coverage for those users. On
the other hand, users at the cell edge will most likely be scheduled only on a single
carrier. There is no difference in cubic metric in this case, and coverage may
actually be improved since the eNB has the ability to dynamically assign users to
the best UL carrier.


Figure5 : Transmitter block diagram for uplink carrier aggregation.


Table 2. Cubic metric comparison for N SC-FDMA.






DATA TRANSMISSION AND CONTROL SIGNALING
At the data layer, each component carrier operates independently with a separate
data stream that is aggregated or segmented at the medium access control (MAC)
layer. This MAC-PHY interface provides support for carrier aggregation with
minimal changes to the PHY specifications. In this case separate HARQ
processing and associated control signaling is required for each of the component
carriers. This allows separate link adaptation and MIMO support for each carrier,
which should improve throughput since each data transmission can be
independently matched to channel conditions on each carrier. As a result, data
processing at the PHY can be thought of as independent per carrier.

Control signaling serves several main purposes in LTE: to provide signaling
related to scheduling assignments, to provide acknowledgment in response to data
transmission, to provide channel state feedback information, and to provide power
control information. To support carrier aggregation, these functions must be
extended to support multiple component carriers. It should be noted that, in
general, the principle is to extend the LTE design to multiple carriers when
possible.

In the DL three control channels are presented: the PDCCH, used for scheduling
assignments and power control; the PCFICH, used to indicate the size of the DL
control region; and the PHICH, used to provide acknowledgment in response to
UL data transmission.

For the PDCCH, a separate scheduling grant has been adopted where each grant
addresses one carrier. To allow for scheduling flexibility, each grant will contain a
carrier indication field (CIF) to indicate to which carrier the grant applies, which
thus allows the reuse of existing LTE scheduling grant formats with only the
addition of the CIF. By using a separate grant structure, the eNB has the ability to
schedule different grant formats to the same UE in different component carriers,
and perform dynamic grant load balancing and interference coordination among
the component carriers on a subframe basis.

For the PCFICH, the channel is expected to be configured independently for each
component carrier. A UE will therefore independently decode the PCFICH and
determine the data boundary as in LTE. In the case of a PDCCHless carrier (e.g.,
extension carrier in a heterogeneous network), this would be known to the user and
no special handling is necessary.

Since separate HARQ processing is used, associated control signaling will be
required for each of the component carriers. In addition, LTE-A UE can be
configured with UE-specific UL/DL carrier aggregation configurations that are a
subset of the system configuration shown in Fig. 5. The ACK is transmitted on the
same DL carrier as the UL scheduling assignment, which allows many
simultaneous UE-specific configurations to be supported. This feature solves the
issue with UL-heavy aggregation and also does not require an explicit DL-UL
pairing relationship to be defined. Although this will require the eNB to manage
the PHICH, it will be similar to the PDCCH management that will also be
required.

UL control signaling carries ACKs for theDL data transmission and channel state
feedback (e.g., CQI, precoding matrix indication [PMI], or rank indication [RI]). In
the UL, the simplest possible approach is to support an individual UL control
channel per carrier based on the Release 8 structure. Thus, each UL control
channel may be configured independently to support one DL-UL carrier pair. In
the case of asymmetric carrier aggregation, additional UL control signaling
(ACK/NACK and CQI) may be configured to support multiple DL carriers on a
single UL carrier. One concern for the UL is the limited transmission power
available to the UE. As a result, it may not be possible to transmit feedback or
acknowledgments from many component carriers simultaneously. In this case it is
possible that some form of control information multiplexing such as ACK/NACK
bundling may be used. In addition to carrier aggregation, within the same carrier,
UE has the ability to simultaneously transmit both the physical uplink shared
channel (PUSCH) and physical uplink control channel (PUCCH). With carrier
aggregation in addition to simultaneous PUSCH and PUCCH transmission, the
total cumulative power the UE is allowed to transmit across all component carriers
should not exceed the maximum UE power. Under a transmit power limitation,
power reduction based on per-channel scaling is used where scaling factors are
used such that power reduction is first performed on the PUSCH.





DL/UL SPATIAL MULTIPLEXING AND DL COMP

The DL MIMO transmission schemes already supported in LTE Release 8 include
transmitdiversity and open-loop and closed-loop spatial multiplexing with up to
four layers. These MIMO schemes are supported by common reference signals
(common pilots) broadcast from an eNB. Codebook-based precoding is used to
support closed-loop spatial multiplexing. In addition, a single layer of a UE-
specific reference signal (also known as dedicated pilots) is available that may be
used for beamforming. The UE-specific reference signal enables single-layer
beamforming in TDD scenarios where the DL spatial channel information may be
obtained at the base station using channel reciprocity. A multi-user MIMO (MU-
MIMO) transmission scheme may also be applied based on codebook feedback.
However, the performance of the MUMIMO scheme is limited by coarse
quantization (using codebook) and the lack of support for cross-talk suppression at
the UE. As a result, MU-MIMO in Release 8 does not provide any performance
gains compared to SU-MIMO. Fortunately, LTE-A will fix the MU-MIMO
performance, as we will see.

The performance of DL and UL transmission in LTE Releases 9 and 10 will
be enhanced by the following features :

Enhanced single-cell DL MU-MIMO support

Extension to eight-layer DL spatial multiplexing

DL CoMP support

Extension to four-layer UL spatial multiplexing

ENHANCED SINGLE CELL DL MU-MIMO SUPPORT

In LTE Release 9 two layers of orthogonal UEspecific reference signals have been
introduced, as shown in Fig. 6. This enables an eNB to transmit two layers of data
to a UE set using spatial multiplexing in a closed-loop mode by constructing
antenna weights using channel reciprocity. Additionally, this enables an eNB to
transmit two layers of data to two UE sets (one layer each) using the same time-
frequency resource in MU-MIMO fashion. Switching between singleand dual-layer
transmission to a single UE set, as well as between SU-MIMO and MU-MIMO
is supported in a dynamic fashion. The control signaling overhead for supporting
dynamic and transparent MU-MIMO transmission is small because UE is not
explicitly informed of the presence of co-scheduled UE, for the purposes of
feedback or demodulation. The two layers of UE-specific reference signals are
overlaid on top of each other (separated by a length 2 orthogonal cover code) and a
UE, after subtracting out its channel estimate, may estimate a covariance matrix
representing the combined interference from a co-scheduled UE and outer cell
transmissions. This feature can be used by a receiver to significantly suppress the
interference due to MU-MIMO. The MU-MIMO enhancements in Release 9
provide substantial gains in sector throughput when the eNB is able to form
transmit beams using reciprocity-based techniques. Note that with channel
reciprocity based MUMIMO, there is no restriction on the number of transmit
antennas that can be employed at the eNB. In addition, up to four layers of
quasiorthogonal UE-specific reference signals is available for MU-MIMO enabling
co-scheduling of up to four UE sets in the same time-frequency resource. In
Release 10 an enhancement to provide high-resolution spatial channel feedback is
being considered to support MU-MIMO in cases where channel reciprocity is not
reliable.



Figure 6: Two layers of UE-specific reference signals (dark shade) separated by
code in a resource block. The lightlyshaded resources indicate common reference
signals for four antenna ports present in Release 8.


EXTENSION TO
8-LAYER DL SPATIAL MULTIPLEXING

The extension to eight-layer spatial multiplexing is aimed at improving the DL
peak throughput of a Release 10 system employing eight or more transmit
antennas. Accordingly, eight layers of UE-specific reference signals will be
introduced for demodulation. Since the density of UE-specific reference signals
can be adapted based on the number of layers of data transmission and the size of
the resource allocation, it is more economical to use UE-specific reference signals
compared to common reference signals for supporting higher-order spatial
multiplexing. The maximum number of transport blocks transmitted over eight
layers will remain two with support for separate modulation and coding schemes
(MCSs) and separate HARQ ACKs. In order to support feedback of channel
quality, choice of frequency bands, or channel spatial information (e.g., codebook)
for up to eight layers, additional reference signals will be broadcast from an eNB.
These channel state information reference signals (CSI-RSs), however, will not be
used for demodulation, and will be designed to be sparse in time and frequency
with an overhead of around 1 percent or less. The CSI-RS design will support up to
eight transmit antennas and will potentially enable UE at the cell edge to measure
CSI-RSs transmitted from adjacent cells (for CoMP support). The power, density,
and placement of CSI-RSs will have to be designed carefully as they may interfere
with the data transmissions of a Release 8 or 9 user in the same subframe.

DL COMP SUPPORT

The motivation of the CoMP feature is to provide air interface support to enable
cooperation among eNBs that may or may not be collocated. Multiple eNBs may
cooperate to determine the scheduling, transmission parameters, and transmit
antenna weights for a particular UE. This cooperation will depend on a high-
capacity backhaul link being available between eNBs. Closed loop beamforming or
precoding-based transmissions will be supported in CoMP. This framework is
depicted in Fig. 7, where three eNBs may coordinate to create a multipoint
transmission to UE1 (served by eNB1) and UE2 (served by eNB2). The objective
of CoMP is to reduce interference for a UE set in the network that is close to
multiple eNBs and therefore experiences an interference-limited environment. The
interference to these UE sets may be reduced and can be predicted if there is some
coordination between the interfering eNBs and the serving eNB. The control
overhead, feedback methods, and scheduling restrictions necessary to enable
CoMP are under investigation. Within the timeframe of Release 10 a standardized
interface of direct communication between two eNBs (i.e., the X2-interface) will
not be redefined. Therefore, the coordination needed for CoMP will depend on
proprietary interfaces that may easily be possible for intersector coordination. A
similar framework for eNB cooperation for UL transmissionis also under study.



Figure 7: A CoMP framework for DL transmission where eNB1, eNB2, and
eNB3 can coordinate and create a multipoint transmission to UE1 and UE2.

EXTENSION TO FOUR-LAYER
UL SPATIAL MULTIPLEXING
It is envisioned that Release 10 compatible UE may support up to four transmit
antennas, and the extension to four-layer spatial multiplexing is targeted to
improve the UL peak rate in such cases. As in Release 8, the UL transmission
scheme for Release 10 will be DFT-precoded OFDM that will retain the single
carrier property and restrict the peak-to-average power ratio (PAPR) or the cubic
metric [t] of the UL waveform to a low value. Open-loop and closed-loop spatial
multiplexing will be supported, and in the case of closed-loop operation,
codebook-based precoding will be used. In contrast to the DL, the transmission
schemes in the UL are optimized to preserve the single-carrier property or reduce
PAPR/cubic metric. As an example, layer shifting is a mechanism introduced to
transmit data from multiple layers through a given antenna while maintaining
single-carrier properties. Layer shifting has been proposed to provide robustness to
spatial multiplexing against antenna gain imbalances and CQI mismatches due to
velocity. Codebook designs are also being studied to enable precoded transmission
with low PAPR/cubic metric. In addition, the codebook will also provide the
capability to turn off certain antennas (for power saving) at the instruction of an
eNB.

HETEROGENEOS NETWORKS

In heterogeneous networks low-power nodes are distributed throughout macrocell
networks. Lowpower nodes can be micro eNBs, pico eNBs, home eNBs (for
femtocells), relays, and distributed antenna systems (DASs), the latter of which
may employ remote radio head (RRH) cells. These types of cells operate in low-
geometry environments and produce high interference conditions. Such
deployments enable optimization of network performance at relatively low cost.
The various types of heterogeneous deployments are summarized in Fig. 8, and the
definitions of the various low power nodes are summarized in Table 4 [7, 8]. In
addition to CoMP, intercell interference coordination (ICIC) techniques can play a
critical role in obtaining good performance within heterogeneous deployments.
In this article one of the components of heterogeneous networks, relay nodes
(RNs), is discussed in detail. Repeaters can also offer coverage extension for eNBs
by amplifying and forwarding received waveforms, but suffer from the following
disadvantages: they cannot distinguish between signals and interference/noise, and
RF isolation is a big issue. Relays typically yield performance improvement by
providing higher throughputs to UE that would otherwise be located in poor
geometry locations with respect to the macrocell sites. There are two types of
relays being discussed in the context of 3GPP standards, Type 1 and Type 2 relays.
Type 1 and Type 2 relays have the following characteristics:




Type 1:
A relay cell should have its own physical cell ID and transmit its own
synchronization channels, reference symbols, and so on, and will be distinct from
the donor cell.
The UE should receive scheduling information and HARQ feedback directly from
the RN and send its control channels (SR/CQI/ACK) to the RN.
The RN appears as a Release 8 eNB to Release 8 UE.

Type 2:
The RN does not have a separate Physical Cell ID and thus would not create any
new cells.
It is transparent to Release 8 UE; Release 8 UE is not aware of its presence.
It can transmit physical downlink shared channel (PDSCH) but does not transmit
CRS and PDCCH.



Figure 8. Classification of heterogeneous deployments.


Table 4. Characteristics of low power nodes.

The RNs are connected to the radio access network (RAN) via a donor macrocell.
Two types of backhaul connections can be supported: in-band (IB), where the
eNB-to-relay link shares the same band with direct eNB-to-UE links within a cell,
and out-of-band (OOB), where the eNB-to-relay link does not share the same band
with direct eNB-to-UE links . For the IB scenario, backhaul traffic is facilitated by
configuring certain subframes as multicast broadcast single-frequency network
(MBSFN) subframes in the relay cell. These subframes are then used by a relay
node to receive DL backhaul traffic from its donor eNB. This configuration is
illustrated in Fig. 9, where the utilization of subframes at eNBs, RNs, macrocell
UE (UE1), and relay cell UE (UE2) is shown. The arrows show the direction of
transmission for radio links in each subframe. Thus, subframes are normally used
for access links; that is, DL transmission from an eNB or RN to its UE except
during the MBSFN subframes, when UE in the relay cell does not receive data,
whereas eNBs may transmit DL traffic to both RNs (i.e., backhaul traffic) as well
as macrocell UE (access traffic). The set of such DL backhaul subframes is semi-
statically assigned, whereas the set of UL backhaul subframes is either semi-
statically assigned or implicitly derived from the DL backhaul subframes using the
HARQ timing relationship. For simplicity, our simulation study assumes that the
same DL backhaul subframe configuration is maintained throughout the network.
The DL subframe boundaries for the access and backhaul links are aligned at the
RN while allowing an adjustment for RN transmit/receive switching. A new
physical control channel, RPDCCH, is defined for the eNB to assign backhaul
resources to the RN. A set of physical RBs (PRBs) is semi-statically assigned for
the transmission of the R-PDCCH. The actual resources used within these PRBs
may vary dynamically between subframes. Furthermore, these resources may
correspond to either the full set of OFDM symbols available for the backhaul
link or a subset of them. The R-PDCCH is transmitted starting from an OFDM
symbol within the subframe that is late enough so that the RN can receive it (i.e.,
such that the RN can complete transmission of its own PDCCH and switch to
receive). The assignment of DL and UL backhaul resources on the R-PDCCH may
be done either dynamically or semi-persistently. DL backhaul resources may be
assigned for either the same subframe and/or a later subframe. UL resources may
be assigned for one or more later subframes.



Figure 9. Radio links and subframe utilization at different nodes in the network.

Four models are considered for the IB backhaul link. The models to be used for
path loss, antennas, and lognormal shadowing on the access and backhaul links are
described in [7]. The first path loss model is that of non-optimized relay site
planning with a single omnidirectional antenna set at the RN. We refer to this
model as backhaul A. The backhaul model for optimized relay site planning with a
single omnidirectional antenna set is referred to herein as backhaul B. The next
two models assume the presence of two antenna sets at the RN, an omnidirectional
antenna set for the relay access links, and a directional antenna set for the backhaul
link (i.e., for receiving DL data from the eNB). The backhaul model with a
directional antenna and non-optimized relay site planning is called backhaul C.
Finally, the model that combines a directional antenna with optimized relay site
planning is referred to as backhaul D.

A two-ring, 19-macrocell, three-sectored site hexagonal grid system layout is
simulated with dual-port UE receiver operation and assuming TU channels using
cell wrap-around for two systems, each operating in a 10 MHz bandwidth,
corresponding to deployment scenario (DS) cases 1 and 3. One thousand four
hundred twenty- five UE sets are randomly dropped with uniform spatial
probability density over the entire 57-cell network. For the results presented here,
a deployment with 228 relays is considered. The relays are dropped randomly over
the entire network with a uniform spatial distribution.

The number of in-band backhaul subframes per frame (SFpF) shared among
all RNs in a sector is a parameter. By controlling the number of backhaul SFpF, the
size of the backhaul pipe can be controlled (at the expense of the resources
available for the macrocell access links). Note that this is the total number of
subframes shared by all backhaul links in the sector. For IB backhaul, the
scheduler of each RN is constrained to allocate resources to its UE only when the
amount of data it has transferred to the UE does not exceed the amount of data the
RN has received from the donor eNB. This constraint ensures that the relay-cell
throughput does not exceed the corresponding backhaul throughput. For OOB
backhaul simulations, it is assumed that the backhaul is ideal and unconstrained.

LTE-A PERFORMANCE

DOWNLINK

The simulations are based on the IMT-A evaluation document [9], and the
simulation assumptions are outlined in Table 5. Downlink simulation results using
LTE Release 8 features are shown in Table 6 for the base station antenna
configuration shown in Table 5. Although the results shown in Table 5 are for an
FDD deployment, results for TDD systems are similar in terms of spectral
efficiency.

From the results, it is seen that IMT-A requirements can be met for indoor (InH)
and rural macrocell (RMa) environments. This is especially true for the InH
environment, where only a 2 2 antenna configuration is needed. However, DL
performance for urban microcell (UMi) and urban macrocell (UMa) are below the
IMT-A requirements. The shortfall is especially pronounced for the UMa
environment, where approximately 70 percent of the sector and 60 percent of the
cell edge requirements can be achieved.


Table 5. Main simulation assumptions.

Table 7 provides DL performance results using the LTE-A MIMO technique
described earlier enhanced single-cell DL MU-MIMO. From the table it is seen
that the performance improves substantially in all cases and that the IMT-A
requirements can be met in all environments. Next, the DL performance using a
different workload (bursty traffic model) is summarized. The bursty traffic model
is important to characterize the dependency of the user data rate as a function of
network load and user geometry. The characteristics of this model are as follows:

Model users only for the duration of the file transfer.

Users are randomly created and start file transfers in the simulated network at a
certain rate according to a Poisson process.

Data transfer users are modeled for the duration of the file transfer and then
dropped from the simulation once the file data volume is transferred.

For the period of the file transfer the application is assumed to download as
quickly as the network allows.

In the simulation the UE arrival is random between 0 and 64 s with a total
run time of 128 s. Statistics are collected between 16 and 64 s, and the file size
used is 2 Mbytes. A new metric called resource utilization is defined which is
given by the fraction of the total available air interface resources utilized for user-
generated traffic (except for MU-MIMO or beamforming). In other words, it
equals the average number of RBs used for user traffic over the duration of the
simulation, divided by the number of RBs available to carry user traffic. The DL
performance using 2 2 SU-MIMO with closely placed cross-pole antennas is
summarized in Table 8. It may be observed from the table that the mean user data
rate at 50 percent resource utilization (normal operating point) is around 11 Mb/s,
while the 5 percent edge throughput is approximately 1.8 Mb/s.

Table 6. LTE Release 8 DL performance summary

Table 7. LTE-A DL performance summary


Table 8. Bursty traffic performance with 2 2 SU-MIMO, UMi, 10 MHz.


UPLINK

Uplink simulation results using LTE Release 8 features for two antenna
configurations (1 4 and 1 8 linear arrays of vertically polarized elements with
0.5 separation) under FDD systems are shown in Table 9. Frequency selective
scheduling based on the proportional fair metric is used together with fractional
UL power control. From the results, it is seen that with four receive antennas at the
eNB, all IMT-A requirements can be met using the existing Release 8 functionality
and 1 4 antenna configuration. With eight receive antennas, the cell spectral
efficiency is significantly increased; therefore, more UL overhead can be
accommodated. For instance, under the UMa scenario, the number of PUCCH RBs
can be increased from 4 to 18 (36 percent control overhead) while still satisfying
the IMT-A requirement. For UMi, the number of PUCCH RBs can be similarly
increased from 4 to 16 RBs (32 percent control overhead). Note that the
interference-over-thermal-noise ratio (IoT) in each case was constrained to be in
the range of approximately 911 dB.

Although not shown here, results for TDD systems are similar in terms of spectral
efficiency. In addition, performance under different antenna configurations, with
10 separation and multiple pairs of cross-polarized antennas, was also similar (but
slightly better in this case). Finally, the performance of UL for the bursty traffic
model is summarized in Table 10. It may be noted that the mean user data rate at
50 percent resource utilization is around 10 Mb/s, while the 5 percent edge
throughput is approximately 1.5 Mb/s.


Table 9. LTE Release 8 UL performance summary.







Table 10. Bursty traffic performance with 1 2 SIMO, UMi, 10 MHz.






















CONCLUSION

This article gives a brief overview and description of the concepts of the LTE-A
features and the performance achievable using some of these features. It may be
observed that requirements for IMT-A can be met using these advanced concepts.
Finally, it may be noted that the work on LTE-A is still ongoing, and a follow-up
to this article will be published at a later date.