Beruflich Dokumente
Kultur Dokumente
Table 4-6 LTE Carrier Parameters
AIRCOM International ASSETV8.0- LTE Application Notes Page 28 of 64
4.5 Bearers
Bearers represent the air interface connections, performing the task of transporting voice and data
information between cells and terminal types. After bearers have been defined, you can then decide
which ones will be supported by your different services. The following figure presents the ASSET GUI
for the definition of LTE Bearers.
Figure 4-8 Definition of LTE Bearers in ASSET
The Default Uplink and Downlink LTE bearers are defined per CQI providing 15 DL bearers and 4 UL
bearers. CQI is a report sent from the UE to the eNodeB suggesting the appropriate Modulation and
Coding to be used by the eNodeB when transmitting in order to maintain a Block Error Ratio (BLER)
less than 10% at the RLC level. The eNodeB is finally deciding upon MCS depending on CQI and
other (vendor dependent) related measurements. Downlink MCSs are 32. Each default Bearers has
Control & Traffic SINR requirements according to Table 3-8.
Figure 4-9 Bearer SINR requirements
AIRCOM International ASSETV8.0- LTE Application Notes Page 29 of 64
The following table is a vendor specific mapping of CQIs to MCSs.
MCS Index Modulation Coding rate x 1024 Efficiency Comments Code Rate
0 2 120 0.2344 from CQI table 0.1171875
1 2 157 0.3057 Average Efficiency 0.15332031
2 2 193 0.377 from CQI table 0.18847656
3 2 251 0.4893 Average Efficiency 0.24511719
4 2 308 0.6016 from CQI table 0.30078125
5 2 379 0.7393 Average Efficiency 0.37011719
6 2 449 0.877 from CQI table 0.43847656
7 2 526 1.0264 Average Efficiency 0.51367188
8 2 602 1.1758 from CQI table 0.58789063
9 2 679 1.3262 Average Efficiency 0.66308594
10 4 340 1.3262 overlap 0.33203125
11 4 378 1.4766 from CQI table 0.36914063
12 4 434 1.69535 Average Efficiency 0.42382813
13 4 490 1.9141 from CQI table 0.47851563
14 4 553 2.1602 Average Efficiency 0.54003906
15 4 616 2.4063 from CQI table 0.6015625
16 4 658 2.5684 Average Efficiency 0.64257813
17 6 438 2.5684 overlap 0.42773438
18 6 466 2.7305 from CQI table 0.45507813
19 6 517 3.0264 Average Efficiency 0.50488281
20 6 567 3.3223 from CQI table 0.55371094
21 6 616 3.6123 Average Efficiency 0.6015625
22 6 666 3.9023 from CQI table 0.65039063
23 6 719 4.21285 Average Efficiency 0.70214844
24 6 772 4.5234 from CQI table 0.75390625
25 6 822 4.8193 Average Efficiency 0.80273438
26 6 873 5.1152 from CQI table 0.85253906
27 6 910 5.33495 Average Efficiency 0.88867188
28 6 948 5.5547 from CQI table 0.92578125
29 Implicit TBS signalling with QPSK
30 Implicit TBS signalling with 16QAM
31 Implicit TBS signalling with 64QAM
Table 4-7 CQIs to MCS mapping
4.6 Services
To account for the different services offered to the subscriber, you can set up your own services and
then allocate the services to terminal types. For example, services might have different costs, data rates,
and other requirements such as quality of service (QoS). Some of these factors are determined by the
bearers that you assign to a service. The parameters that you specify will influence how the simulation
(Chapter 5) behaves and will enable you to examine coverage and service quality for individual types
of services.
The standard LTE services correspond to QoS Class Identifier (QCI) values of 1 to 9 and are available
in ASSET by default. The following figure presents the ASSET GUI for the definition of LTE
Services.
AIRCOM International ASSETV8.0- LTE Application Notes Page 30 of 64
Figure 4-10 Definition of LTE Services in ASSET
QoS differentiation, i.e. prioritisation of different services according to their requirements becomes
extremely important when the system load gets higher. The most relevant parameters of QoS classes
are
Transfer Delay: This represents how delay sensitive the traffic is. For example, the VoIP
class is meant for very delay sensitive traffic while the P2P File Sharing class is delay
insensitive.
Guaranteed Bit rate: Delay sensitive QoS Classes have guaranteed bit rate requirements.
This defines the minimum bearer bit rate that the E-UTRAN must provide and it can be used
in admission control and in resource allocation. Each guaranteed bit rate service also has a
maximum bit rate demand, i.e. it can't exceed this limit.
Allocation and Retention Priority (ARP): Within each QoS class there are different
allocation and retention priorities. The primary purpose of ARP is to decide whether a bearer
establishment / modification request can be accepted or needs to be rejected in case of
resource limitations (typically available radio capacity in case of GBR bearers). In addition,
the ARP can be used (e.g. by the eNodeB) to decide which bearer(s) to drop during
exceptional resource limitations (e.g. at handover).
It is important to remember that pure prioritisation in packet scheduling alone is not enough to provide
full QoS differentiation gains. Users within the same QoS class and ARP class will share the available
capacity. If the number of users is simply too high, then they will suffer from bad quality. In that case it
is better to block a few users to guarantee the quality of existing connections, like streaming videos.
The radio network can estimate the available radio capacity and block an incoming user if there is no
room to provide the required bandwidth without sacrificing the quality of existing connections.
Table 4-8 presents the standard LTE Services per QCI. Gaming, VoIP, Signalling and Web Browsing
are treated as the most delay sensitive Classes while Streaming, E-mail, P2P File Sharing and Chat are
not that delay sensitive. Highest Priority in terms of ARP is given to Signalling followed by VoIP and
lowest to Chat. Packet Error Loss Rate (PELR) requirements vary with values starting from 10
-2
down
to 10
-6
. The loosest PELR requirements hold for VoIP at 10
-2
.
AIRCOM International ASSETV8.0- LTE Application Notes Page 31 of 64
Name QCI
Resource
Type
Priority
Packet
Delay
Budget
Packet Error Loss
Rate
Example Services
VoIP
QCI-1
1 2 100 ms 10
-2
Conversational Voice
Video Call
QCI-2
2
GBR
4 150 ms 10
-3
Conversational Video (Live
Streaming)
Gaming
QCI-3
3 3 50 ms 10
-3
Real Time Gaming
Streaming
QCI-4
4 5 300 ms 10
-6
Non-Conversational Video
(Buffered Streaming)
Signalling
QCI-5
5 1 100 ms 10
-6
IMS Signalling
E-mail
QCI-6
6
6
300 ms
10
-6
Video (Buffered Streaming)
TCP-based (e.g., www, e-mail,
chat, ftp, p2p file sharing,
progressive video, etc.)
Web browsing
QCI-7
7 Non-GBR
7
100 ms
10
-3
Voice,
Video (Live Streaming)
Interactive Gaming
P2P File
Sharing
QCI-8
8
8
300 ms
10
-6
Video (Buffered Streaming)
TCP-based (e.g., www, e-mail,
chat, ftp, p2p file
Chat
QCI-9
9 9 sharing, progressive video, etc.)
Table 4-8 Definition of Default LTE Services
After defining the General Service Parameters one or more Carriers can be related to the Service. Since
a supporting Carrier has been assigned to the Service, all UL and DL Bearers will be available for
selection as the Supporting Bearers. For example in Figure 4-11 all DL Bearers have been assigned to
VoIP Service. A Minimum Bit Rate (Min-GBR) and a Maximum Bit Rate (Max-MBR) have been
specified for the service. If a terminal achieves connection to one or more of the available bearers then
the eNodeB will firstly allocate enough resources to it in order to achieve the Min-GBR. It will keep
allocating more resources to it until the terminal either reaches the Max-MBR ceiling or until there not
more resources available due to cell loading. When many services are competing to get assigned to
resources from the same eNodeB then the services priorities and the eNodeBs scheduling algorithm
(Round Robin, Proportional Fair, Proportional Demand or Max SINR) will determine the proportion of
resources to be allocated to each one of them. The most preferable bearer is DL-CQI-15 and the least
preferable bearer is DL-CQI-1. ASSET sorts the bearers automatically in descending Throughput or
Data Rate.
Figure 4-11 LTE Service, Example of Supporting Bearers
AIRCOM International ASSETV8.0- LTE Application Notes Page 32 of 64
4.7 eNodeB and Cell parameters
The LTE Radio Access Network consists of eNodeBs. In ASSET eNodeBs can be defined on the GIS
or imported from a spreadsheet using the xml editor. Cell and eNodeB templates may be used to speed
up this procedure. The configuration values for the eNodeBs should normally be taken from vendor
specifications. As a sort check list the following should be set accordingly:
Antennas with propagation models
Carriers, ICIC schemes, schedulers defined and applied on per cell basis
Advanced Antenna Systems (AAS) Settings
Transmit Power and Power Channel Offsets
Figure 4-12 Site Database
4.8 LTE Planners
4.8.1 Physical Cell ID Planner
In LTE, the Primary and Secondary Synchronisation (P-SCH and S-SCH) signals are employed for
initial cell search and detection of Physical Cell Identities (Physical Cell IDs). The LTE Physical Cell
ID Planner in ASSET is designed to assign these Physical Cell IDs automatically to each sector with a
sophisticated (fixed\automatic) reuse distance algorithm, using multiple filters and schemas and
Neighbour relations.
Figure 4-13 PCI Planner
AIRCOM International ASSETV8.0- LTE Application Notes Page 33 of 64
4.8.2 LTE Frequency Planner
ASSET incorporates an Automatic Frequency Planner (AFP) for the optimum assignment of carriers
(Channel Bandwidths) to LTE sectors. In addition to a simple distance based algorithm, AFP uses the
Interference matrix for minimizing inter-cell interference. Carrier assignments and conflicts can be
visualised and further analysed in an enhanced reporting engine to determine the quality of produced
frequency plans.
Figure 4-14 Frequency Planner
For a detailed description of the LTE planners functionality please refer to ASSET User Reference
Guide.
4.9 Terminal Types
The following figure presents the ASSET GUI for the definition of LTE Terminal Types.
Figure 4-15 Definition of LTE Terminal Types in ASSET
In ASSET, terminal types represent the different types of mobile devices in your network, and their
distribution. In a modern cellular network, subscribers can have different types of terminals with
different characteristics. In ASSET, you can define a variety of terminal types to represent current or
projected distribution profiles of the subscribers in your network. You can associate these terminal
types with specific or multiple services. Importantly, you can then determine how the traffic will be
spread for each service, according to specified distributions in relation to the mapping data.
In summary, a terminal type defines the following key characteristics that will in their turn determine
the accuracy of the Simulations:
AIRCOM International ASSETV8.0- LTE Application Notes Page 34 of 64
How much traffic will the terminal type generate in total?
How will the traffic be spread geographically?
What is the expected mobile speed distribution for this terminal type?
With which service will the terminal type be associated?
What are the mobile equipment characteristics?
The purpose of master and slave terminal types is so that you can specify geographical distribution
settings on a 'master' terminal type, and then experiment with various scaling factors when you spread
traffic. If you want to do this, you can define separate 'slave' terminal types with appropriate scaling
percentages. Each slave must always be associated with a 'master' terminal type.
4.9.1 Creating a Traffic Raster
This is usually done per clutter type by assigning a terminal density or a relative weight to each one of
the clutters. It is also possible to spread traffic on user defined points, polygons and inside polygons.
The percentage of in-building traffic per clutter type can also be specified.
For in-building terminals a different fading standard deviation, indoor loss and angular spread will be
applied as defined in LTE Clutter Parameters. It is also possible to define the terminals Mean Speed,
Speeds Standard deviation, Minimum and Maximum Speed per clutter type.
To complete the traffic modelling the Traffic Wizard is run to spread the actual terminals in the area
under examination. The Resolution Option for the outcome array should be in alignment with the
lowest resolution of the propagation models in use.
There is an option to Restrict Traffic to Coverage that ensures that traffic will be spread only in areas
where there is coverage. This option should not be used if only initial estimates of the site locations,
equipment and configuration needed for a new or expanding network are required.
The option to Restrict Traffic to Coverage should be used when the 3G available coverage is required
to match that of the LTE network under planning. Having already a set of 2G and 3G sites/cells and
assuming that they have been finely optimised over years to cover and serve all wanted areas and
traffic it may be required that LTE coverage matches the coverage provided by the old technologies. In
this case the coverage predictions for the 2G and 3G cells should be created and the LTE traffic should
be restricted to the contour created by them.
The actual number of terminals to be served should come from the operators OSS statistics and traffic
forecasts also taking into account the churn rates and the 2G/3G to LTE customer conversion predicted
rates.
Figure 4-16 Example of Traffic Raster or Geographical Traffic Distribution
AIRCOM International ASSETV8.0- LTE Application Notes Page 35 of 64
5 LTE Network Performance- Coverage and Capacity Predictions
LTE network performance using the Monte-Carlo simulator can be performed in two different manners
with regards to how the cell load levels are specified. In the first case they are specified in the Site
Database and specifically under the LTE Parameters tab in the fields of Downlink Load (as a
percentage) and Mean UL Interference Level (in dB). The second option is to create a traffic raster
spreading the defined LTE Terminal Type(s) and then the cell load levels get calculated by running
Simulator Snapshots. In both cases a reference terminal type has to be specified for the calculation
process.
Figure 5-1 LTE Simulator Wizard
Figure 5-2 Selection of Filters and Cells
The decision on what resolution should be used for the simulations is based on what propagation
models are assigned to the cell antennas.
Firstly, it is suggested to use a propagation model at the resolution it has been tuned for.
Secondly, it is suggested to use two propagation models.
o The first one (Primary) should be calculated at high resolution (2-20 meters) and for
a relatively small radius (1-3 km).
o The second one (Secondary) should be calculated at relatively lower resolution (20-
100 meters) and for a larger radius (3-30km).
This setup will provide high accuracy at the expected serving area of the cell which usually doesnt
span farther than 3km (for urban type of environment). It will also provide good enough accuracy for
the calculation of interference caused by the bespoken cell far away from it but and at the same time it
will be computationally effective (relatively fast to calculate as the resolution is low). The number of
covering cells mainly affects the accuracy of the interference based calculations. The more cells taken
into account, the more accurate the interference values are. A typical value would be 6 to 10. The
typical values for Fading Correlation Coefficients are 0.8 for Intra-Site antennas and 0.5 for Inter-Site
antennas.
Figure 5-3 Selection of Terminal Type(s)
Figure 5-4 Line of Sight Settings
AIRCOM International ASSETV8.0- LTE Application Notes Page 36 of 64
It is also important to consider cells with prediction area within the specified region (2D View) as those
cells will possibly pick up some of the traffic in the specified area and also increase interference.
The Line of Sight Settings allow deactivating certain MIMO schemes based on LOS information.
MIMO schemes rely on a low correlation between the signal paths to the transmit elements of an
antenna; locations that have LOS to an antenna are more likely to have a high correlation, therefore
MIMO gains should not be considered for such locations.
Following, a comprehensive presentation and discussion of the ASSET LTE module is presented. It
will focus on four main areas, namely basic coverage (RSRP, RSSI and RSRQ), MIMO schemes, Inter-
Cell Interference Coordination (ICIC) schemes and Schedulers. An urban area was chosen expanding
6.4 by 4.8 km and covered by 78 eNodeBs bearing 224 cells.
AIRCOM International ASSETV8.0- LTE Application Notes Page 37 of 64
5.1 Basic Coverage (RSRP, RSSI, RSRQ)
RSRP, RSSI and RSRQ are defined in detail chapter 3.8. An informal definition of these quantities is
given hereby:
Received Signal Reference Power (RSRP) is the indicator of the signal strength coming
from the serving cell experienced by a UE at a certain point and time.
Received Signal Strength Indicator (RSSI) is the indicator of how much power is received
by the UE over its operating bandwidth. The sources of this power include co-channel serving
and non-serving cells, adjacent channel interference, thermal noise and so on.
Reference Signal Received Quality (RSRQ) is the indicator of the quality of the signal. In
this context, quality is expressing how stronger the signal is compared to noise and
interference. It is thereafter proportional to RSRP and diversely proportional to RSSI, however
it is not a ratio of RSRP over RSSI.
From a terminal point of view a pixel is covered if the required RSRP, RSRQ and BCH/SCH SINR are
met.
Figure 5-5 Trial Area
Figure 5-6 LTE Terminal Type
Well examine how cell load levels affect the satisfaction of these requirements. Cell Load Setting live
on the Cell Params. To start with, SU-MIMO as well as MU-MIMO was disabled on both uplink and
downlink at the Cell Params as well as at the Terminal Type.
Figure 5-7 Site DB Settings - Cell Load Levels
AIRCOM International ASSETV8.0- LTE Application Notes Page 38 of 64
RSRP (Figure 5-8) is not affected by cell loads. This is the reason why a network is usually firstly
dimensioned to provide adequate signal strength at the desired areas.
Figure 5-8 RSRP
RSRQ (Figure 5-9) on the other hand is affected by cell loads. Cell loads in essence express how many
users are connected to the network. The more active users there exist, the more resources are consumed
in the DL and UL interference level rises. This explains why cells service areas shrink as the number
of users increase. The following figure illustrates an unloaded network and the one after this a heavily
loaded network.
Figure 5-9 RSRQ changes depending on Cell Loads
Unloaded Network Loaded Network
AIRCOM International ASSETV8.0- LTE Application Notes Page 39 of 64
Finally, the third requirement for BCH/SCH SINR (Figure 5-10) is not affected by the cell load.
Although BCH/SCH SINR includes the term I for interference it is not affected by load as the BCH and
SCH channels are positioned in the 6 central RBs of the BandWidth and dont get interference from
RBs carrying traffic.
Figure 5-10 DL BCH/SCH SINR
The change in RSRQ (Figure 5-9) coming from the change of the cell load levels is causing changes in
the allocated bearers (Figure 5-11), white areas not covered at all. Grey, red and yellow areas covered
by the higher CQI bearers. Performance deteriorates as the number of users increase. It is obvious that
the area covered by high CQI bearers (or high MCS bearers) that provide high throughput decreases as
cell load rises. The mitigation of this effect is one of main objectives of MIMO schemes, ICIC schemes
and Schedulers.
Figure 5-11 Achievable DL Bearer
5.2 MIMO Schemes
The SU-MIMO and MU-MIMO modes need to be enabled on the Cell, Bearer and Terminal Type. To
demonstrate the use of MIMO we will assess the network without MIMO, then with Diversity Only,
Spatial Multiplexing (SM) Only, Adaptive Switching (Diversity-SM) and finally MU-MIMO.
Unloaded Network Loaded Network
AIRCOM International ASSETV8.0- LTE Application Notes Page 40 of 64
Roughly speaking Diversity is used to improve coverage, SM is used to increase single users
throughput and MU-MIMO is used to improve the network throughput or to serve more simultaneous
users. We will elaborate on each of schemes in more detail.
5.2.1 SU-MIMO Diversity
As shown in the LTE AAS Parameters look-up table (Figure 5-12) the effect of having 1 TX antenna at
the eNodeB and 2 RX antennae at the UE is equivalent to having 2 TX antennae on the eNodeB and
one RX antenna on the UE. We will take the simple example of 2 RX elements and 1 TX. Note that we
are now examining downlink and the number of RX elements is defined on the Terminal Type.
Figure 5-12 LTE AAS Parameters - Diversity
When applying diversity the RSRP plot and the SCH/BSC SINR plot stay the same. RSRQ stays the
same as well. What changes, are the SINR requirements for the bearers that are divided by the
corresponding table value. The SINR Adjustment can be refined per clutter type using the LTE Cletter
Parameters.
Figure 5-13 LTE Clutter Parameters Diversity
As previously mentioned Diversitys main purpose is to increase coverage and this is done by
decreasing the bearers SINR requirements. The bearers with the decreased SINR requirements are
easier to achieve.
By increasing the coverage for each bearer respectively the result will be larger areas with higher CQI
bearers. So from a system perspective Diversity not only increases coverage but network throughput as
well. Figure 5-14 and Figure 5-15 depict the change of coverage that results in higher system
throughput.
AIRCOM International ASSETV8.0- LTE Application Notes Page 41 of 64
Figure 5-14 Achievable DL Bearer without and with Diversity Coverage Improvement
Figure 5-15 DL Data Rate Improvement with Diversity (2TX by 1 RX)
5.2.2 SU-MIMO Spatial Multiplexing
Spatial Multiplexing (SM) targets increasing users throughput. Depending on the number of TX and
RX antennae the user experiences a Rate Gain as shown below:
Figure 5-16 LTE ASS Parameters - SM
The SM Rate Gain can be refined per clutter type using the LTE Clutter Parameters if required.
50% Loaded Network without MIMO
50% Loaded Network with Diversity
50% Loaded Network without MIMO 50% Loaded Network with Diversity
AIRCOM International ASSETV8.0- LTE Application Notes Page 42 of 64
Figure 5-17 LTE Clutter Parameters - SM.
Spatial Multiplexing does increase throughput but this comes at an expense of higher SINR
requirements as shown on the LTE bearers
Figure 5-18 MIMO SINR Delta for Spatial Multiplexing
We will examine how the Achievable bearers plot and data rates change with the application of SM. As
shown in Figure 5-16 a combination of 1TX by 2 RX provides no improvements so we will go for 2TX
by 2RX elements that should roughly provide a doubling in data rates subject to clutter parameters. In
general, the Rate Gain is equal to the minimum of TX and RX elements and is also affected by the
clutter specific parameters (Figure 5-17).
As shown in Figure 5-19 and Figure 5-20 coverage slightly shrinks with SM because of the higher
SINR requirements but at the same time data rates almost double. Take for example eNodeB 032451 at
South-West of the map. Areas providing 3-4Mbps data rate without MIMO jump up to 6-7Mbps with
Spatial Multiplexing.
Figure 5-19 Achievable DL Bearer without and with SM Minor Coverage Change
50% Loaded Network without MIMO
50% Loaded Network with SM (2TX by 2RX)
AIRCOM International ASSETV8.0- LTE Application Notes Page 43 of 64
Figure 5-20 DL Data Rate Improvement with Spatial Multiplexing
5.2.3 SU-MIMO Adaptive Switching
As shown in the previous two subchapters Diversity and Spatial Multiplexing provide significant gains
to the network. Both of them can be deployed at the same time in Adaptive Switching mode by
eNodeBs so as to provide higher throughput to users close to the cell and extended coverage to users
further away at the cell edge.
Observing how the Achievable DL bearer changes with Adaptive Switching (Figure 5-21) there is an
obvious difference compared to operating the network without SU-MIMO. Take for example eNodeB
032451 at South-West of the map. Starting from the vicinity of the cell and moving further away the
following bearers are deployed DL-CQI-15, DL-CQI-14, DL-CQI-13, DL-CQI-12, DL-CQI-15, DL-
CQI-14, DL-CQI-13. The switch from DL-CQI-12 to DL-CQI-15 is the point where Diversity takes
over Spatial Multiplexing.
Figure 5-21 Achievable DL Bearer without and with SU-MIMO (2TX by 2RX)
The DL Transmission array (Figure 5-22) is more indicative of the type of SU-MIMO that is selected
for every pixel. The aim is always to provide the higher possible data rate at every pixel. This aim
governs the decision of whether to choose SM or diversity and which specific bearer to deploy for
every pixel.
50% Loaded Network without MIMO 50% Loaded Network with SM (2TX by 2RX)
50% Loaded Network without MIMO 50% Loaded Network with SU-MIMO (Diversity
and Spatial Multiplexing in Adaptive Switching md)
AIRCOM International ASSETV8.0- LTE Application Notes Page 44 of 64
Figure 5-22 Transmission Modes with adaptive switching (no user defined thresholds)
Examining the DL Data Rate (Figure 5-23) and comparing against the cases when either Diversity or
SM is deployed we can convey that adaptive switching is, as expected, combining the gains from both
types of MIMO by deploying the most appropriate one for every pixel.
Figure 5-23 DL Data Rate with Adaptive Switching (no user defined thresholds)
No MIMO Diversity
Spatial Multiplexing
Adaptive Switching (Diversity - SM)
AIRCOM International ASSETV8.0- LTE Application Notes Page 45 of 64
The user can set specific thresholds per cell to govern the switch between Diversity and SM depending
on DL Reference Signal SNR. In this case the Transmission-Modes-plot changes as shown in Figure
5-25 and the achievable bearer plot changes similarly as indicated in Figure 5-26.
Figure 5-24 User Defined Threshold for Adaptive Switching
Figure 5-25 Transmission Modes with adaptive switching (with user defined thresholds)
Figure 5-26 Achievable DL Bearer without and with SU-MIMO with user defined thresholds
With User Def. Thresholds Automatic Adaptive Mode
With User Def. Thresholds Automatic Adaptive Mode
AIRCOM International ASSETV8.0- LTE Application Notes Page 46 of 64
Figure 5-27 is the comparing all different options for SU-MIMO and how they affect Data Rates.
Figure 5-27 DL Data Rate with Adaptive Switching
No MIMO Diversity
Spatial Multiplexing Adaptive Switching (Diversity - SM)
Adaptive Switching (Diversity - SM) user defined Thresholds
AIRCOM International ASSETV8.0- LTE Application Notes Page 47 of 64
5.2.4 MU-MIMO
MU-MIMO is used to increase the cells throughput. This is achieved by co-scheduling terminals on
the same Resource Blocks. Applying MU-MIMO will make no obvious changes to a network unless it
is overloaded. To demonstrate the use of MU-MIMO we will spread terminals and run the SIM in
snapshot mode. The density of terminals will be high enough for many of them to fail due to
insufficient capacity. Then we will enable MU-MIMO and observe how the network is now capable to
serve more of the terminals that were previously dropped because of the eNodeBs resources maxing
out.
In order for MU-MIMO to be used there is a higher Traffic & Control SINR requirement defined
.
Figure 5-28 Bearers - MU-MIMO SINR Delta
Figure 5-29 Cell AAS Settings - MU-MIMO Average Co-scheduled Terminals
RSRQ changes when MU-MIMO is deployed because the number of served terminals changes.
In Figure 5-30 we can observe how multiuser MIMO Improves the Data Rates.
Figure 5-30 DL Data Rate without and with MU-MIMO
Figure 5-31 depicts the actual difference of DataRate
with MU-MIMO
- DataRate
without MU-MIMO
. We can
observe that when MU-MIMO is deployed everywhere, it provides small improvements close to the
cell, large improvements close to the cell edge and mediocre improvements at the cell edge.
Without MIMO
MU-MIMO deployed
AIRCOM International ASSETV8.0- LTE Application Notes Page 48 of 64
Figure 5-31 Improvement in DL Rate with MU-MIMO deployed
The following figure demonstrates how total DL Cell Throughout (per cell) increases when MU-
MIMO is enabled. This is an effect of the eNodeB now being capable to serve a higher number of users
by scheduling them on the same resources. These users would be otherwise failing to connect.
Figure 5-32 DL Cell Throughput without and with MU-MIMO
The following table indicates how a highly loaded network can accommodate extra users by deploying
MU-MIMO.
Cell Composite Report On Service hidata
Without MIMO With MU-MIMO
Mean Attempted 4930.286 4889.714
Mean Served 2743.857 55.65% 3484.929 71.27%
Mean Failed 2186.429 44.35% 1404.786 28.73%
Contributions to Failure
without MIMO
MU-MIMO deployed
AIRCOM International ASSETV8.0- LTE Application Notes Page 49 of 64
DL RSRP 0.37% 0.47%
RSRQ 0.16% 0.15%
DL BCH/SCH SINR 0.00% 0.00%
UL SINR 0.00% 0.00%
DL SINR 1.88% 1.89%
UL Capacity 0.00% 0.00%
DL Capacity 97.62% 97.51%
No valid connection scenarios 0.00% 0.00%
No covering cells 0.00% 0.00%
Note: Terminals can fail to connect for multiple reasons so the failure reason
percentages can sum to more than 100%.
Table 5-1 Composite failure report with and without MU-MIMO.
Doing some further analysis to these results we observe that the MU-MIMO provides great gains to
heavily loaded cells and relatively smaller gains to lightly loaded cells. Moreover, from Figure 5-31 we
observe that this improvement in is mainly coming from the cell edge rather than the cell centre.
Figure 5-33 Comparative Cell Throughput Improvement with MU-MIMO versus Cell Load
5.2.5 SU-MIMO and MU-MIMO
We will now enable SU-MIMO (Diversity and Spatial Multiplexing) and MU-MIMO at the same time.
Our strategy will be to use SM close to the eNodeBs to increase data rates, Diversity further away from
the eNodeBs to increase coverage and MU-MIMO for heavily loaded cells to reduce the number of
failures due to capacity.
AIRCOM International ASSETV8.0- LTE Application Notes Page 50 of 64
Figure 5-34 DL Transmission Mode with SU-MIMO and MU-MIMO
Figure 5-35 DL Data Rate with SU-MIMO and
MU-MIMO
Figure 5-36 Achievable DL Bearer with SU-
MIMO and MU-MIMO
Figure 5-37 DL Transmission Mode with SU-MIMO and MU-MIMO
Unloaded Network with MIMO Loaded Network with MIMO
AIRCOM International ASSETV8.0- LTE Application Notes Page 51 of 64
In this chapter the advantage of SU-MIMO and MU-MIMO over single antenna transmission were
demonstrated. For the trial area examined, the deployment of MU-MIMO increased the number of
served terminals by 15% and the combined deployment of SU-MIMO and MU-MIMO increased the
number of served terminals by 33%.
Cell Composite Report On Service
Without MIMO With MU-MIMO
With SU-MIMO and
MU-MIMO
Mean Attempted
4930
4889
4943
Mean Served
2743 55.6% 3484 71.27% 4350 88.01%
Mean Failed
2186 44.3% 1404 28.73% 592 11.99%
Contributions to Failure
DL RSRP
0.37%
0.47%
1.20%
RSRQ
0.16%
0.15%
0.11%
DL BCH/SCH SINR
0.00%
0.00%
0.00%
UL SINR
0.00%
0.00%
0.00%
DL SINR
1.88%
1.89%
1.64%
UL Capacity
0.00%
0.00%
0.00%
DL Capacity
97.62%
97.51%
97.08%
No valid connection scenarios
0.00%
0.00%
0.00%
No covering cells
0.00%
0.00%
0.00%
Note: Terminals can fail to connect for multiple reasons so the failure reason percentages can sum
to more than 100%.
Table 5-2 Composite failure report with SU and MU-MIMO.
AIRCOM International ASSETV8.0- LTE Application Notes Page 52 of 64
5.3 ICIC
Inter-Cell Interference Coordination (ICIC) schemes aim to maximize spectral efficiency of LTE
systems by re-using the available resource blocks (RBs) as often and in as many cells as possible while
keeping the overall ICI in the system to an acceptable level. There are various schemes that are
designed to mitigate ICI, and their implementation in the live network scenario is largely governed by
the equipment vendors. The following ICIC schemes are supported in ASSET:
Reuse 1 (Prioritisation)
Soft Frequency Reuse
Reuse Partitioning
They are defined on the carrier layer (Figure 5-38). Fundamental to each of these methods is a division
of the network into two areas in relation to the cell coverage, i.e. Cell Centre Users (CCUs) and Cell
Edge Users (CEUs). This spatial separation of cell service area is controlled in ASSET by the Cell
Edge Thresholds defined per cell in the Site Database. The available thresholds are RSRP and
Relative RSRP. RSRP is self-explanatory while the latter is defined in dBs and can be expressed as
the difference between the RSRPs of the serving and the strongest interfering cell.
Figure 5-38 ICIC Schemes
Figure 5-39 Cell Edge Thresholds
For this Analysis we will use the Relative RSRP option (Figure 5-40) as it captures better the Cell Edge
compared to RSRP threshold that captures the Site Edge. The result of the two Cell Edge Threshold
option is shown in the following figure.
Figure 5-40 Cell Centre / Cell Edge
Based on Relative RSRP Based on RSRP
AIRCOM International ASSETV8.0- LTE Application Notes Page 53 of 64
5.3.1 Reuse 1 (Prioritisation)
Reuse-1 is implemented by splitting the total bandwidth in partitions and prioritising their use to
different cells in a coordinated format. ICIC schemes target the improvement of Traffic and Control
SINR. This will affect positively the achievable bearers and subsequently the number of served
terminals and network throughput.
Figure 5-41 Traffic & Control SINR without and with ICIC (Reuse-1, Prioritisation)
The main factor improved by ICIC is Traffic & Control SINR which is shown in Figure 5-41. This
improvement allows for a more favourable allocation of bearers (Figure 5-42) and consequently higher
data rates (Figure 5-43).
Figure 5-42 Achievable DL bearer without and with ICIC (Reuse-1, Prioritisation)
Without ICIC With Reuse 1 (Prioritisation)
Without ICIC
With Reuse 1 (Prioritisation)
AIRCOM International ASSETV8.0- LTE Application Notes Page 54 of 64
Figure 5-43 DL Data Rate without and with ICIC (Reuse-1, Prioritisation)
The improvement of Traffic & Control SINR with the deployment of Prioritisation is dependent on the
Cell Loading and on the Coordination factor (Figure 5-44). The Coordination Factor takes into
account that the modelling is probabilistic and scales the maximum theoretical gains accordingly. A
coordination factor of 1 means perfect coordination and a coordination factor of 0 assumes no
coordination at all. For the purpose of this trial a value of 0.7 is used.
Figure 5-44 Average Traffic&Control SINR Improvement with Prioritisation as a function of Cellss
Load Level and the Coordination Factor (CF)
Without ICIC With Reuse 1 (Prioritisation)
AIRCOM International ASSETV8.0- LTE Application Notes Page 55 of 64
As shown in Figure 5-44, irrespective of the Coordination Factor, if a Network is heavily loaded (over
70%) then gains are negligible.
The gains of Prioritisation diminish as if the network load is high. To explain this, we will take the
example of a 3sectored network. If the cell load is more than 1/3 or 33% then the prioritised recourses
for each one of the cells is not enough to serve the load and overlapping (thus interference) is
unavoidable. For an 80% loaded network the use of prioritisation doesnt provide any gains as shown
in Figure 5-45.
Figure 5-45 Highly Loaded (80%) Network without and with Prioritisation Same Performance
DL Traffic/Ctrl SINR Without ICIC DL Traffic/Ctrl SINR With Reuse-1
(Prioritisation)
DL Data Rate With Reuse-1 (Prioritisation) DL Data Rate Without ICIC
AIRCOM International ASSETV8.0- LTE Application Notes Page 56 of 64
5.3.2 Soft Frequency Reuse and Reuse Partitioning
Soft Frequency Reuse
This scheme is an extension of the Reuse-1 (Prioritisation) scheme, where in addition to prioritising
RBs in each cell, a power difference in the DL between cell centre users (CCUs) and cell edge users
(CEUs) is also introduced. This difference in power between RBs effectively divides the cell into an
inner and an outer region, and the users located in these regions can be classified as the CCUs and
CEUs, respectively. Hence, by prioritising and using more power on CEUs' RBs, overall ICI mitigation
in the network and performance gain in terms of coverage and cell-edge capacity can be realized.
To define a Soft Frequency Reuse Scheme:
Split the carrier bandwidth (RBs) into two dedicated portions or zones, i.e. CC zone and CE
zone, one for CCUs and the other for CEUs, using the Soft Bandwidth Ratio parameter.
Specify a Coordination Factor. The Reuse-1 (Prioritisation) scheme (as described in the
previous section) is implemented within both zones; each sector has a prioritised partition in
each zone, which it tries to use before the non-prioritised partitions in that zone. The
Coordination Factor can be used to adjust the calculated probability of collision.
Specify a power difference between the DL RBs of the CEUs and CCUs, using the Power
Ratio parameter. The concept is that the CCUs can be reached with the reduced power
whereas the CEUs need the higher power level for successful transmission. This results in a
possible performance gain for the CEUs.
This picture shows an example of the Soft Frequency Reuse scheme:
Figure 5-46 Soft Frequency Reuse Scheme (Power Ratio 50%, Bandwidth Ratio 50%)
Figure 5-47 Soft Frequency Reuse Scheme Example Settings
AIRCOM International ASSETV8.0- LTE Application Notes Page 57 of 64
Reuse Partitioning
In addition to the prioritisation of RBs and different DL power levels, this scheme also divides the
available spectrum into multiple partitions. Reuse Partitioning is similar to Soft Frequency Reuse,
because it divides the available carrier bandwidth (RBs) into two dedicated zones, one for CCUs, the
other for CEUs. As in Soft Frequency Reuse, the CC zone uses Reuse 1 (Prioritisation). However,
unlike Soft Frequency Reuse, the CE zone does not use Reuse 1 (Prioritisation), but instead employs
the traditional frequency reuse of N, where N is the number of sectors on the eNodeB. Each sector can
only consume CE resources from its own dedicated CE partition. Restricting each sector to its own
dedicated CE partition results in power concentration for the CE partition, which means that the
spectral density of the power transmitted over a fraction of the CE RBs is higher than the spectral
density of the same power transmitted over the entire RBs.
Figure 5-48 Reuse Partitioning
Figure 5-49 ReUse Partitioning Example Settings
We will now compare Soft Frequency Reuse and Reuse Partitioning against no ICIC at all. The first
factor to examine is DL Traffic and Control SINR (Figure 5-50). In Figure 5-51 we can see how the
Achievable bearer is improved with the deployment of the ICIC schemes and finally in Figure 5-52 we
can observe the obvious improvement in DL Data Rate.
AIRCOM International ASSETV8.0- LTE Application Notes Page 58 of 64
Figure 5-50 DL Traffic & Control SINR without and with ICIC
No ICIC
Soft Frequency Reuse
Reuse Partitioning
AIRCOM International ASSETV8.0- LTE Application Notes Page 59 of 64
Figure 5-51 Achievable DL bearer without and with ICIC
No ICIC
Soft Frequency Reuse
Reuse Partitioning
AIRCOM International ASSETV8.0- LTE Application Notes Page 60 of 64
Figure 5-52 DL Data Rate without and with ICIC
No ICIC
Soft Frequency Reuse
Reuse Partitioning
AIRCOM International ASSETV8.0- LTE Application Notes Page 61 of 64
Taking a closer look at the results, we have created the following graphs to prove the superiority of the
aforementioned ICIC schemes:
Figure 5-53 Traffic & Control SINR without and with ICIC
The above figure demonstrates how the ICIC schemes improve the SINR Cumulative distribution.
Figure 5-54 DL Data Rates without and with ICIC
The above figure comprehensively demonstrates the effect of ICIC schemes on the ultimate operators
goal; to increase the Data Rates.
AIRCOM International ASSETV8.0- LTE Application Notes Page 62 of 64
Figure 5-55 DL Data Rates without and with ICIC Cumulative Distribution
Without any ICIC scheme only 67% of the covered area can support speeds up to 40Mbps. The
deployment of Soft Frequency Reuse increases this figure by 5%, up to 72% and the deployment of
Reuse Partitioning improves it by 24%, up to 91%.
5.4 Schedulers
LTE services consist of two traffic types Real Time and Non-Real Time. Real Time services have an
associated Maximum Bit Rate (MBR) demand in addition to the (minimum) Guaranteed Bit Rate
(GBR), whereas Non-Real Time services have only a GBR demand.
When running a simulation, ASSET first attempts to serve the GBR demands of both Real Time and
Non-Real Time services, taking into account the Priority values of the different services. Resources are
first allocated to the service with the highest priority, and then to the next highest priority service, and
so on. Terminals are only served if there are enough resources available to satisfy their GBR demand.
In the event that there are not enough resources to fulfil the GBR demand of all Real Time and Non-
Real Time services, then only the Priority values of the services determine the precedence of resource
allocation.
If resources are still available after the GBR demands have been met, then different scheduling
algorithms can be employed to attempt to serve the MBR of Real Time services. To satisfy the
Maximum bit rate of Real Time services (RT-MBR), there are four different scheduling algorithms
available:
On the General LTE Parameters Page, at the Cell layer, there exists an option of four different
schedulers to select from:
Max SINR
Proportional Demand
Proportional Fair
Round Robin
AIRCOM International ASSETV8.0- LTE Application Notes Page 63 of 64
Figure 5-56 Schedulers LTE PArams
Lets draw a simple example.
We have two services only:
Real Time Service : RT_S1, GBR=100kbps, MBR=300kbps
Non Real Time Service : NonRT_S1, GBR=140kbps.
RT_S1
NonRT_S1
Now lets assume a cell is serving five RT_S1 users and four NonRT_S1 users with that much RBs
available:
Figure 5-57 Schedulers LTE PArams
Round Robin
The aim of this scheduler is to share the available/unused resources equally among the RT terminals
(i.e. the terminals requesting RT services) in order to satisfy their RT-MBR demand. This is a recursive
algorithm and continues to share resources equally among RT terminals, until all RT-MBR demands
have been met or there are no more resources left to allocate.
Proportional Fair
The aim of this Scheduler is to allocate the available/unused resources as fairly as possible in such a
way that, on average, each terminal gets the highest possible throughput achievable under the channel
conditions.
This is a recursive algorithm. The remaining resources are shared between the RT terminals in
proportion to their bearer data rates. Terminals with higher data rates get a larger share of the available
resources. Each terminal gets either the resources it needs to satisfy its RT-MBR demand, or its
weighted portion of the available/unused resources, whichever is smaller. This recursive allocation
process continues until all RT-MBR demands have been met or there are no more resources left to
allocate.
NRT_S_GBR
RT_S_MBR RT_S_GBR
Unused resources, available for
RT_S_MBR -1 -2 -3 -4 and -5
to be allocated according to
the selected Scheduler.
RT_S_GBR -2
RT_S_GBR -3
RT_S_GBR -1
NRT_S_GBR -4
NRT_S_GBR -3
NRT_S_GBR -1
NRT_S_GBR-2
RT_S_GBR -5 RT_S_GBR -4
AIRCOM International ASSETV8.0- LTE Application Notes Page 64 of 64
Proportional Demand
The aim of this scheduler is to allocate the remaining unused resources to RT terminals in proportion to
their additional resource demands. This is a non-recursive allocation process and results in either
satisfying the RT-MBR demands of all terminals or the consumption of all of the resources,.
Max SINR
The aim of this Scheduler is to maximise the terminal throughput and in turn the average cell
throughput. This is a non-recursive resource allocation process where terminals with higher bearer rates
(and consequently higher SINR) are preferred over terminals with lower bearer rates (and consequently
lower SINR). This means that resources are allocated first to those terminals with better SINR/channel
conditions, thereby maximising the throughput.
Figure 5-58 The effect of different schedulers on a fairly loaded network
Figure 5-59 The effect of schedulers on a heavily loaded network
It becomes obvious that Max SINR Scheduling will maximise the network throughput as terminals
with the best RF conditions are served first. However this is not optimising the user experience as users
with worse RF conditions are neglected. Also Max SINR approach is unfriendly to mobile users as
their SINR is often changing from good to bad and they would possible get complete interruption of
service when moving through bad RF conditions. The Round Robin approach is completely random as
it simply allocates the same resources to all terminals in turns. Proportional Fair seems as the fairest
choice because it provides resources analogous (or proportional) to the RF conditions. This is in
contradiction to Max SINR where the allocation of resources is not analogous to RF conditions but
absolute. In other words with Proportional Fair scheduling terminals experiencing bad RF conditions
will not be cut off all together but will be simply allocated fewer resources which is important for the
continuity of the service. Finally, Proportional Demand is trying to satisfy the more demanding users
by allocating more resources to them and this results in decreased overall net throughput. Proportional
Demand completely ignores RF conditions and the effect of this is to totally waste resources by trying
to serve very demanding users happening to be at very bad RF conditions.
Service
Average
Users/
Cell
GBR MBR
Average
Users
Served
per cell
Av. Cell
Load
Av. Cell
Throughput
(kbps)
Average
Users
Served
per cell
Av. Cell
Load
Av. Cell
Throughput
(kbps)
Average
Users
Served
per cell
Av. Cell
Load
Av. Cell
Throughput
(kbps)
Average
Users
Served
per cell
Av. Cell
Load
Av. Cell
Throughp
ut (kbps)
RT_S1 64 16 28 99.70% 1708 99.70% 1750 99.70% 1750 99.70% 1690
RT_S2 11 154 720 99.64% 7257 99.62% 7170 99.62% 7128 99.62% 7017
NonRT_S1 19 121 X 98.39% 2243 98.30% 2241 98.30% 2241 98.30% 2241
Sum: 11208 Sum: 11161 Sum: 11119 Sum: 10948
Round Robin
49.93%
Proportional Demand
49.93% 49.93%
Proportional Fair
49.93%
Max SINR
Service
Average
Users/
Cell
GBR MBR
Average
Users
Served
per cell
Av. Cell Load
Av. Cell
Throughpu
t (kbps)
Average
Users
Served
per cell
Av. Cell
Load
Av. Cell
Throughput
(kbps)
Average
Users
Served
per cell
Av. Cell
Load
Av. Cell
Throughpu
t (kbps)
Average
Users
Served
per cell
Av. Cell
Load
Av. Cell
Throughpu
t (kbps)
RT_S1 64 16 28 98.74% 2004 98.77% 2240 98.72% 2262 98.70% 1851
RT_S2 11 154 720 98.28% 7130 97.92% 6562 98.05% 6083 98.03% 5549
NonRT_S1 19 121 X 87.27% 3066 86.98% 3058 87.33% 3050 87.34% 3030
Sum: 12200 Sum: 11860 Sum: 11395 Sum: 10430
Max SINR Proportional Fair Round Robin
Proportional Demand
97.45% 97.67% 97.44% 97.61%