W H I T E PA P E R
Ralf Bendlin, Ph.D.,
Member of Technical Staff
Runhua Chen, Ph.D.,
Member of Technical Staff
Anthony Ekpenyong, Ph.D.,
Member of Technical Staff
Zhihong Lin,
Strategic Marketing Manager
Eko Onggosanusi, Ph.D.,
Manager, Senior Member of Technical Staff
Texas Instruments
Introduction
With smartphones like the iPhone5, Sam-
sung Galaxy and HTC Droid all equipped
with LTE capabilities, the demands for high-
er data rates from the network will soon be-
come commonplace. Wireless operators are
seeking cost-effective, high-performance
solutions that meet the demands of user
traffic today, and position their network
architecture for the increased demands of
the future. The Third Generation Partnership
Project (3GPP) has been working at length
to establish standardized solutions to ad-
dress these network requirements, mainly
through the Long Term Evolution (LTE) and
LTE-Advanced (LTE-A) initiatives. Texas
Instruments has been a key player in this
standard-making process and has been on
the forefront of incorporating supporting
technology into TI’s KeyStone multicore ar-
chitecture, ensuring availability of LTE-A in
wireless base station System on Chip (SoC).
This paper discusses the major elements
of the LTE-A standard and how it achieves
higher throughput with increased efficien-
cies, as well as how LTE-A is addressed in
TI’s KeyStone multicore SoC architecture.
Embracing LTE-A with
KeyStone SoCs
Abstract
With the rapid acceleration of LTE deployments and the enhanced mobile user experience it
delivers, LTE is clearly emerging as the winning technology for mobile communications. LTE
features a simplified network architecture, higher performance and lower cost per bit, all
things that operators worldwide crave. LTE-A further enhances the spectral efficiency of both
macro cells and small cells. It increases coverage for macro cells while using small cells to
boost capacity. Re-farming existing 2G/3G spectrum into LTE/LTE-A operation with carrier
aggregation will improve spectral efficiency for mobile networks while using higher order
Multiple Input Multiple Output (MIMO) antennas can increase user data throughput to achieve
the required LTE-A data rate of 1Gbps. Finally, coordinated multi-point and enhanced inter-cell
interference coordination at the eNodeB will optimize network efficiency. With more than 10
years of base station expertise and successful field deployments, Texas Instruments is well
positioned to answer the LTE-A network challenges with its KeyStone II SoC architecture,
a highly integrated and scalable SoC for macro and small cell solutions. KeyStone II SoCs
enable flexible network deployment and unlock the potential of LTE-A while providing both
energy and cost efficiency.
LTE overview
LTE is defined in 3GPP release 8 and 9 with flexible carrier spectrum ranging from 1.4-MHz,
3-MHz, 5-MHz, 10-MHz, 15-MHz and 20-MHz bandwidth. It supports peak downlink data
rates of 300Mbps and peak uplink date rates of 75Mbps. LTE offers a simplified network
architecture compared to 2G and 3G deployments with much lower latency and an efficient
transport mechanism enabling higher performance with lower cost compared to legacy mobile
networks. Motivated by further improvements in the cellular user experience in scenarios with
bursty asymmetric traffic, a reduction in cost per bit, an increase in spectrum and energy ef-
ficiency, and the support of heterogeneous networks for higher capacity and better coverage,
3GPP working groups have been working to advance LTE with new features to LTE-A.
2 Texas Instruments

LTE-A requirements 3GPP started an LTE Advanced feasibility study in release 9. The LTE-A specifications are defined in releases
10 onward. In addition to backward compatibility with existing LTE networks, LTE-A lays out new requirements
aimed at improving throughput, peak data rate, and spectral efficiency at reduced cost. The requirements are
defined in the following categories:

Capability requirement
The capability requirements define peak data rate, latency and capacity. LTE-A shall support peak data rates
of 1Gbps in the downlink and 500Mbps in the uplink. Control plane (C-Plane) and user plane (U-Plane)
latency shall be significantly decreased from LTE. The C-Plane latency target is less than 10ms for transition
from dormant state to connected state, and less than 50ms for transition into camped state with the user
plane established. User plane latency is further reduced. Control plane capacity shall support at least 300
users per 5-MHz bandwidth.

System performance requirement

System performance requirements are defined in terms of peak, average and cell edge spectral efficiency.
Peak spectral efficiency is based on perfect channel conditions with all radio resources concentrated into
a single user equipment (UE). The target peak spectral efficiency of LTE-A is 30bps/Hz in the downlink and
15bps/Hz in the uplink.
Average spectral efficiency is defined as the aggregate throughput of all users in one cell normalized
by the overall cell bandwidth. The average spectral efficiency is measured in bps/Hz/cell. It is use-case
dependent and LTE-A targets the average spectral efficiency to be as high as possible due to various channel
conditions.
The cell-edge-user-spectral efficiency is defined as the 5% point of the cumulative distribution function
(CDF) of the user throughput normalized to the overall cell bandwidth. Cell-edge-user throughput is mea-
sured in bps/Hz/cell/user and LTE-A targets the cell edge user throughput to be as high as possible.
Other system performance requirement improvements in LTE-A over LTE include VoIP capacity improve-
ment, up to 10km/h mobility enhancement over LTE and further enhanced Multimedia Broadcast/Multicast
Services (MBMS) for better spectral efficiency.

Deployment requirement
LTE-A is expected to deploy heterogeneous networks with increased small cell density. It provides spectrum
flexibility in addition to new spectrum bands and enables operation in wider spectrum with up to 100-MHz
bandwidth possibly aggregated through contiguous or non-contiguous inter- or intra-frequency band carrier
aggregation. Wider bandwidths enable higher peak throughput in LTE-A deployments. LTE-A shall also con-
tinue to support co-existence and interworking with legacy 2G/3G networks as was required for LTE.

Embracing LTE-A with KeyStone SoCs October 2012

 Texas Instruments 3

To meet those requirements, LTE-A defines multiple techniques to achieve the capacity, performance and
deployment goals, so let’s look at those techniques in more detail. (In the following sections, we also use the
term eNodeB to denote an LTE base station entity. User equipment will also be denoted as UE.)

LTE-A enabling Significant work has been done in several areas of 3GPP release 10 and onward to fulfill LTE-A requirements.
technologies Carrier aggregation is aimed at improving spectrum flexibility to provide wider bandwidths resulting in higher
data throughput and is also useful to reuse legacy spectrum for LTE operation. Higher order and multi-user
MIMO, up to 8×8 in the downlink and 4×4 in the uplink are proposed to increase spectral efficiency. In 3GPP
release 11, coordinated multi-point (CoMP) is adopted with the goal of improving coverage, enhancing cell
edge throughput and increasing the overall system performance. In addition, enhanced inter-cell interference
coordination (eICIC) is introduced for increased network efficiency and a better user experience.

Carrier aggregation (CA)

One of the solutions to increase peak data rate is to increase the transmission bandwidth. LTE-A leverages
the carrier aggregation (CA) concept to combine multiple discrete carriers called component carriers (CC)
into a single virtual wide-band carrier, up to five component carriers and up to 100MHz of spectrum can be
aggregated in LTE-A systems.
Each component carrier in carrier aggregation systems can be adjacent or discrete inside one frequency
band with carrier separation to be a multiple of 300KHz; this is called intra-band carrier aggregation. Inter-
band carrier aggregation is also supported in LTE-A where different component carriers can reside in differ-
ent frequency bands. Figure 1 illustrates how carrier aggregation can be constructed through different carrier
frequencies using two carriers as an example.

CC1 CC2

f Band 1
Intraband Contiguous CA

Carrier
separation

CC1 CC2

f Band 1
Intraband Non-Contiguous CA

Carrier+band
separation

CC1 CC2

f Band 1 f Band 2
Interband Non-Contiguous CA
Figure 1: Component carrier’s scenarios in carrier aggregation

Embracing LTE-A with KeyStone SoCs October 2012

4 Texas Instruments

LTE-A release 10 is restricted to two component carriers for carrier aggregation, and each CC can be of
1.4-MHz, 3-MHz, 5-MHz, 10-MHz, 15-MHz or 20-MHz bandwidth. The maximum number of aggregated car-
riers may be increased in later releases. Each component carrier is considered as a serving cell with different
coverage area. The primary component carrier constitutes the primary serving cell (PCell) and the rest of CCs
constitute the secondary serving cells (SCell). PCell handles radio resource management (RRM) for the UE,
it cannot be deactivated and is only changed by a handover. The SCells can be configured to the same UE
to boost data rate. Independent configuration of downlink and uplink transmission modes is handled in the
Media Access Control (MAC) and Physical (PHY) layer.
There are many benefits to using carrier aggregation in LTE-A systems, e.g., combined and wider spec-
trum to increase the data throughout, to increase the cell coverage, and especially to improve the cell-edge
performance. CA can also mitigate interference in heterogeneous networks where a small cell and a macro
cell are using the same carrier frequency. If a UE experiences too much interference, CA in the macro cell
provides an additional carrier and the macro cell can schedule the UE to a different carrier to mitigate the
interference and improve the user experience.
Carrier aggregation imposes additional complexity in base station designs. CA with two component
carriers can be viewed as transmitting with two RF chains to the same UE. At the MAC layer, a UE can be
configured with two data streams each of which is destined for a different RF component in the PHY layer.
Determination of whether a UE should be configured for carrier aggregation is done by the scheduler after
assessing the amount of the data received from higher layers for a specific UE. The downlink scheduler has
to make sure it can allocate a Physical Downlink Control Channel (PDCCH) for scheduling the Physical Down-
link Shared Channel (PDSCH) for each component carrier and some priority must be given to a user that is
configured for CA. At the PHY layer, each CC independently goes through the same PHY procedure as before,
and a new Physical Uplink Control Channel (PUCCH) format 3 signaling is introduced to support downlink CA.
CA complexity can also affect the analog front end, especially with inter-band CA. Typical band separation
can be up to 1GHz, two independent RF data paths for each data converter and power amplifier (PA) will be
needed and careful synchronization between the two data paths need to be devised to ensure the integrity
of the signal chain. For intra-band CA, if the carrier separation is narrow enough so that it can fit into the
dynamic frequency range of one analog front end, two CC data paths can be aggregated using a single ana-
log device. In this case, single band CA could reduce the complexity of an eNodeB. Figure 2 on the following
page illustrates the base station data flow for downlink carrier aggregation on separate data paths.

Multi-antenna techniques and higher order MIMO

MIMO techniques are widely used in commercial wireless systems as one of the most effective means of
obtaining spectral efficiency improvements. By creating additional degrees of freedom in the spatial domain,
MIMO significantly improves the signal-to-noise ratio (SNR), radio link robustness and spectrum efficiency
yielding substantial performance gain over single-antenna technologies. Two categories of MIMO schemes

Embracing LTE-A with KeyStone SoCs October 2012

 Texas Instruments 5

CC1

DAC/
PA

scheduler
IQMod

Multiplexer
HARQ PHY/
RRC PDCP RCL Digital
HARQ Radio Front
DAC/
End IQMod PA

MAC

CC2

Figure 2: Carrier aggregation with two CC eNodeB downlink data path example

are supported in LTE: transmit diversity and spatial multiplexing. Transmit diversity sends multiple encoded
copies of a single data stream resulting in improved signal robustness against adverse channel fading and
cell coverage. Spatial multiplexing, on the other hand, leverages channel-dependent beamforming to transmit
multiple streams concurrently, thereby achieving higher data throughput.
Advanced MIMO techniques in release 10 include higher order MIMO (up to 8×8) and multi-user MIMO
(MU-MIMO) beamforming. Increasing the antenna array size to 8 effectively doubles the peak data rate com-
pared to releases 8/9 and is the only solution to meet the ITU-R requirement of 30bps/Hz spectral efficiency.
Advanced MU-MIMO beamforming allows multiple spatially separated users to be scheduled in the same
spectrum, thereby achieving higher spatial reuse factor and improving the spectral efficiency. MU-MIMO is
arguably one of the major enhancements in LTE-A that directly results in significant cell-average through-
put gains (e.g., 20–30% in favorable channel condition). This, however, leads to several challenges in the
beamforming and scheduler implementation which often requires cross-layer optimization between the PHY
and MAC layers. Texas Instruments has conducted extensive research in the area of PHY/MAC scheduler
optimization to fully reap the benefits of LTE-A MIMO techniques.
Flexible and energy-efficient operation of wireless networks is increasingly critical for mobile operators.
Toward this goal, LTE-A brings a paradigm shift in facilitating UE-specific MIMO beamforming enabled by
newly introduced demodulation reference signals (DMRS) and channel state information reference signals
(CSI-RS). In contrast to release 8 codebook-based (use of a predefined precoding look-up table for MIMO
precoding) MIMO transmission with cell-specific reference signals (CRS), LTE-A beamforming allows more
flexibility for the base station with non-codebook-based transmission as enabled by DMRS. The UE-specific
and low-density nature of DMRS/CSI-RS allows the eNodeB to dynamically adjust transmission activities (e.g.,
ON and OFF) in accordance with network traffic variations, making it possible to reduce the overall energy
consumption. In addition, non-codebook-based beamforming with DM-RS allows arbitrary beamforming at

Embracing LTE-A with KeyStone SoCs October 2012

6 Texas Instruments

1
Mux CW1

Scheduler

Precoding
UE1 Channel UE1
coding
Data Modulation CW2
8
DL SU-MIMO
eNodeB

UE1
Mux
1
CW1

Scheduler

Precoding
UE1 Channel
coding
UE2 Modulation CW2
Data
8
UE2
eNodeB
DL MU-MIMO
Figure 3: Downlink SU-MIMO and MU-MIMO

the eNodeB, which is a prerequisite for advanced MU-MIMO operation. Figure 3 shows the DL SU-MIMO and
MU-MIMO data paths: SU-MIMO increases peak user throughput by dedicating all antenna resources to a
single UE whereas MU-MIMO improves the network spectral efficiency.
In the uplink, LTE-A adds an enhancement in PUCCH to allow the UE to use two antennas by adopting spe-
cial orthogonal resource transmit diversity (SORTD) in all PUCCH formats. This ensures the orthogonality of
each transmit antenna achieves better SNR at the receive antenna with the combined signal. Uplink transmit
diversity can reduce transmit power at the UE to meet the eNodeB PUCCH SNR requirement and it can lower
the UL inter-cell or intra-cell interference and prolong UE battery life.
Early LTE releases supported a single transmit antenna at the UE. Accordingly, UL spectral efficiency can
be increased by MU-MIMO whereby two users are assigned the same UL frequency allocation. To reduce
mutual interference, the eNodeB scheduler may select a pair of users whose channels are mutually orthogo-
nal or near-orthogonal. LTE-A further enhances UL spectral efficiency by introducing support of two or four
transmit antennas at the UE making possible up to 4×4 SU-MIMO transmission from a single UE. Another
key enhancement of LTE-A is the scheduling flexibility now available to the eNodeB. For example, the eNodeB
can schedule a single UE for 4-layer SU-MIMO transmission, or it can operate in MU-MIMO by scheduling
two UEs each with two-layer transmission as shown in Figure 4 on the following page.
To facilitate UL MIMO transmission, LTE-A also introduces a time-domain spreading technique across the
UL demodulation reference signals (UL DMRS) known as orthogonal cover codes. This feature allows pairing
of users with unequal UL allocations and is also used to ensure orthogonal DMRS across different antennas.

Coordinated multi-point (CoMP)

Cells in conventional wireless networks operate in a semi-independent manner, where downlink transmission
and uplink reception are performed in each cell independently. Consequently, severe inter-cell interference

Embracing LTE-A with KeyStone SoCs October 2012

 Texas Instruments 7

1
DeMux CW1

Scheduler

Receiver
Channel UE1

MIMO
UE1
Decoding
Data DeMod CW2
4
UL SU-MIMO
eNodeB

UE1
DeMux
1
CW1

Scheduler
UE1

Receiver
Channel

MIMO
Decoding
UE2 DeMod CW2
Data
4
UE2
eNodeB
UL MU-MIMO
Figure 4: Uplink SU-MIMO and MU-MIMO examples

may become a fundamental bottleneck particularly for cell edge users. This problem will be increasingly
severe if there is a rapid proliferation of small cells without interference management.
CoMP is an LTE release 11 feature that allows multiple geographically distributed points (e.g., eNodeBs
or cells) to optimize their transmission/reception activities in a coordinated fashion, so that strong inter-cell
interference can be avoided or even transformed into signals that a UE can decode to improve coverage. By
leveraging multiple cells or remote radio heads (RRH) in either homogeneous and heterogeneous networks,
joint transmission and reception to and from a single UE enable a more flexible network topology and optimi-
zation, mitigate multi-cell interference, enable dynamic load balancing, allow UEs to select the closest base
station point for transmission and reception, which in turn will reduce the UE power consumption resulting in
increased battery life in addition to a boost in data throughput and user experience.
More importantly, CoMP allows radio resources of different cells to be centralized and adaptively allocated
to better handle the traffic fluctuations in each eNodeB, thereby achieving increased resource utilization
across the entire network. For instance, a cell with low traffic loading may hand over its users to a neighbor-
ing cell and be turned OFF to reduce inter-cell interference and improve the overall system performance. Dy-
namic radio resource management like this is one of the primary motivations for C-RAN (Cloud/Centralized/
Clean/Coordinated Radio Area Network) and is critical for future high-throughput and low-energy network
operation. It also facilitates improved network architectures, e.g., distributed antenna systems (DAS) with
geographically separated RRHs which are connected to a centralized baseband processor via optical fiber.
Clearly, efficient PHY/MAC scheduler designs must be considered to realize these benefits without incurring
prohibitive network processing complexity.
The CoMP participants are grouped into CoMP sets. Such grouping is performed in homogeneous networks
featuring a single eNodeB with multiple sectors or multiple high-transmit power RRH, and similarly in hetero-
geneous networks where macro cells and small cells are deployed either with the same or different cell IDs.

Embracing LTE-A with KeyStone SoCs October 2012

8 Texas Instruments

CoMP Set

Signal

Signal

Figure 5: CoMP scheme: joint transmission and reception

There are two main CoMP schemes, joint processing and coordinated scheduling/beamforming. Joint
processing (JP) includes joint transmission (JT) and joint reception (JR) which enable simultaneous data
transmission and reception from multiple points within a CoMP set to a single or multiple UEs, respectively.
Dynamic point selection/blanking (DPS/DPB), as part of joint processing, requires PDSCH data availability at
multiple points as well, but the data are only transmitted from one point at any given time to mitigate interfer-
ence and achieve the best performance. Figure 5 shows a CoMP set in a heterogeneous network where a
macro cell and one small cell perform joint transmission and reception, UE receive valid signals from both
eNodeBs, where in a non-CoMP network, one of the eNodeB transmissions would result in interference to
the UE.
When user data is only transmitted from one point, coordinated scheduling/beamforming (CS/CB) can
be used to improve performance by coordinating frequency assignments and beamforming vectors across
transmission points.
Heterogeneous networks create new challenges in uplink interference handling. New cell boundaries are
created by small cells within the macro area and higher uplink interference is seen due to the smaller path
loss between a UE and a victim cell. Interference randomization using cell IDs is insufficient when small cells
and the macro cell share the same cell ID because PUSCH/PUCCH capacity is limited. CoMP introduces the
virtual cell IDs (VCID) for PUSCH and PUCCH to increase the degrees of freedom for the uplink-cell-ID-based
randomization resulting in improved inter-cell orthogonality and reduced interference.
CoMP introduces new dimensions in scheduling complexity. The scheduler needs to be coordinated across
several points, e.g., user data and scheduling decisions need to be shared, and CoMP across different
geographically dispersed eNodeBs also requires additional backhaul capacity to carry the CoMP data. Link
adaptation and scheduling are more complicated especially when multiple CoMP schemes are combined.
Simulation results show that leveraging CoMP can improve cell edge spectral efficiency by 20–25%, while
only marginal gain is seen in cell average throughput. The following information, as well as Figure 6 on the
following page, show cell average and cell edge performance comparisons for the following scenarios (the
performance gain is depending on the network topology and traffic patterns):

Embracing LTE-A with KeyStone SoCs October 2012

 Texas Instruments 9

• Homogenous macro – 57 macro eNodeBs, 4 Tx antennas, 30 users per macro

• Heterogeneous uniform: non-CoMP – same homogenous macro cells overlaid with four small cells
per macro, uniform user distribution, without CoMP
• Heterogeneous uniform: CoMP – same heterogeneous network with CoMP enabled
• Heterogeneous cluster: non-CoMP – heterogeneous network with clustered user distribution, simu-
lating hot spot situation, without CoMP
• Heterogeneous cluster: CoMP – heterogeneous network with clustered user distribution with CoMP
enabled

0.16
homogenous macro homogenous macro

5% -perc entile c ell-edge spec ial effic iency (bps /Hz )

4 heterogenous uniform: non-CoMP
heterogenous uniform: non-CoMP 0.14
heterogenous uniform: CoMP heterogenous uniform: CoMP
cell average special efficiency (bps/Hz)

3.5 heterogenous cluster: non-CoMP heterogenous cluster: non-CoMP

0.12
heterogenous cluster: CoMP heterogenous cluster: CoMP
3
0.1
2.5

0.08
2

0.06
1.5

1 0.04

0.5 0.02

0
1 0
1

Cell-average gain 5% cell-edge gain

Figure 6: Cell average and cell edge throughput comparison with and without CoMP

Enhanced inter-cell interference coordination (eICIC)

Over the course of a day, the load in the network fluctuates as users move around the network. Residential
areas see most of the activity during the evening hours when people are home from work whereas during
business hours peak traffic occurs in offices, schools and downtown areas. On the weekend, users don’t go
to office parks but to malls, sport events, and bars or restaurants. These dynamics in user behavior render
static network planning sub-optimal. One way to dynamically partition resources is by means of cell range
expansion (CRE). CRE uses cell-specific offsets called CRE bias to favor one layer of eNodeB over another. For
instance, as depicted in Figure 7 below, small cells could be favored during business hours when users are

Cell Range Expansion(CRE)

‰ UE connects to macro
eNodeB without CRE and to
pico eNodeB with CRE
‰ UE does not connect
CRE to strongest cell with CRE
Pico eNodeB
Macro eNodeB

Figure 7: HetNet load balancing through Cell Range Expansion (CRE)

Embracing LTE-A with KeyStone SoCs October 2012

10 Texas Instruments

at work or school whereas the macro layer could be favored during evening hours when users are at home.
This way, both CAPEX and OPEX can be reduced for the operator and user experience can be enhanced for
the end-consumer.
When no bias is configured, one in three users is connected to a macro base station. However, less than
one in five users connects to the macro layer when the CRE bias exceeds 6dB or more. This is shown in
Figure 8.

Figure 8: Percentage of small cell user offload using different CRE bias

Unfortunately, there is no free lunch and consequently, user offloading by means of CRE increases the
interference seen by the offloaded users connected to the small cells. To mitigate this inter-cell interference,
LTE-A introduces a time-domain inter-cell interference coordination feature called almost blank subframes
(ABS). This feature is often referred to as enhanced ICIC or eICIC. ABS creates protected resources, namely
subframes during which the small cell layer experiences significantly reduced inter-cell interference. The
small cell layer can take advantage of this by scheduling cell edge users during this ABS whereas only cell
interior users are scheduled in normal subframes. The macro layer, on the other hand, schedules no or only
cell interior users during protected subframes and cell-edge users otherwise. eICIC requires tight synchroni-
zation and coordination among eNodeBs due to its time-domain nature. Since the CRE bias is user specific,
the network can use even more aggressive biases for advanced UEs which have interference canceling
receivers. These can tolerate a higher level of inter-cell interference resulting from larger CRE biases for they
are still able to detect and communicate with small cells under such extreme configurations.
Since CRE and ABS are adaptive and dynamic they can be thought of as an aspect of self-organizing
networks (SON). In particular, the eNodeB provides radio resource management functions which allow for
optimization of both the CRE bias as well as the ratio of protected ABS resources. Thus, resources can be
optimally tailored towards the current load in the network resulting in more energy- and spectral-efficient
operation as well as lowered costs for the operator. eICIC can also be thought of as a complement to CoMP
as both aim towards improving the throughput at the edge of a cell. CoMP tries to harness the fact that cell
edge users receive multiple strong signals from several eNodeBs whereas eICIC tries to avoid dominant

Embracing LTE-A with KeyStone SoCs October 2012

 Texas Instruments 11

interference from nearby neighboring cells. This gives the network operator ample control and choice to opti-
mize heterogeneous networks where small cells increase the number of cell-edge users due to the introduc-
tion of additional cells.

Downlink and uplink signaling enhancement

There is a general trend in LTE-A to move from cell-specific signals and channels to user-specific ones.
The reasons for this change are crucial to advancing LTE towards a more efficient and thus greener mobile
communications standard while meeting the demand for improvements in throughput, spectral efficiency, and
end-user experience that is expected from LTE-A. Cell-specific signals and channels are transmitted across
the entire system bandwidth in each physical resource block (PRB). In addition, they are present in each sub-
frame. Being cell specific, they are static and suboptimal for modern dense heterogeneous network deploy-
ments which strive for utmost energy and spectral efficiency.
In LTE, cell-specific reference signals (CRS) are used, amongst other things, for estimating the channel
from the eNodeB to the UE for demodulation as well as CSI feedback. CRS is un-precoded, an important fea-
ture since all UEs in a cell use the same CRS irrespective of their configured transmission mode, speed, loca-
tion, throughput requirements and so forth. Consequently, CRS-based MIMO operation requires codebook-
based precoding because the UEs use unprecoded reference signals to estimate the channel. In addition,
CRS is tied to the Physical Cell ID (PCI) of the transmitting cell which imposes undesired restrictions to the
operation of certain deployment scenarios in heterogeneous networks with dense small cell eNodeBs. Other
potentially detrimental characteristics of CRS in dense small cell deployments are that they create interfer-
ence to neighboring cells and that they consume a lot more energy because they are transmitted across the
entire system bandwidth in all sub-frames.
Some of these properties have been addressed in LTE-A with the introduction of user-specific reference
signals (RS), namely, CSI-RS and demodulation RS (DMRS). As the name suggests, CSI-RS is used by the
UE to estimate the channel for CSI feedback to the serving cell. Signaling overhead and energy consumption
can be significantly reduced, e.g., by configuring larger reporting intervals for low mobility UEs and shorter
ones for high-speed UEs, respectively. Consequently, CSI-RS are not present in each sub-frame as they were
with legacy CRS, resulting in reduced overhead and greater efficiency. CSI-RS is also required to support the
advances LTE-A brings forward relating to MIMO and CoMP as mentioned before.
Similarly, DMRS can be used by the UE to estimate the channel for demodulation purposes. In contrast
to CSI-RS and CRS, DMRS is pre-coded and only present in those PRBs actually used to transmit data to a
given UE. This not only further increases the efficiency of the system’s energy consumption and signaling
overhead, but more importantly, allows for non-codebook-based precoding which is desirable, for instance, in
TDD deployments to piggyback channel reciprocity.
By making reference signals user-specific, they can also be decoupled from the PCI, an important feature
for several CoMP operation scenarios both in the downlink and uplink. By configuring “virtual” cell IDs per

Embracing LTE-A with KeyStone SoCs October 2012

12 Texas Instruments

user, the network can choose the optimal cell in a semi-static or even dynamic fashion for greater energy
efficiency and increased user throughput. These enhancements especially favor cell-edge users who might
receive signals of almost equal strength from more than one eNodeB. In addition, utmost care was taken in
the design of these new features to facilitate seamless inter-operability with legacy UEs, for example to reap
MU-MIMO, cell splitting, or diversity gains through interference averaging or avoidance.
Similar considerations have led to the standardization of an enhanced Physical Downlink Control Chan-
nel (ePDCCH) which has been cell specific in LTE, and was similarly transmitted across the entire system
bandwidth. The ePDCCH is configurable for transmission in a subset of PRBs allowing for coordination among
eNodeBs to reduce inter-cell interference. Borrowing from the design of the LTE-shared data channel, the
ePDCCH harnesses multi-user diversity and beamforming gains to increase robustness and system perfor-
mance. The ePDCCH extends LTE as it develops into even more energy-efficient deployments and supports
the evolution of many features envisioned beyond LTE-A.

LTE-A system KeyStone II is the second generation of Texas Instruments’ multicore SoC architecture. It is optimized to
solutions with deliver high throughput and low latency as well as well as scalability and flexibility to support LTE-A. Figure 9
KeyStone II SoCs
shows the KeyStone II architecture.

Multicore Navigator

Power Mgr SysMon

+= * +
ARM Cores - << - DSP Cores Debug EDMA
TeraNet

Layer 1 Acceleration
Radio processing
ARM shared L2 DSP L2
Digital Radio Front End
DUC/DDC/CFR/DPD

Multicore Shared Memory Controller Shared L3

Packet and Security
Acceleration

DDR3 L DDR3 L Ethernet Switch

64/72b 64/72b

EMIF I2C, UART SPI PCIe CPRI/OBSAI JESD204B Hyperlink USB 3 SRIO Ethernet Ethernet Ethernet Ethernet

Figure 9: KeyStone II architecture

At the heart of the KeyStone II architecture are the SoC infrastructure elements, Multicore Navigator and
TeraNet. Multicore Navigator provides architecture parallelism with a unified interface for cores, accelerators
and I/O using hardware queues and packet Direct Memory Access (DMA) for communication, data transfer
and task management. TeraNet is the non-blocking SoC interconnect that provides high throughput to make
possible the high data rates required in LTE-A. The KeyStone II SoC infrastructure offers highly scalable

Embracing LTE-A with KeyStone SoCs October 2012

 Texas Instruments 13

LTE solutions from small cells to macro cells and cloud RAN eNodeBs with lowest communication latency.
Multicore Navigator also enables hardware virtualization and facilitates load balancing in advanced network
deployments.
The KeyStone II architecture also features Layer 1 radio acceleration to enable cost- and energy-efficient
Layer 1 processing. A Bit Rate Coprocessor (BCP) that is incorporated into the Layer 1 accelerationPac al-
ready has many LTE-A features built in to offload cycle-intensive bit-level operations from the DSP cores. DL
and UL shared-channel processing tasks including rate matching, scrambling, interleaving, Hybrid automatic
repeat request (HARQ) combining and Log-likelihood ratio (LLR) computations are all performed by the BCP
hardware.
A primary use case that showcases the fast and efficient bit-level processing offered by the BCP is in itera-
tive processing that may be required for UL SU and MU-MIMO in LTE and LTE-A. For LTE UL MIMO process-
ing, DFT-spread transmission in Single Carrier Frequency Division Multiple Access (SC-FDMA) results in pro-
hibitive complexity for maximum likelihood-based receiver algorithms due to the DFT spreading. Alternatively,
iterative processing solutions, based on the turbo equalization concept, offer satisfactory performance with
reasonable complexity. The BCP offloads iterative processing in the form of parallel interference cancellation
(PIC) and successive interference cancellation (SIC) techniques.
When applied to UL MIMO – both SU-MIMO and MU-MIMO in LTE-A – SIC cancels out the reconstructed
signal corresponding to a selected MIMO user from the original received signal. This effectively cancels
out the interference experienced by the other users due to the first user. The “cleaner” residual signal after
cancellation is then further processed to successively decode the other MIMO users. Multiple stages of re-
construction and cancellation are also possible to improve signal detection. A similar concept is used for PIC,
where cancellation takes places simultaneously on all users in one stage.
The BCP also supports an advanced PUCCH receiver that shows 1–2 dB gain over a conventional MRC
receiver. The processing complexity associated with an advanced receiver is offloaded to the BCP and can
be used to improve performance for PUCCH formats with large codebooks such as PUCCH Format 2 and the
new LTE-A PUCCH Format 3.
LTE-A introduces new implementation complexity especially in the MAC layer scheduler for many of its new
features such as carrier aggregation, CoMP and eICIC. The KeyStone II architecture with heterogeneous low-
power and high-performance DSP and ARM® cores, together with the Multicore Shared Memory Controller
(MSMC), provides low-latency access to DDR3 external memory yielding an excellent low-latency computa-
tion platform for LTE-A’s performance needs.
With powerful I/O connectivity, KeyStone II SoCs are the ideal solution for many different network topolo-
gies and scalable network deployments. The antenna interface supports CPRI 5.0 with 9.8Gbps per lane and
built-in antenna switching capabilities. It enables eight LTE 20-MHz channels per lane and allows flexible
solutions for LTE-A carrier aggregation and higher-order MIMO with distributed remote radio heads. Gigabit
Ethernet with a built-in Ethernet switch reduces the overall system processing latency for LTE-A transport

Embracing LTE-A with KeyStone SoCs October 2012

14 Texas Instruments

data processing. 50-Gbps HyperLink provides a low latency and transparent inter SoC connection so com-
plicated software protocols are unnecessary. HyperLink also enables scalable macro to cloud RAN solutions,
while JESD204B provides direct data converter interfaces enabling low-power and low-cost small-cell solu-
tions. With optimized DSP and ARM® processing horsepower, KeyStone SoCs enable lowest power-per-bit to
process all of the new LTE-A features for macro and small cells. Figure 10 depicts some exemplary macro
and cloud RAN solutions implemented with KeyStone II SoCs.

Macro or Cloud RAN eNodeB, Layer 1, 2+

CPRI CPRI
KeyStone II KeyStone II KeyStone II
SoC
Ethernet SoC
Ethernet SoC

HyperLink HyperLink

KeyStone II KeyStone II KeyStone II

SoC SoC SoC

Integrated on-chip Ethernet and antenna switching

CPRI over Fiber

RRH
Remote Radio Head

Figure 10: Building LTE-A enabled macro or Cloud RAN eNodeBs with the KeyStone II architecture

Conclusion With the promise of better coverage, much higher throughput and lower latency, the LTE-Advanced standard
is poised to deliver a vastly improved mobile user experience. TI’s KeyStone II multicore architecture provides
highly integrated Layer 1, 2, 3 and transport SoC solutions that unleash the abundant throughput and spec-
tral efficiency offered through various LTE-A features. In addition, Keystone II boasts the lowest processing
latency for low-cost and energy-efficient LTE-A solutions. With a flexible architecture, efficient hardware ac-
celerators, a superior SoC infrastructure, high-speed I/O and programmable low-power DSP and ARM cores,
KeyStone II is the ideal architecture to drive high-performance and scalable heterogeneous networks and to
unlock the potential of LTE-A.

Important Notice: The products and services of Texas Instruments Incorporated and its subsidiaries described herein are sold subject to TI’s standard terms and
conditions of sale. Customers are advised to obtain the most current and complete information about TI products and services before placing orders. TI assumes no
liability for applications assistance, customer’s applications or product designs, software performance, or infringement of patents. The publication of information
regarding any other company’s products or services does not constitute TI’s approval, warranty or endorsement thereof.

© 2012 Texas Instruments Incorporated SPRY218

 IMPORTANT NOTICE
Texas Instruments Incorporated and its subsidiaries (TI) reserve the right to make corrections, enhancements, improvements and other
changes to its semiconductor products and services per JESD46, latest issue, and to discontinue any product or service per JESD48, latest
issue. Buyers should obtain the latest relevant information before placing orders and should verify that such information is current and
complete. All semiconductor products (also referred to herein as “components”) are sold subject to TI’s terms and conditions of sale
supplied at the time of order acknowledgment.
TI warrants performance of its components to the specifications applicable at the time of sale, in accordance with the warranty in TI’s terms
and conditions of sale of semiconductor products. Testing and other quality control techniques are used to the extent TI deems necessary
to support this warranty. Except where mandated by applicable law, testing of all parameters of each component is not necessarily
performed.
TI assumes no liability for applications assistance or the design of Buyers’ products. Buyers are responsible for their products and
applications using TI components. To minimize the risks associated with Buyers’ products and applications, Buyers should provide
adequate design and operating safeguards.
TI does not warrant or represent that any license, either express or implied, is granted under any patent right, copyright, mask work right, or
other intellectual property right relating to any combination, machine, or process in which TI components or services are used. Information
published by TI regarding third-party products or services does not constitute a license to use such products or services or a warranty or
endorsement thereof. Use of such information may require a license from a third party under the patents or other intellectual property of the
third party, or a license from TI under the patents or other intellectual property of TI.
Reproduction of significant portions of TI information in TI data books or data sheets is permissible only if reproduction is without alteration
and is accompanied by all associated warranties, conditions, limitations, and notices. TI is not responsible or liable for such altered
documentation. Information of third parties may be subject to additional restrictions.
Resale of TI components or services with statements different from or beyond the parameters stated by TI for that component or service
voids all express and any implied warranties for the associated TI component or service and is an unfair and deceptive business practice.
TI is not responsible or liable for any such statements.
Buyer acknowledges and agrees that it is solely responsible for compliance with all legal, regulatory and safety-related requirements
concerning its products, and any use of TI components in its applications, notwithstanding any applications-related information or support
that may be provided by TI. Buyer represents and agrees that it has all the necessary expertise to create and implement safeguards which
anticipate dangerous consequences of failures, monitor failures and their consequences, lessen the likelihood of failures that might cause
harm and take appropriate remedial actions. Buyer will fully indemnify TI and its representatives against any damages arising out of the use
of any TI components in safety-critical applications.
In some cases, TI components may be promoted specifically to facilitate safety-related applications. With such components, TI’s goal is to
help enable customers to design and create their own end-product solutions that meet applicable functional safety standards and
requirements. Nonetheless, such components are subject to these terms.
No TI components are authorized for use in FDA Class III (or similar life-critical medical equipment) unless authorized officers of the parties
have executed a special agreement specifically governing such use.
Only those TI components which TI has specifically designated as military grade or “enhanced plastic” are designed and intended for use in
military/aerospace applications or environments. Buyer acknowledges and agrees that any military or aerospace use of TI components
which have not been so designated is solely at the Buyer's risk, and that Buyer is solely responsible for compliance with all legal and
regulatory requirements in connection with such use.
TI has specifically designated certain components which meet ISO/TS16949 requirements, mainly for automotive use. Components which
have not been so designated are neither designed nor intended for automotive use; and TI will not be responsible for any failure of such
components to meet such requirements.
Products Applications
Audio www.ti.com/audio Automotive and Transportation www.ti.com/automotive
Amplifiers amplifier.ti.com Communications and Telecom www.ti.com/communications
Data Converters dataconverter.ti.com Computers and Peripherals www.ti.com/computers
DLP® Products www.dlp.com Consumer Electronics www.ti.com/consumer-apps
DSP dsp.ti.com Energy and Lighting www.ti.com/energy
Clocks and Timers www.ti.com/clocks Industrial www.ti.com/industrial
Interface interface.ti.com Medical www.ti.com/medical
Logic logic.ti.com Security www.ti.com/security
Power Mgmt power.ti.com Space, Avionics and Defense www.ti.com/space-avionics-defense
Microcontrollers microcontroller.ti.com Video and Imaging www.ti.com/video
RFID www.ti-rfid.com
OMAP Applications Processors www.ti.com/omap TI E2E Community e2e.ti.com
Wireless Connectivity www.ti.com/wirelessconnectivity

Mailing Address: Texas Instruments, Post Office Box 655303, Dallas, Texas 75265
Copyright © 2012, Texas Instruments Incorporated