OpNET - LTE

Planning LTE Network Deployments R&D Solutions for Commercial and Defense Networks

Session 1599

CONFIDENTIAL – RESTRICTED ACCESS: This information may not be disclosed, copied, or transmitted in any format without Riverbed’s prior written consent. © 2013 Riverbed Technology, Inc. Riverbed, OPNET, OPNETWORK, and all Riverbed hardware and software product names are trademarks of Riverbed.

All other trademarks are the property of their respective owners.

CONFIDENTIAL–RESTRICTED ACCESS: This information may not be disclosed, copied, or transmitted in any format without Riverbed’s prior written consent. © 2013 Riverbed Technology, Inc. Riverbed, OPNET, OPNETWORK, and all Riverbed hardware and software product names are trademarks of Riverbed. All other trademarks are the property of their respective owners.

2

1599 Planning LTE Network Deployments

Agenda

LTE Network Architecture, Features and Capabilities

Deploying Realistic LTE Networks in OPNET Modeler • Basic Configurations and Analysis

Lab 1: Deploying and Analyzing Performance of an LTE Network

Using OPNET Modeler for Planning Studies

Cell Planning Study with OPNET Modeler • Lab 2: Cell Planning Lab

Capacity Planning Study with OPNET Modeler • Lab 3: Capacity Planning Lab

Battery Life Planning Study with OPNET Modeler • Lab 4: Optimizing DRX Parameters Within Application Delay SLAs


3


Brief Technology Introduction

Goals

• To improve the UMTS standard to cope with future technology evolutions

• User demand for higher data rates and QoS

~300 Mbps downlink, ~100 Mbps uplink

• Continued demand for cost reduction (CAPEX and OPEX)

• Low complexity

• Compatibility and inter-working with earlier 3GPP Releases

The resulting architecture is called EPS and comprises

• E-UTRAN on the radio access side

• EPC on the core side

Introduced in 3GPP specification 36-series (Releases 8 and 9)

• OFDMA in the downlink

• SC-FDMA in the uplink

Marketed as 4G

• Actually a 3.9G technology

• Does not fully comply with the IMT Advanced 4G requirements.

• LTE-Advanced (Release 10) to be 4G compliant


4


OPNET's Model Development Consortia

LTE Model Development Consortium

• Prominent network equipment manufacturers, service providers, defense organizations

• Benefits to Consortium Members

Early access to LTE model

Opportunity to influence design requirements

• Some current members include Aerospace Corporation, AT&T, DoCoMo Euro-Labs, InterDigital, NIST, Samsung, Sony, Toshiba, Hitachi and Panasonic

• Successful past consortia

WiMAX, UMTS, MANET, MPLS, and DOCSIS

Phased release schedule

Phases I, II, III, IV, V, VI and VII released so far

Future features planned


5


LTE Model Features Up to Phase VII

PHY

• FDD and TDD technologies supported

• OFDMA for downlink & SC-FDMA for uplink

• Supported channels: PDCCH, PUCCH, PHICH, PDSCH, PUSCH, PRACH

• BLER modulation curves with turbo coding and circular buffer rate matching algorithm for each modulation and coding scheme (MCS)

• Multiple path loss models

• Multipath channel model for uplink and downlink

• Interference on data channels from other data and control channels

• Intra- and inter-cell interference

• MIMO

• Transmit Diversity

• Spatial Multiplexing

HARQ

• Synchronous retransmissions with implicit grants on uplink

• Asynchronous retransmissions on downlink

• Type-II incremental redundancy

• ACK to NACK and NACK to ACK error modeling

• Accurate timing support for the TDD mode

MAC

• GBR/Non-GBR EPS bearers

• Logical and Transport Channels

• Random Access Procedure

• Frame generation and Scheduler

• Channel dependent scheduling

• CQI and rate adaptation

• Scheduling Requests

• Buffer Status Reporting

• Admission Control

• MIMO Spatial multiplexing

• DRX in RRC_Connected

• Idle mode support

Mobility and Handovers

• Initial cell selection

• Radio link monitoring

• Intra-E-UTRAN and intra-frequency handover with and without X2 support


6


* This information is provided for planning purposes only and is subject to change without notice. This does not represent a

commitment by OPNET to deliver any or all capabilities in any particular timeframe.

RLC

• Acknowledged, Unacknowledged and Transparent Modes

• Segmentation of retransmitted PDUs in case of small grants into PDU segments

• Configurable RLC parameters for each radio bearer for each direction

PDCP: Compression for TCP/IP and UDP/IP headers

EPS Mobility Management (EMM)

EPS Session Management (ESM)

• S1 Signaling and EPS Bearer Setup/Modification/Release

General

• Efficiency mode to disable PHY layer

• Tagged EPS/radio bearer related statistics

• 3 and 6 sector eNodeBs

• Router UE node

• Application Delay Tracking

• Energy consumption model

• Single-cell downlink broadcast

General

• Multimedia Broadcast Multicast Service (MBMS)

• LTE Network Deployment Wizard

• Dynamic failure/recovery of base stations

• GGSN services by EPC to legacy SGSNs

• Device Creator support

• LTE Network View

• Jammer node support for LTE

• IPv6 support

Considered New Enhancements

• Improved PUCCH Modeling

• Uplink Power Control

• Semi-persistent scheduling

LTE Model Features Up to Phase VII (cont.)


7


eNodeBs (1, 3 and 6 sectors)

Evolved Packet Core (EPC) Network with

IP/GTP Support

UE with complete TCP/IP stack

Typical Modeled Network Architecture


8


Data Traffic Flow in LTE Networks

GTP Encapsulation/Decapsulation EPS Bearer

Radio Bearer S1 Bearer

IP packets

entering the LTE

network are

mapped to GTP

tunnels

Uplink data on

radio bearer

Corresponding radio

bearer carrying the

downlink data


9


LTE eNodeB

• lte_enodeb_atm4_ethernet4_slip4

• lte_enodeb_ethernet4

• lte_enodeb_slip4

• lte_enodeb_3sector_slip4

• lte_enodeb_6sector_slip4

LTE Attribute Configuration Object

• lte_attr_definer

Simulation Model Entities

LTE UE

• lte_wkstn

• lte_server

• lte_ue_ethernet_gtwy (Router UE)

LTE EPC

• lte_epc_atm8_ethernet8_slip8


10


Agenda









11


LTE Network Deployment

Fast Deployment: Wireless Network Wizard

• Choice to select the bandwidth, number of cells, cell radius, etc.

• Most of the other settings automatically taken care of

Deploying realistic conditions

• Terrain Modeling Module

• Mobility modeling

Trajectories, random mobility, programmatic mobility

• Application/traffic modeling

Standard applications, custom applications, real application traces, application demands, IP traffic flows

Traffic must be mapped to EPS bearers to achieve QoS

• Unmapped traffic will be handled by Default bearer

• Interference modeling

Jammer nodes


12


LTE Network View

EPC ID


13


How to Configure Core Network Connectivity

OPNET models both E-UTRAN and EPC

• A UE is connected to a core network with IP connectivity

A UE is allowed to connect to only one EPC

• The UE also cannot change EPC in the simulation

EPC builds automatic GTP tunnels with an eNodeB for both uplink/downlink traffic communication with the UE

• No attributes need to be configured


14


How to Analyze Core Network Connectivity

Usually necessary only during troubleshooting when the UE does not receive any data when expected

Associated eNodeB statistic (discussed later)

Also lte_emm specifically traces connectivity with the core The first attach

Accept is received

around 99 seconds

Also, UE’s NAS state can be checked in graphical debugger at any time

Must be in

EMM_Connected.


15


How to Configure the Serving Cell

Scanning and Initial Cell Connectivity

• A UE can automatically detect nearby eNodeBs, possibly on different channels, and connect to them

• Criteria: First suitable or the best eNodeB

• Can also force eNodeB selection at a UE

What if no eNodeB is found nearby?

• All uplink/downlink packets dropped

• UE continues scanning for a new eNodeB

Unique for each eNodeB


16


How to Identify UE’s Current Serving Cell

Note: Connecting to an eNodeB also gives core network connectivity to that UE

eNodeB ID is configured at each eNodeB


17


How to Use the Serving Cell Knowledge for Rapid Analysis

UE may undergo radio link failures and experience disconnection from LTE core network

“-1” stands for no eNodeB


18


How to Deploy QoS in LTE

LTE Configuration node

• Define the EPS bearers and their properties

• Identify each by a unique name

• QCI determines scheduling priority: {5} > {1-4} > {6-9}

UE

• Deploy the EPS bearers by their names

• Define which application packets are mapped to each bearer

• Packets that are not mapped to any bearer (or all packets when no bearer defined) are mapped to “Default” bearer with QCI = 9


19


How to Analyze QoS in LTE

Per EPS bearer stats are available for collection

Stats are annotated with bearer names


20


How to Configure the Physical Profile

Step 1: Define Physical Profile in the LTE Configuration Node

• FDD and TDD profiles can be configured in the LTE configuration node

FDD: UL and DL subframes are configured separately

TDD: Common channel for UL and DL


21


How to Configure the Physical Profile (cont.)

Step 2: Deploy the required physical profile on the eNodeB

• Once a physical profile is deployed, it cannot be changed during the simulation


22


Use FDD when the utilization of both UL and DL subframes is almost the same

• Example: Voice or video conferencing traffic among the users in a cell

TDD when there is a asymmetric utilization of UL and DL subframes

• TDD Channel Index decides the UL:DL Division

Frame type cannot be changed during the simulation

Example: FTP or HTTP application traffic

When to Use FDD and TDD ?

TDD Channel

Index

UL: DL

Division

DL %

0 3:2 40%

1 2:3 60%

2 1:4 80%

3 3:7 70%

4 2:8 80%

5 1:9 90%

6 3:3:2:2 50%


23


How to Configure Downlink MIMO

Step 1: Configure the MIMO Transmission technique

• A UE may choose to use the MIMO Transmission technique configured on the eNodeB or use its own custom downlink MIMO transmission technique

By default “Downlink MIMO Transmission Technique” is set to “Use Serving eNodeB Setting”

• The MIMO transmission technique configured on the eNodeB is a cell wide setting for UEs without custom setting


24


Step 2: Configure the number of transmit antennas on eNodeB and receive antennas on UE

• Number of transmit antennas supported by eNodeB : 1, 2 or 4

• Number of receive antennas supported by UE : 1, 2 or 4

• A minimum antenna criterion must be satisfied to support a spatial multiplexing MIMO transmission technique

If not satisfied then “Transmit Diversity” will be used as the MIMO technique

How to Configure Downlink MIMO (cont.)

MIMO Transmission Technique

(Spatial Multiplexing)

Minimum number of

transmit antennas at eNodeB

Minimum number of

receive antennas at UE

2 Codewords- 2 Layers 2 2




25


How to Configure Downlink MIMO (cont.)

Step 3: Downlink multipath channel model must be configured to realize detailed physical layer effects of MIMO in “PHY Enabled” efficiency mode

• Transmit and Receive diversity are supported only for “PHY Enabled” efficiency mode

• Spatial Multiplexing is supported for “PHY Enabled” and “PHY Disabled” efficiency modes

“PHY Layer Enabled”

• MAC layer effects and detailed PHY layer effects of spatial multiplexing can be realized

“PHY Layer Disabled” : Efficiency mode which bypasses the PHY Layer

• Only MAC layer effects of spatial multiplexing can be realized

• Packet drops can be modeled statistically


26


How to Configure Uplink MIMO

Only receive diversity is feasible as only one UE Tx antenna is supported

• Must use “PHY Enabled” efficiency mode

Step 1: Uplink Multipath Channel Model must be configured

Step 2: Configure the number of receive antennas

• Number of transmit antennas supported by UE : 1

• Number of receive antennas supported by eNodeB : 1, 2 or 4


27


Default configuration

• Downlink : Spatial Multiplexing 2 codewords to 2 Layers with 2 Tx and 2 Rx antennas

• Uplink: Receive diversity with 1 Tx and 2 Rx antennas

Pros and Cons

• MIMO Spatial Multiplexing increases throughput but is more prone to physical layer impairments

Can potentially degrade performance when the link quality is bad

• MIMO Antenna Diversity (Transmit and Receive diversity) reduces the effects of multipath fading

Not so useful if the link quality is good

When the link quality is bad, using MIMO Antenna diversity technique yields better throughput than any other transmission schemes

MIMO Configuration


28


How to Analyze the Impact of PHY in LTE

Channel capacity depends upon:

• Channel Bandwidth - The higher the bandwidth, the higher the capacity

• Modulation and coding index (MCS) - The higher the MCS index, the greater the capacity

• Spatial Multiplexing – When enabled increases the capacity

Capacity estimate available in an OT table at the end of the simulation for each cell for both uplink and downlink

Channel

Bandwidth

(FDD only)

Capacity Estimate (Downlink) Mbps

Transmit Diversity 2x2 SM 2CW-2L SM 2CW-3L SM 2CW-4L

Low

Estimate

(MCS 0)

High Estimate

(MCS 28)

Low

Estimate

(MCS 0)

High Estimate

(MCS 28)

Low

Estimate

(MCS 0)

High Estimate

(MCS 28)

Low

Estimate

(MCS 0)

High Estimate

(MCS 28)

1.4 MHz 0.15 3.27 - 4.09 0.33 6.52 – 8.12 0.48 8.63 – 10.54 0.66 11.50 – 14.04

3 MHz 0.39 8.86 – 10.65 0.81 17.78 – 21.33 1.20 24.45 – 28.19 1.62 32.61 – 37.40

5 MHz 0.68 14.88 – 17.84 1.38 29.84 – 35.70 2.06 41.20 – 47.03 2.77 55.01 – 62.73

10 MHz 1.38 30.13 – 36.20 2.79 60.80 – 74.22 4.18 81.84 – 95.48 5.58 109.31 – 127.55

15 MHz 2.09 45.20 – 54.24 4.14 90.53 – 108.50 6.22 120.99 – 140.69 8.27 161.36 – 187.60

20 MHz 2.79 60.80 – 73.06 5.54 122.58 – 145.52 8.34 163.96 – 192.27 11.09 218.61 – 256.99


29


Agenda









30


Lab 1: Deploy and Analyze an LTE Network

An LTE “transit network” where static application users and an application server are connected to the Internet with LTE links

Deploy QoS in LTE to provide differentiated services to multiple applications

Improve throughput and performance for all users by analyzing statistics

Time: 35 minutes


31


Lab 1: Conclusions

Deployment of QoS improves the service offered to high priority traffic in congested LTE networks

Spatial Multiplexing may not benefit all users

• Use spatial multiplexing only when the signal quality is really good

Deployment of TDD can help serve asymmetric traffic better and improves performance potentially for everyone


32


Agenda









33


Analysis Tools

Statistics and ODB

• Discover the causal relationships between multiple observations

Application delay tracking

• Can track standard applications end to end – including the LTE portion

Parametric studies

• Parameter value that achieves the best performance

• Distributed simulations: Run multiple simulations on multiple CPUs

Visualization tools

• Time controller – helps correlating statistics with each other

• Terrain viewer – helps understand pathloss and terrain profile quickly

• Graphical ODB – reference session 1502

Reports

• Performance analysis web reports

• OT tables


34


Typical Planning and Analysis Workflow

Set/revise

assumptions Define

constraints

Deploy

scenario Analyze

Constraints not satisfied due to

unrealistic assumptions

Constraints

satisfied

Find optimal

configuration


35


Some Tips for Effective Planning Studies

LTE model offers the “efficiency mode” that bypasses PHY for speedup • Physical layer can be abstracted by properly configuring a global HARQ block error rate

• UEs can be configured with static MCS indexes to reflect their typical link quality

Raw traffic: IP flows, application flows

Large packets: Fewer events • Packets that are too large can cause undesirable effects; segmentation at TCP, IP, and

MAC may diminish the benefits eventually

Jammer nodes: Eliminating the need to model interfering neighbors explicitly

• Great acceleration potential for studies requiring interference


36


LTE Analysis: Jammer Nodes Study

Statistically same results with scenario without jammers

Example Networks Project: LTE Scenario:

video_perf_under_coch_interference_w_jammers

Jammer nodes abstracting neighbor cell interference – one

for the eNodeB (downlink) and one for all the UEs (uplink)

Time saving in large scale simulations for rapid analysis

Explicit neighbor cells

Jammer nodes


37


Statistical Validity of the Planning Studies

Reference – Session 1576

Good practice to run the simulation with many random seeds in a typical planning study

• Random seed acts as the promoted attribute

• Parametric studies workflow

OPNET provides results with their mean values as well as the statistical confidence interval (95% by default or user entered)

Observations should be statistically indifferent

• That is, their time series and mean values should look similar

• Especially useful for R&D + planning studies

Incorrect custom algorithms might give a wrong notion because the results “by chance” look good

But different random seeds can reveal those problems

Reference example: Lab 2, Session 1941, OPNETWORK 2008


38


Agenda









39


Provide the required coverage while minimizing one of the resources and constraining the others: • Number of cells

• Cell tower location/height

• Transmission power

Where the following are assumed to be known • Radio spectrum and the bandwidth

• Number of users

• Traffic per user

• Density of users per square units of a given geographic area

• Maximum transmission power of the users

Some other variations of the cell planning problem are also available

The Cell Planning Problem


40


How to Analyze the Cell Planning Problem

Cell planning should mainly provide coverage

• “Coverage” can be defined as that point from the center of the cell where the UE’s performance is deemed “acceptable”

At a minimum, the UE should connect to the eNodeB

More performance criteria are defined as well

Cell planning should account for mobility

• Need to plan cells so that handovers are as smooth as possible without service disruption

If the UE sees the strength of the current eNodeB is falling, it should find a new eNodeB in its vicinity


41


What Affects the Coverage of a UE?

UE’s coverage is affected by physical layer effects such as

• Terrain of the region

• Frequencies used for communication

Interference from neighboring cells

• Signal fading due to the pathloss

• Signal variance due to the multipath

Node mobility affects the coverage of UE

• A UE may suffer radio link failures which causes a loss of coverage

Remedies

• MIMO transmit or receive diversity can be used to reduce the effect of multipath

• eNodeB can make use of link adaptation to maintain a consistent link quality to reduce the physical layer effects

• Have additional cells so that UEs can handover without experiencing radio link failures


42


Link Adaptation

UE’s MCS index changes based on link quality

• Good signal – high MCS index, bad signal – low MCS index

Price paid for low MCS index is consumption of extra radio resources lowering the data rate of the channel

The eNodeB balances signal quality and channel capacity by keeping the MCS index at a maximum possible level


43


Link Adaptation: In-Cell Mobility

As the UE moves away:

• MCS index reduces to sustain link quality

• PDSCH utilization increases as more radio resources are required

Why did the traffic stop after some time?

UE was using a GBR bearer – must always be admitted

eNodeB can preemptively delete a GBR bearer if it can no longer guarantee the contract

Use “lte_adm” trace in ODB

• mltrace <enb_objid> lte_adm


44


Link Adaptation: Conclusions

If the UE ventures farther from the base station, its MCS index is lowered and it consumes more resources as a result

• Additionally, the UE will also spend more power to account for more transmissions

• In some cases, the system may become overloaded and UE’s services may be dropped by the “Admission Control” module

Thus the cell should be planned such that UEs should be guaranteed a certain level of service in the worst case

• The worst case could be defined by deciding a worst case MCS index value

• What if the level of service cannot be guaranteed in the worst case?

UE must handover to another cell so that there is no interruption in the cell coverage


45


Mobility and Handovers

Handovers

• Allowed across cells (different eNodeBs) belonging to the same core network

eNodeB in any frequency or even different technology (FDD/TDD) is allowed

• Not allowed across core networks (different EPC nodes)

EPC 0

EPC 0

EPC 1


46


How Handover Takes Place

Good behavior:

• eNodeB triggers handovers based upon UE’s measurement reports when a certain condition is not satisfied (we will see this soon)

Bad behavior

• UE encounters radio link failure, starts scanning for a new eNodeB and initiates attachment to the eNodeB that satisfies attachment criteria

Why would UE encounter radio link failure?

• Measurement reports are not received by the eNodeB due to interference

• There are no nearby altenative eNodeBs and the serving eNodeB cannot sustain communication with the UE

• Too much interference causes frequent failures even when the UE is relatively close to the serving eNodeB

A good eNodeB placement, reduction of interference and even data distribution across many eNodeBs can reduce the radio link failure problem


47


Why Does a UE Undergo Radio Link Failure?

mltrace <ue_objid> lte_rlf

N310 timeout: Typical when the UE is far away from the eNodeB

Other factors causing radio link failure

• RACH access failure

• RLC-AM maximum retransmission exceeded threshold

In ideal situations, eNodeB should have handed over the UE before


48


Handover Triggers

Triggers based upon two main statistics: RSRP and RSRQ

• RSRP: measures the strength of signal from the current eNodeB

• RSRQ: measures the “quality” of signal from the serving cell by considering interference from neighbors

Trigger handover if the strength of serving

eNodeB <= -112 dBm

Trigger handover if the quality of signal from

the serving eNodeB <= -5 dB

Above trigger values are standard recommended

• But they can still be customized based upon a given scenario’s requirements

Target eNodeB has to satisfy entry threshold too

• Note: Selection threshold >= RSRP Threshold


49


Using Statistics to Analyze Handover Issues

RSRP of the scanned eNodeB below threshold (-110 dBm)

eNodeBs are too far apart

• Either reduce the cell range or increase the transmission power

Radio link failure


50


Using ODB to Analyze Handovers

mltrace <src_enb_objid> lte_handover_for_<ue_name>

RSRQ threshold exceeded for some time now (-7.00 dB)

eNodeB 2 is the only eligible eNodeB

• RSRP and RSRQ values of that eNodeB mapped to an index

• Index to value map: RSRP: (-140 + index) dBm, RSRQ: (index - 40)/2 dB


51


The Three Resources Affecting Cell Coverage

Tower location/height

• Taller towers better LOS

Costs more

May increase pathloss beyond a point

• Choose an optimal location based on the terrain

The location may be unavailable or expensive

Transmission Power

• Higher transmission power More coverage

Also can increase interference at the cell edges

Number of cells

• More cells UE’s can handover without radio link failure

Adding more cells can potentially increase interference

Adding more cells can be costly and yield diminishing returns


52


The “Resources” for Cell Planning: Recap

Number of cells

Per

form

ance

Diminishing returns

Tower height

Per

form

ance

Increasing pathloss Increasing LoS

Transmission Power

Per

form

ance

Increasing interference Increasing coverage

OPNET Modeler helps in identifying the optimal operating points on similar performance curves under the presence of realistic terrain, mobility and physical layer data


53


Lab 2: Planning LTE Cell Placement to Provide Coverage and Minimize Interference

Deploy an LTE network on a terrain in Nevada to provide coverage to 30 UEs

• Start the deployment with a single cell and check if it is sufficient to provide coverage

• By assuming a maximum of 50 meters of tower height, deploy multiple eNodeBs to provide 100% coverage

Adjust the transmission power of one or more eNodeBs to minimize downlink interference and improve application traffic performance

• Also find out if some eNodeBs are redundant

Time: 35 minutes


54


Lab 2: Conclusions

The terrain modeling module (TMM) allows us to model a realistic LTE cell deployment study

Using the in-built LTE statistics, it is easy to rapidly analyze the cell coverage

Using the TMM visualization tool, it is easy to check the line of sight (LoS) coverage and deploy multiple cells to provide the basic LoS coverage

Parametric studies tool allowed us to lower transmission power of two eNodeBs and improve performance

• At first, interference was reduced

• Later, we discovered that the eNodeBs were redundant and could be eliminated from the network


55


Agenda









56


The Capacity Planning Problem

Number of users

Maximize the number of users while

• Satisfying the application delay constraint

• Maximizing the throughput

Sometimes, increasing number of users can affect throughput negatively due to TCP congestion window effects, increased interference, scheduling overheads and retransmissions, etc.

Relationship between capacity planning and cell planning

• A well designed cell will also have a higher “capacity potential”


57


How to Analyze the Capacity Planning Problem

Need to define “acceptable” application performance

• Service level agreements (SLAs) can help define constraints on application delay

Need to understand factors affecting channel capacity

• Bandwidth: The higher the bandwidth, the higher the capacity

• MCS Index: The higher the MCS index, the higher the capacity

• Spatial Multiplexing: Increases the capacity when enabled

• Other factors: Overheads (MAC overheads, LTE physical overheads), retransmissions (TCP, RLC, HARQ)

Need to understand which channels can experience saturation

• Three channels: PDSCH (downlink shared), PDCCH (downlink control), PUSCH (uplink shared)

• For good performance, all three channels should be below saturation level

OPNET Modeler provides statistics, reports, and traces to help analyze channel capacity


58


Channel Capacity Reports

Capacity reports published for both uplink/downlink for each eNodeB

Assumes a single-UE case

• That is, a single UE saturates the channel with its traffic

Overheads will consume some reported capacity; the rest is available for good throughput

• MAC, RLC, PDCP

• Hence application throughput will be lesser

If all UEs are approximately similar (same MCS indexes), it is easy to estimate channel capacity

• With a mix of UEs, it is difficult, and that’s where planning studies are useful in OPNET Modeler


59


Channel Utilization Statistics

There are Three channels of importance:

• PDCCH: Physical Downlink Control Channel

• PDSCH: Physical Downlink Shared Channel

• PUSCH: Physical Uplink Shared Channel

Utilization statistics track how much channels are utilized to detect if they are overloaded

Overloaded

downlink


60


ODB Tracing into an LTE Frame

mltrace <enb_objid> lte_frm

Complete breakdown of uplink and downlink subframes

• Number of resource blocks and corresponding bits fitted for a given UE

Per codeword information in case of spatial multiplexing transmission in downlink subframes

• Higher layer payload in the LTE MPDU

• Frequency information.

It sometimes helps to look at the ODB output instead of just statistics

• Per UE contribution to a subframe can be found

• Bit carrying capacity per UE per subframe can be found


61


Role of Admission Control Module

Relevant only for GBR bearers

• Before admitting into the system, admission control performs a “rough check” on radio resources to check if there is space to admit

• Some things such as retransmissions cannot be anticipated apriori

Admission control can be made flexible using “loading factor”

To account for retransmissions, set the loading factor < 1

• Usually admitted GBR bearers have satisfactory traffic performance

All admitted, none preempted/rejected


62


Understanding Traffic Statistics

PDCP/RLC

MAC

PHY

Total higher layer traffic

sent to PDCP/RLC

Includes new transmissions

and HARQ retransmissions

Transmitter

PDCP/RLC

MAC

PHY

Receiver

Total higher layer traffic

forwarded by PDCP Higher Layer Higher Layer

Recorded when HARQ decoding

is successful

Higher layer traffic sent to

PDCP/RLC per EPS bearer

Includes new transmissions ,

retransmissions and status

reports

Higher layer traffic forwarded by

PDCP per EPS Bearer - “Good

throughput” per EPS Bearer

Delay

Delay for all traffic that is

delivered to the higher layer

Delay for traffic that is

delivered to the higher layer

per EPS bearer


63


Understanding LTE Traffic via Statistics (cont.)

Examples

• Collecting “MAC Traffic sent” at an eNodeB shows the traffic load on the downlink channel (PDSCH)

• Applying the “Adder” filter for “Traffic received” by the UEs in a cell can show traffic throughput for the downlink channel (PDSCH)

• Applying the “Adder” filter for “Traffic sent” by the UEs in a cell can show traffic load for the uplink channel (PUSCH)

• Collecting “MAC traffic received” at the eNodeB shows the throughput on the uplink channel (PUSCH)

The difference between load and throughput is dropped traffic due to:

Congestion at the MAC

Dropped packets at the physical layer due to interference/fading

Additionally traffic sent (bps, packets/sec), traffic received (bps/packets per second), and delay statistics is available per GTP tunnel as well


64


Configurations of Physical Channels and Effect of Control Channels on Capacity

Downlink

• PDSCH: Physical Downlink Shared Channel

• PDCCH: Physical Downlink Control Channel

• The size of PDCCH is dynamic…Model automatically “resizes” PDCCH to maximize the number of resource elements left to PDSCH

Uplink

• PUSCH: Physical Uplink Shared Channel

• PRACH: Physical Random Access Channel – bigger PRACH subtracts capacity from PUSCH

• PUCCH: Physical Uplink Control Channel – bigger PUCCH subtracts capacity from PUSCH


65


“Resources” Affecting Channel Capacity

Capacity Planning Resource

Effect of more resource Cost-benefit

Number of Users

More users more load offered

to the channel

Less users less load

offered to the channel

Having more users increases the offered load

and can worsen performance for everyone

More users mean more revenue, so as long as the application SLAs are

satisfied, number of users should be maximized

Channel Bandwidth

More bandwidth More space to

carry data

Less bandwidth less space to

carry data

More channel bandwidth means there is more space

to carry the same amount of data which leads to lower

channel utilization and better application

performance

More bandwidth is clearly desirable, but it can cost more money to buy this

resource

Multiple Antennas (MIMO spatial

multiplexing)

Enabling MIMO spatial

multiplexing increase in

capacity

Disabling MIMO spatial multiplexing

lesser capacity

Enabling MIMO spatial multiplexing increases the subframe capacity without increasing the bandwidth

MIMO spatial multiplexing will only be beneficial if the link quality of the UE is really good otherwise there will be a lot

of retransmissions as MIMO spatial multiplexing is very susceptible to

physical layer effects.

Transmission Power

More TX power better coverage

and higher MCS index unless

interference is high

Less TX power less

coverage and lower MCS

index

Increasing power can improve the UE’s MCS

index and effective capacity is increased. However too much TX power can cause

interference and the capacity gain is offset by

retransmissions

Increasing power drains UE’s battery faster and can cause interference. Hence power should be minimized as long as

coverage criterion is satisfied


66


The “Resources” for Capacity Planning: Recap

Number of users

Per

form

ance

Bandwidth

Per

form

ance

Transmission Power

Per

form

ance

Increasing

retransmissions

Increasing MCS

OPNET Modeler helps in identifying the optimal operating points on similar performance curves under the presence of realistic terrain, mobility and physical layer data

Application SLA

Operator budget

Link quality

Per

form

ance

MIMO Transmit Diversity

MIMO Spatial Multiplexing


67


Lab 3: Planning an LTE Cell to Determine Maximum Number of Users

Deploy a single cell LTE network on a Nevada terrain region to determine the number of users that can be supported in that cell

Define an acceptable application service level agreement (SLA) criterion. Determine if 50 users can be supported with the given traffic profile at the beginning.

If the SLA is not satisfied, progressively reduce the number of users. • Choose which users to eliminate intelligently; the users that achieve the lowest MCS

index should be eliminated to maximize the number of supported users

Determine using the above algorithm the maximum number of users that can be supported in the cell

Time: 25 minutes


68


Lab 3: Conclusions

OPNET’s LTE solution provides an easy way to estimate the channel capacity (OT table reports) and channel utilization (in-built statistics) to rapidly analyze capacity planning problem

The Top Statistic utility allowed us to find the UEs with the lowest MCS index rapidly

The capacity planning problem was solved by finding the best 32 UEs that could be supported without compromising the application quality using the iterative algorithm


69


Agenda









70


Battery Life Planning Problem

Cell planning and capacity planning studies are throughput oriented since we are attempting to solve

• How to increase the system throughput by planning well placed cells covering a region

• How to squeeze as many users as possible for maximum revenue

However to achieve those objectives, the UE may end up spending its battery by increasing its transmission power

Why would the battery life be important?

• Autonomous and unmanned sensor networks: Static UEs

• Mobile UEs that do not have readily available charger

Power saving feature in LTE:

• “Discontinuous Reception” (DRX) in “RRC Connected” state

• Idle mode


71


DRX in RRC Connected State

A UE starts a DRX cycle when there is no activity on the medium for a certain time

• A UE is in either “DRX Inactive” or “DRX” phase at all the times

• If there is no activity on the medium for a duration of inactivity timer, “DRX” phase begins

• UE runs DRX cycle in DRX phase

DRX cycle consists of an “active” period and “sleep” period

During “active” period, UE listens to PDCCH for downlink activity

At the beginning of a DRX phase UE runs short DRX cycle first

• If the short DRX cycle completes successfully without transitioning to “DRX Inactive” phase, then from that point on UE will run only the long DRX cycle until it transitions to inactive period

DRX Configuration Enables or Disables “DRX in RRC

Connected State” for the UE

Duration of active

period

Duration of short DRX cycle

which includes active period

Inactivity Timer

Multiplication factor to the short

DRX cycle gives the duration of

long DRX cycle


72


Idle Mode

A UE transitions to “idle mode” when there is no activity on the medium for a certain time

When in idle mode a UE…

• Is disconnected from the core

• Runs “DRX” cycles to save power

• Keeps track of the best eNodeB by performing “Cell Reselection” procedure

Sends a Tracking Area Update message to the core if the current eNodeB belongs to a different TA than the TA of the previous one

• Reconnects to the core if uplink or downlink data activity is detected

Monitors the paging channel to detect downlink traffic


73


Idle Mode—Tracking Area Update and Paging

An idling UE must update the EPC of its current tracking area

• A group of eNodeBs can be in the same tracking area

EPC pages eNodeBs in UE’s current tracking area when there is a DL traffic

• eNodeBs broadcast the page to UE

Tracking area size must be wisely chosen

• A larger tracking area may potentially have higher paging load and therefore lesser resources for PDSCH

• If the tracking areas are small, idling UEs may need to send frequent tracking area updates, which reduces its battery life


74


Idle Mode Attributes

RRC Connection Release Timer • eNodeB Trigger

• Upon expiry of this timer for a UE, the serving eNodeB starts UE context release with the EPC

Idle Mode Support • eNodeB Triggered, Enabled or Disabled

• Enabled : eNodeB or UE Triggered

T3440 • UE Trigger

• Upon expiry of this timer UE enters Idle mode if UE triggered

Cell Reselection attributes used only during idle mode

Attributes related to Paging

Tracking area update attributes


75


How to Analyze the UE’s Battery Power Expenditure

Battery and power expenditure model

OT Report


76


Pros and Cons of DRX in RRC Connected State and Idle Mode

DRX in RRC Connected State and Idle mode are two independent power saving features in LTE

DRX in RRC Connected State Idle Mode

• UE is attached and bearers are active

• No additional signaling required after

the initial attachment

• Usually used during short periods of

inactivity

• A UE in idle mode is disconnected and

bearers are inactive

• Signaling required to go into and to

come out of idle mode

• Usually used during long periods of

inactivity

OPNET Modeler helps can identifying the optimal operating parameters that will yield maximum battery life and still have good application performance


77


Lab 4: Maximize UE’s Battery Life Within Application SLAs

Deploy a single cell sensor network where the UEs receive a command to send sensor data to a command center

• Unless the command is received, sensor data is not sent

• Sensor data is time critical

• Sensors are costly to replace…hence their battery life is extremely important

Define an acceptable application SLA

Verify the battery life with just the idle mode only enabled and idle mode and DRX both enabled

Conduct a parametric study to find the optimal configuration resulting in most battery life

Time: 30 minutes


78


Lab 4: Conclusions

With default configuration we noticed that have idle mode and DRX enabled yielded the maximum power savings

OPNET Modeler allows the users to define the application SLA and the number of violation instances very easily.

Using the parametric studies feature and the distributed grid computing, one can easily determine the optimal value of the DRX sleep parameter for the given application that can satisfy the application SLA.

Excise caution when deciding the idle mode and DRX setting as they depend on the application behavior


79


Documentation References

Some important 3GPP Standards

• 36213-880: for the physical layer

• 36300-910: for the overall description of E-UTRAN

• 36321-900: for the MAC operation

• 36322-870: for the RLC operation

• 36331-900: for the RRC protocol

• 23203-830: for the policy and control architecture

• 23401-860: for the EUTRAN access network

OPNET Published (LTE consortium website)

• LTE Phase I, II, III, IV, V, and VI Requirements Documents

• Requirements document for the idle mode operation (LTE Phase VII)

• LTE Frame Generator and Scheduler Description

• LTE Modulation Models

• LTE Multipath Fading Models


80


User Community and Technical Support

Join the OPNET products online user forum: https://splash.riverbed.com/community/product-lines/opnet

Riverbed Technical Support

• https://support.riverbed.com/

• 1.415.247.7381 or 1.888.782.3822 toll-free in the US or Canada

International phone support numbers are available at:

https://support.riverbed.com/contact/index.htm

• Knowledge base:

https://supportkb.riverbed.com/support/index?page=home

https://splash.riverbed.com/community/product-lines/opnet





https://support.riverbed.com/

https://support.riverbed.com/







81


riverbed | splash for Modeler Users

riverbed | splash is the place to connect with OPNET product experts

• Extensions—Download new product enhancements submitted by Riverbed’s OPNET product staff and customers

• Community—Discuss your challenges; announce your successes

Login with your customer username and password at https://splash.riverbed.com/

Is there anything on riverbed | splash for Modeler users?

• Sure! For example, the contributed model for IEEE 802.15.3/3b WPAN technology can be found under this link: https://splash.riverbed.com/docs/DOC-3079




https://splash.riverbed.com/docs/DOC-3079





82


Related Sessions

1580: Modeling Custom Wireless Effects

1598: Productivity and Code Efficiency Tips for OPNET Modeler Users

1576: Obtaining Statistically Valid Simulation Results: Generation, Interpretation, and Presentation

1586: Building Realistic Application Models for Discrete Event Simulation

OPNETWORK 2011 – 1581: Understanding LTE Models Internals and Interfaces


83


Take-Away Points

OPNET Modeler supports easy deployment and auto-configuration of LTE networks using Wireless Network Deployment Wizard

Terrain modeling module can be used to model terrain and custom propagation models, which works seamlessly with LTE models

Urban propagation model coming up shortly (Session 1574 -New and Improved Features for Modeling and Simulation)

LTE model library in OPNET Modeler is comprehensive with a variety of features which can be used in your planning and analysis studies

• Cell planning, Capacity planning and battery life planning are some examples

OPNET provides extensive capabilities to provide what-if analysis

• Perform parametric studies with distributed execution

• Evaluate protocol setting and network parameters to optimize performance


84


Acronyms

3GPP: 3rd Generation Partnership Project

QoS: Quality of Service

OFDMA: Orthogonal Frequency-Division Multiple Access

SC-FDMA: Single-Carrier Frequency-Division Multiple Access

LTE: Long Term Evolution

• 4G: 4th Generation

UMTS: Universal Mobile Telecommunications System

• 3G: 3rd Generation

EPS: Evolved Packet System

EPC: Evolved Packet Core

E-UTRAN: Evolved UMTS Terrestrial Radio Access Network

GTP: GPRS Tunneling Protocol

eNodeB: Enhanced NodeB

UE: User Equipment

FDD: Frequency Division Duplex

TDD: Time Division Duplex

RSRP: Reference Signal Received Power

RSRQ: Reference Signal Received Quality

HARQ: Hybrid Automatic Retransmission reQuest

RLC-AM: Radio Link Control – Acknowledgment Mode

OpNET - LTE

Documents

Transcript of OpNET - LTE