LTE RNP Guidelines

Nokia Siemens Networks

LTE Radio Network Planning Guidelines

RL20 FDD

Revision HistoryDateRev.Summary of Change

30.04.2009v 0.3Draft version for review

02.06.2009v 1.0Official version after feedbacks review

30.12.2009v 2.0Additions: RACH planning, 6-sector site and Link Budget Comparison sections. Updates: LTE power budget to DimTool v2.3.0, Scenarios and use cases and smaller updates to ensure contents are up to date.

09.07.2010V 3.0Updates LTE power budget to DimTool v2.3.2, RACH Planning signatures, KPI, Parameters, LTE performance.

12.01.2011V 4.0Update for RL20, removal duplicities and adding references, adding SINR RSRQ relation, and review comments.

Editor/s:T. Novosad, L. Serna, , C. Johnson

Date:12.01.2011

Version:v 4.0

Table of contents

1 Scope62 LTE Overview72.1 Expectations and Markets72.2 Architecture92.2.1 Network Elements102.2.2 Interfaces112.3 Air-Interface122.3.1 OFDMA122.3.1.1 Subcarrier types152.3.1.2 OFDMA Symbol162.3.1.3 LTE Physical Layer Structure162.3.1.4 Physical Resource Block and Resource Element172.3.1.5 Downlink Resource Allocation182.3.1.6 OFDMA Parameters192.3.1.7 Downlink Reference Signals192.3.2 SC-FDMA202.3.2.1 SC-FDMA Symbol 212.3.2.2 Uplink Reference Signals232.3.2.3 Uplink Physical Resource Block 232.3.2.4 Uplink Resource Allocation242.4 Radio Resource Management262.4.1 Scheduler262.4.1.1 Downlink Scheduler272.4.1.2 Uplink Scheduler292.4.2 Power Control302.4.2.1 Slow Uplink Power Control312.4.2.2 Downlink Power Control322.4.3 Admission Control332.4.4 Link Adaptation (Adaptive modulation and coding)332.4.4.1 Downlink link adaptation procedure352.4.4.2 Uplink link adaptation procedure372.4.5 Connection Mobility Control382.4.5.1 Handovers within NSN implementation412.4.6 MIMO Mode Control432.4.6.1 Transmit Diversity442.4.6.2 Spatial Multiplexing (MIMO)442.4.6.3 MIMO Mode Control462.4.7 Outer Link Quality Control (OLQC)472.5 Comparison with UMTS483 NSN LTE Solution513.1 Flexi eNodeB513.2 eNodeB Configurations533.3 6-sector Sites 563.4 RL09 RL10, and RL20 Features583.4.1 RRM and Telecom583.4.2 Performance and Capacity653.4.3 Overview of main changes in RL20 release against previous releases 664 Planning Process684.1 Scenarios and Use cases704.1.1 Greenfield704.1.2 Existing Operator704.2 Propagation Modelling724.3 Planning Tools754.4 Atoll754.5 Mentum Planet764.6 NetAct Planner764.7 Simulation Based Planning774.8 Path Loss Based Planning814.9 Common Practice844.10 Link Budget and Planning Thresholds864.10.1 RSSI versus RSRP and RSRQ measurements864.10.2 LTE Power Budget894.10.2.1 MCS selection Downlink954.10.2.2 Uplink Specifics974.10.3 Planning thresholds974.10.4 Link and system level simulations984.10.5 Link Level Simulation Results994.10.6 System Level Simulation Results1004.11 Link Budget Comparison Example 1024.12 Special scenario consideration for fix wireless broadband network 1034.12.1 Special consideration example for National Broadband Network (NBN) project1035 Performance Expectations1075.1 Peak Throughputs and Peak Spectral Efficiency1085.2 System Level Performance1095.2.1 Cell Level Performance1095.2.2 Latency1115.2.3 Mobility and cell ranges1115.3 Comparison with other technologies1145.4 LTE UE Categories1155.5 LTE user throughput based on HSPA Performance1166 Detailed Planning1176.1 Frequency Bands and Spectrum Re-farming1176.2 Site Selection1216.3 Site Design1246.4 Antenna and eNB Parameters1266.5 Physical Layer Cell Identity Planning1286.5.1 Planning129 : . 1306.6 eNodeB and Cell Identity Planning1316.6.1 Planning1316.7 RF Carrier Allocation1326.8 Neighbour Planning1346.8.1 LTE1346.8.2 UMTS1366.9 Tracking Area Planning1376.9.1 Planning1376.10 MME Area Planning1396.11 PRACH Planning1406.12 PUCCH Capacity & Configuration Planning 1426.13 PDCCH Capacity & Cofiguration Planning1436.14 UL DM RS Planning1436.15 Co-Siting1446.15.1 LTE and GSM1456.15.2 LTE and UMTS1496.15.3 WiMAX1526.15.4 Achieving Sufficient Isolation1566.16 Indoor Solutions1596.17 Microcells1666.18 Coverage and Capacity1696.18.1 Coverage169 Site Design and Network Planning169 Equipment Capability and Performance170 Parameter Configuration1706.18.2 Capacity1706.19 Network Evolution1726.19.1 LTE1726.19.2 LTE Advanced1747 Parameters1758 Key Performance Indicators1779 References18010 Glossary18311 Comparison between UMTS and LTE186

1 ScopeThis document is intended to provide a basic understanding of LTE (Long Term Evolution) radio aspects together with a set of radio network planning guidelines and recommendations that can be used to develop radio network planning processes. The detailed specification of these processes is outside the scope of this document. These processes are typically customized to suit the specific requirements of an operator.

The present version of the guideline covers NSNs LTE RL10 and RL20, the first commercial releases. It includes also aspects of the RL09 (Trial release). RL10 is dedicated to the FDD mode of operation for LTE. RL15TD is the counterpart release for TD LTE and it is not covered in the this version of the planning guidelines. TD LTE Planning guidelines are a separate document nowadays see reference 17]. In some cases, features from future releases are described or mentioned to provide users with a broader view of the system.

The document is organized into the following sections:

Section 2 presents an overview of the LTE technology. Its introduction as the next step in mobile radio communications after 3G/HSPA and the new concepts that define this evolution: new architecture and new air interface. Radio Resource management functionalities and air interface concepts are presented with a level of detail such that planners can obtain a good basic knowledge of LTE.

Whereas Section 2 is, to a certain extent, generic (some implementation details are NSNs specific) Section 3 is purely related to NSNs implementation of LTE: the Flexi Multiradio eNodeB and NSNs RL09, RL10 and RL20 LTE features.

Section 4 describes a general approach to LTE planning: It points out the different planning considerations when facing a greenfield operator case or an existing operator case. It also describes different approaches to planning (simulation based and path loss based) and the link budget, a fundamental element in network planning. Finally, link level and system level simulation results are presented for reference use.

Section 5 focuses on the Planning Tool options currently considered by NSN for LTE planning. Additional tools may be considered in the future. LTE expected performance is discussed in Section 6 including reference to material about of how to calculate LTE throughputs based on HSDPA (Rel.6) throughputs.

Section 7 focuses on practical aspects of LTE planning. Guidelines for site selection, design and configuration are proposed. It explains in detail the different LTE deployment scenarios: macro cells, indoor solutions and microcells and finally, it considers important aspects of the LTE deployment such as spectrum re-farming, mobility (intra-system and inter-RAT) and neighbour planning.

Sections 8 and 9 present an overview of the LTE parameters and LTE KPIs based on the information available at this early stage.This document will be updated at regular intervals to ensure that it considers the latest best practice. Ensuring that the latest best practice is included requires that feedback is provided regularly from NSN projects. This is positively encouraged. Feedback should be provided to Lorena.Serna@nsn.com, Tomas.Novosad@nsn.com or Chris.Johnson@nsn.com.Additional sections can be requested at any time and will be included as and when appropriate.2 LTE Overview

2.1 Expectations and Markets

LTE stands for Long Term Evolution and it represents the next step in mobile radio communications after HSPA. 3GPP is the standardization body behind LTE. The technical specifications for LTE air interface are defined in 3GPP Release 8 and they can be found in the 36-series (TS 36.xxx). Another name for LTE is e-UTRA (Evolved-UMTS Terrestrial Radio Access).

LTE requirements are specified in 3GPP TS25.913. Main requirements are:

Peak data rate of 50 / 100 Mbps (uplink / downlink)

Reduced latency enabling RTT (round trip time) 17, 10MHz ->110, 20 MHz -> 110. Default value depends on uplink channel bandwidth being: 4 users for 5 and 10MHz and 10 users for 20MHz Frequency domain scheduling: How many PRBs and in which location are allocated for each scheduled UE.

2.4.1.2 Uplink SchedulerThe UL scheduler is responsible for the assignment of the UL physical resources (PRBs) in the time and frequency domains. In order to do this, the UL scheduler evaluates first what are the available physical resources per transmission time interval (TTI = 1 ms) and then, selects the UEs that will be scheduled in the next TTI. The result of the UL scheduler evaluation is sent by means of grants to the appropriate UEs via the L1/L2 Control Signalling in DL.

UL scheduler is a Round Robin scheduler and works on a per cell basis. Unlike the DL scheduler, it is channel unaware in NSNs implementation. That is, the scheduling decisions do not take into account the channel conditions. However, channel aware UL scheduling is planned for RL30 onwards.

The maximum number of UEs which can be scheduled per TTI time frame is restricted by an O&M parameter.

When assigning PRBs to a UE, the UL scheduler takes into account the buffered data volumes of the data radio bearers (DRB) and of the signalling radio bearers (SRB). Since the scheduling decisions are performed by the eNodeB and the buffer is in the UE, the UE needs to report the buffer data volume. This reporting is done for each logical channel group (it can not be on a bearer bases). The UE is responsible for selecting from which radio bearer(s) the data is taken. The task/bearer prioritisation for the UL scheduler is as follows:

Hybrid ARQ (HARQ) retransmissions. Note: In case of HARQ retransmissions they will use the same MCS, the same rate matching and the same number of PRBs but may use different PRBs in the frequency domain.

Random access procedure

Scheduling request

Signalling radio bearer

Data radio bearerInteractions of UL scheduler with other RRM functions

Radio Admission Control (RAC): RAC informs the UL scheduler whenever a data or signalling radio bearer is setup or released. RAC also provides the minimum and maximum bit rates of the data radio bearers whenever they are setup. Link Adaptation: UL Adaptive modulation and Coding (AMC) provides per UE the Modulation and Coding Scheme (MCS) and the maximum number of PRBs. UL scheduler provides an indication to the UL AMC for each UE scheduled in the next TTI as well as the amount of PRB which is assigned to the UE. For the DL AMC, the UL scheduler provides a list of UEs scheduled in the next TTI.UL scheduler algorithm

The UL scheduler is divided in two main functions:

The evaluation of the resources that can be used for UL data transfer in the next TTI (step 1 in Figure 20)

The scheduling algorithm that assigns the available resources to the UE (step 2 in Figure 20)

Figure 20 - UL Scheduler algorithm

The PRBs available for UL scheduling are those which are not used for signalling or by the random access procedure.

Scheduling for the available PRBs is done in the time and frequency domain. In time domain, the UEs to be scheduled are prioritised. The maximum number of UEs that can be scheduled per TTI is determined by a configurable parameter: MAX_#_UE_UL (maxNumUeUl). After the decision of how many UEs will be scheduled per TTI the frequency domain scheduling determines the PRBs that will be assigned to those UEs.

In channel unaware scheduling a random function (blind hopping) decides the position of the PRB groups within each TTI. The PRB groups are evenly distributed over the allowed frequency range: if not all of the PRBs are assigned and some physical resources are left over, the unassigned PRBs are considered as physical resources of virtual UEs and the hopping algorithm handles the unassigned PRBs in the same manner as the allocated PRBs.

2.4.2 Power ControlPower control in uplink is based on the slow uplink power control scheme that is a combination of open loop power control and closed loop power control. In downlink, power control is supported via adjustable parameters. In the next sections, both uplink and downlink power control are described in more detail.

Note: Power Control for LTE is standardised within 3GPP in TS 36. 213 whereas for UMTS Power Control is not completely standardised.2.4.2.1 Slow Uplink Power ControlPower control is a mean to improve the cell edge behaviour (improve bit rates and have reliable control plane operation), to reduce inter-cell interference and to reduce the UE power consumption. In the LTE UL, the performance of the system heavily depends on the power control scheme.

The slow uplink power control scheme is applied separately by using different parameter sets for the different uplink physical channels: PUSCH (physical uplink shared channel) and PUCCH (physical uplink control channel) and for the sounding reference signal (SRS). The scope of UL power is UE level, that is, it is performed independently for each UE in a cell.

LTE power control requirements are less stringent than for WCDMA. Power control in WCDMA was designed for radio link adaptation under continuous transmission in circuit-switched services. In LTE, there is already fast scheduling of UEs (1 ms, TTI). So, whereas the power control in WCDMA is periodic with a frequency of 1500 Hz (every 0.67 ms loop delay) with power steps of +-1 dB in LTE it is possible to have larger power steps with lower frequencies (5 ms minimum loop delay) and it does not need to be periodic. This is why it is called slow power control.

Slow UL power control combines two components:

Open loop power control calculated at the UE

Closed loop power control corrections transmitted by the eNodeB

The open loop power control is performed independently by the UE and compensates for long-term channel variations (e.g. path loss and shadowing). It is based on path loss measurements, broadcasted parameters and RRC signalled parameters from the eNodeB. However, open loop power control suffers from errors in UE path loss measurements so closed loop power control is used.

The closed loop power control is less sensitive to errors in measurements because it is based on interactions between UE and eNodeB. The eNodeB performs UL level and quality measurements. Based on these measurements (weighed and averaged) the eNodeB sends power correction commands to the UE to maintain, increase or decrease the transmit power.

The UE uses the closed loop power correction values on top of the open loop component to calculate its total uplink transmit power.

Open loop power control is a basic functionality. Closed loop power control functionality can be enabled / disabled although the best way of operation is when both (open and closed power control) are combined. There can be an overall cell throughput gain of ~20% when using both open and closed power control compared to using only open loop power control.Slow UL power control procedure In UL, the total UE power is accumulated only on the subcarriers that are used for transmission. The UE controls its total output power to keep the transmitted power spectral density (PSD) constant and independent of the allocated transmit bandwidth (number of PRBs)

While there is no feedback from the eNodeB (no UL power control commands) the UE performs open loop power control based on path loss measurements, system parameters and RRC signalling. This is important for the start of the data transfer.

Started the data transfer, the UE receives the power control commands from the eNodeB and corrects the power spectral density.

Figure 21 - Slow UL power control procedure

2.4.2.2 Downlink Power ControlIn downlink, the total eNodeB Tx power is shared equally between all PRBs and subcarriers available in the given channel bandwidth. This is called flat power allocation since each subcarrier is transmitted with the same and constant power independently of the resource occupation. The possibility of boosting the power of certain subcarriers, e.g. control channels, is under study but it is not standardized by 3GPP.

Note that for an eNodeB with the same total Tx power, the power per subcarrier will be higher in smaller bandwidths than in larger bandwidths and consequently, under the same conditions, the downlink coverage will be higher for smaller bandwidths than for larger ones.

DL power control is supported by means of adjustable parameters. It is possible to reduce the cell power (via parameter) to obtain a flat power spectral density for the control of the coverage. In case of multiple transmit antennas (i.e. MIMO) the power is given per transmit antenna. Power for different physical channels is handled separately with different parameter settings for each channel (PDSCH Physical Downlink Shared Channel-, control channels, reference signals, synch channel and common channels). Detailed explaination of Control channel settings is in the Optimization Guide 26].

2.4.3 Admission ControlAdmission control is one of the main functions of the L3 RRC (Radio Resource Control) control plane and is implemented in the eNodeB.

The task of RAC (Radio admission control) is to admit or to reject the requests for establishment of Radio Bearers (RB). Reasons for requesting radio bearers are a paging event, a handover request or a user request for establishing a call. Each of those requests is evaluated with respect to the current cell load and the effect the new bearer can have on existing connections. The scope of RAC is cell level.

In NSNs RL09, the RAC is not-Quality-of-Service-aware and it is based on the number of RRC connections and the number of active UEs (users) per cell. It is a simple implementation for admission control that will be enhanced in future releases.

Radio Bearers in case of cell access due to handover (HO) are handled by the target cell as all or nothing, this is, either both Signalling Radio Bearer (SRB) AND Data Radio Bearer (DRB) are admitted or the UE is rejected. In RL09 all the SRBs are admitted by default. The DRB establishment requests for HO have higher priority than those for normal access.

In RL20, not all the SRBs are admitted by default including the request for SRB coming from handovers and there are different admission priorities for UEs accessing the cell via HO depending on the HO cause.

RAC interacts with the Scheduler by informing it every time a new bearer is set up or released.

2.4.4 Link Adaptation (Adaptive modulation and coding)Link adaptation is a fundamental function of the air interface for efficient data transfer. It dynamically adjusts the data rate for each user based on the channel conditions in order to improve the system capacity and the coverage reliability. It is closely related to the modulation and coding scheme used. This is why the terms Link adaptation and adaptive modulation and coding are used together.

In downlink, the eNodeB decides the modulation and coding scheme based on the downlink channel condition by considering the CQI (channel quality indicator) reports that the UE sends in uplink. This CQI reporting can be done per TTI period and that is why it is also called Fast AMC. The CQI is an indication of the data rate that can be supported by the downlink channel based on the SINR and the characteristics of the UEs receiver.

If SINR is good then higher modulation and coding schemes (MCS) can be used which implies more bits per symbol are sent and higher throughputs are achieved.

If SINR is poor then lower order MCS (i.e. QPSK) should be used which implies fewer bits per modulated symbol are sent and although the achieved throughputs are lower the communication is more robust because it has higher tolerance to interference.

The itneraction between DL AMC and DL power control is not defined / supported in the initial releases. Once it is implemented the downlink AMC will receive from the downlink power control functionality the power control information that will be used, together with other information for the definition of the MCS of the relevant UE.

In uplink, the eNodeB also controls which modulation and coding schemes are used. The main difference with the downlink is that link adaptation is not based on CQI reports, but instead the eNodeB can directly estimate the uplink data rates by channel sounding, i.e. using the sounding reference signals (SRS). In NSNs implementation, the UL link adaptation is based on the BLER after 1st transmission. Additionally, link adaptation can be switched on/off.

The uplink link adaptation interacts with the uplink power control by informing about the selected MCS and getting information about the UE Power Headroom so the uplink AMC can calculate the maximum allowed number of uplink PRBs per TTI for UL scheduling. These interactions are result drivent to reduce the signalling load in the eNodeB internal interfaces (i.e. the delivery of MCS is delivered to PC when the data transfer starts and when changes in the MCS are reported)

Modulation and coding rates are constant over the allocated frequency resources for a given user but they can change on the time domain on TTI basis (both, for uplink and downlink transmissions).

The 3GPP standard defines the following options for modulation and coding schemes: QPSK, 16QAM and 64QAM in both directions. 64 QAM is mandatory in downlink but in uplink is only mandatory for the UEs with highest category (Class 5). 64QAM in uplink is not supported in NSNs RL20.

Figure 22 - Modulation and Coding Schemes defined by 3GPP for LTE

Note that the different physical channels are allowed to use only certain modulation and coding schemes as Physical ChannelMCS

PDSCH, PMCHQPSK, 16QAM, 64QAM

PBCHQPSK

PDCCH (PCFICH, PHICH)QPSK

PUSCHQPSK, 16QAM, 64QAM

PUCCHBPSK and/or QPSK

Table 2 - Allowed MCS per Physical Channel2.4.4.1 Downlink link adaptation procedureThe link adaptation procedure is different for the downlink data channel (PDSCH) and for the downlink control channel (PDCCH): AMC for Downlink Data Channel (PDSCH)The initial modulation and coding scheme is provided by an O&M parameter and it is set as the default MCS. If dynamic DL AMC is not activated, the default MCS is always used.

In case the dynamic link adaptation is activated, HARQ retransmissions are handled differently from initial transmissions. For a HARQ retransmission the same MCS has to be used as for the corresponding initial transmission. The MCS determined for an initial transmission therefore has to be remembered as long as HARQ retransmissions are performed for the same block of data.

For an initial transmission, the DL link adaptation determines the MCS considering the available channel state information, PRB allocation as determined by the downlink scheduler and optionally, potential limitations in the available/allowed amount of data for the considered TTI if indicated by the scheduler

The average channel state is determined from the CQI information corresponding to the PRBs assigned (or considered for being assigned) by the scheduler.

The averaged CQI is mapped to an MCS level taking into account if the downlink scheduler indicates a bit rate limit (throughput). The MCS selected can not pass that limit.

It is possible to enable/disable the use of 64QAM and 16QAM modulation. This limitation also needs to be considered by the link adaptation.

If no new CQI values are received for a UE and the UE is scheduled, the MCS is calculated based on the latest available CQI information or the default MCS is used depending on how old the available CQI information is.

According to 3GPP, the same MCS is used for all resource blocks allocated to a single user in the frequency domain. However, the channel quality may be different for the resource blocks of the same UE within a TTI. In downlink, the set of resource blocks being allocated to a UE is determined step-by-step depending on the UEs channel quality as well as all the other UEs channels quality. For each allocation step the set of RB changes so the MCS for the UE needs to be calculated again. Once scheduling and link adaptation are finalized, the MCS corresponding to the exact set of RBs allocated to the UE is known and it is the same MCS for all RBs.

Note that the link adaptation step may have to be performed multiple times per scheduled UE per TTI. The available number of PRBs (derived from the system bandwidth) is an absolute upper limit for the number of runs of the link adaptation algorithm per TTI.

DL link adaptation supports transmit diversity (single stream of data transmission) and MIMO 2x2 spatial multiplexing (two data streams) as determined dynamically by MIMO Mode Control

Figure 23 - DL link adaptation procedure

AMC for Downlink Control Channel (PDCCH)Link Adaptation/AMC for downlink control channel (PDCCH) is different from that used on the data channel. It is based initially (RL09) on a payload approach (control data volume) and it is independent of the radio conditions. The coding for Broadcast, RACH and Paging (signalling information) is predefined by O&M parameters.

Link adaptation/AMC for PDCCH is cell based (not UE based as the AMC for DL data channel) and it has the scope of a single TTI whereas the AMC for DL data channel can be based on configurable CQI measurement intervals.

Furthermore, the QPSK modulation is mandatory for the downlink control channel. Hence, link adaptation/AMC for PDCCH is only based on the code rate, which can be varied from 1/12 up to 2/3 with intermediate steps of 1/6 and 1/3. The selected code rate is (nearly) equal for all UEs within the same TTI.

The payload approach has a limitation: a UE specific target of 1% BLER on PDCCH cannot be guaranteed and it is not actively controlled. Some UEs will perform better but others might perform worse depending on channel and load conditions. Especially at high system load in tight 1/1 reuse with several UEs multiplexed in the same TTI the PDCCH BLER for UEs at the cell edge might not be sufficient due to high interference load.

In NSNs RL20 there is an evolution towards a UE specific adaptive solution (non payload based).The QPSK is still mandatory in RL20 but the code rate and aggregation level is selected based on CQI reports which means a more efficient usage of the PDCCH resources based on the channel condition.2.4.4.2 Uplink link adaptation procedureUL Link adaptation is based on BLER and it works on a per UE basis: the Link adaptation adapts the optimum MCS according to UE specific link quality. It is called slow AMC because the update rates vary from 10ms-100ms and it is not linked to the UL scheduler nor to the UL power control. As a consequence, it is a robust and stable AMC solution. In addition, the UL link adaptation is also independent from the UL L1 physical measurements and their accuracy. However, to avoid cases of low and high BLER situations, the slow AMC is implemented together with an outer loop link adaptation (OLLA). In this case, two adjustable BLER thresholds are defined allowing for fast MCS downgrade/upgrade in case of long update periods and very poor or excellent radio conditions:

If BLER goes above a MAX BLER threshold (poor radio conditions) there is an Emergency Downgrade of the MCS into a lower one.

If BLER goes below a MIN BLER threshold (good radio conditions) there is a Fast Upgrade of the MCS into a higher one.

With the Emergency Downgrade and the Fast Upgrade functionalities a configurable O&M target BLER is maintained. All the UL link adaptation strategy can be configured via parameters.

Figure 24 - Emergency Downgrade (EDG) and Fast Upgrade (FUG)

The interactions with UL power control and Scheduler are result driven, i.e. to keep signalling load on eNodeB internal interfaces low, the MCS is reported at the start of data transfer and only when there are changes of MCS.

2.4.5 Connection Mobility ControlConnection Mobility Control is one of the basic functionalities of the RRM. As the name suggests, it refers to handovers. Handovers in LTE are: Lossless: Packets are forwarded from the source to the target. Network controlled: The target cell is selected by the network not by the UE. The handover control is in the e-UTRAN not in the Core network. UE-assisted: Measurements are made and reported by the UE to the network although it is the network (eNodeB) which triggers those measurements. Late path switch: Only when the handover is successful, the packet core is involved.Handover TypesHandover types are defined based on the frequency and the technology. Within LTE, the following types are defined:Intra-LTE (or Intra-RAT) handovers: Intra frequency handovers, that is, handovers within the same frequency band. There are two scenarios:

Intra-eNodeB handover (i.e. inter sector) Inter-eNodeB handover (i.e. inter site)In both cases the HO can happen via the X2 interface. or via the S1 interface (RL20).The downlink data forwarding over X2 is applied for lossless data path switching. Inter-frequency handovers that allow service continuity for LTE deployments in different frequency bands and also for LTE deployments within one frequency band but with different centre frequencies. Their implementation is in RL20.Inter-RAT handovers:

These are the handovers to other radio access technologies (RAT). The scope of these HOs is to provide continuous PS coverage. Handovers foreseen in this case are: LTE HO to/from WCDMA

LTE HO to/from GSM

LTE HO to/from HRPD (3GPP2)No inter-RAT handovers are planned for RL20. Inter-RAT handover to WCDMA is in RL30 . Remaining inter-RAT handovers including Inter-RAT handover from WCDMA to LTE are currently planned for future releases.Mobility RequirementsMobility requirements are based on 3GPP 25.915:

Optimised performance (in terms of delay and packet loss) is required for low mobile speeds 0.15 km/h

High performance for mobile speeds between 15120 km/h

Mobility maintained for mobile speeds 120.250/350/500 km/h depending on the frequency band.

Voice and other real-time services via the PS domain must have a handover performance better or equal than GERAN

RL09, RL10 and RL20 support high performance intra-system handover (single layer, intra-frequency), seamless and lossless via data forwarding over X2. There is no specific support for high mobility (250300 km/h and 350500 km/h) in RL20.

Basic handover scenario (Reference 3GPP TS36.300)Figure 25 illustrates the basic HO signalling in the case where neither the MME nor the S-GW change. The HO procedure does not involve the EPC, i.e. the preparation messages are exchanged between eNodeBs. The release of the resources at the source side after HO completion is triggered by the eNodeB.

Figure 25 - Basic HO process (Intra-MME/S-GW)

The source eNodeB contains information about the roaming (area) restrictions for the concerned UE. Based on those restrictions the source eNodeB configures the UE measurement procedures.

After receiving the Measurement Control message from the source eNodeB, the UE is triggered to send measurement Reports to the eNodeB.

The source eNodeB takes a HO decision based on the contents of those measurement reports. If the eNodeB decides to proceed with the HO then it issues a handover request message to the target eNodeB passing the necessary information to prepare the HO at the target side (e.g. the UE X2 and UE S1 signalling references that enable the eNodeB to address the source eNodeB and the EPC respectively).

The target eNodeB may perform admission control depending on the EPS (Evolved Packet System) Bearer QoS and it configures the required resources based on that bearer. After preparing the HO with L1/L2 it sends a handover request ack to the source eNodeB.

As soon as the source eNodeB receives the handover request ack the data forwarding from the source to the target eNodeB may be initiated. The target eNodeB buffers those packets.

The source eNodeB sends the handover command message to the UE that it starts synchronisation and the random access process with the target cell. The network will respond with the UL allocation and timing advance.

When the UE has accessed the target cell, it sends the handover confirm message to the target eNodeB to indicate that the procedure has been completed by the UE. The target eNodeB starts to send data to the UE.

Once the handover is completed it is necessary to involve the core to inform the UE has changed cell. This is done when the target eNodeB sends a path switch request message to the MME, which in turn, after receiving this message, sends a user plane update request to the Serving Gateway (S-GW).

The S-GW switches the downlink data path to the target eNodeB, releases the resources towards the source eNodeB and sends a user plane update response to the MME that confirms the path switch to the target eNodeB with the path switch ack message.

Finally, the target eNodeB sends the release resource message to the source eNodeB to inform the success of HO and trigger the release of resources in the source eNodeB.

2.4.5.1 Handovers within NSN implementationHandover decisions are made in the eNodeB based on DL measurements that the UE provides to the eNodeB by means of event triggered or periodic reporting. The reasons for handover supported in NSN implementation are better cell HO (A3 case) and coverage HO (A5 case).

RL 20 offer possibility for case when no intra-frequency LTE coverage is available. It is the feature LTE_423 - RRC Connection Release with Redirect (event A2). In case serving cell coverage by RSRP is low (defined by certain threshold) the eNodeB can redirect the call to other frequency layer or other RAT (note that this is not a handover).

Inter RAT handover towards WCDMA have been moved into RL30, Measurements of WCDMA are started and if proper inter-RAT neighbour cells are detected, a handover is initiated. Note that the reason for inter-RAT HO to WCDMA in this case is coverage (not better cell).

Assumptions:

Downlink measurements are sufficient for Measurement control and HO decision. UL measurements are not considered in RL20

RSRP (Reference Signal Received Power) is taken into account for the HO algorithm. Carrier RSSI and RSRQ are not yet used by NSNs HO algorithm.

Event triggered measurement report is sent by the UE according to the eNodeB measurement control message

Handover AlgorithmThe HO algorithm handles the UE in RRC_CONNECTED mode only. If the UE is in RRC_IDLE then the UE decides autonomously about cell reselection.

The HO algorithm triggers the UE to provide measurements via the Measurement Configuration. The measurement configuration consists of a list of measurement objects. According the related control parameters (thresholds, offsets, timers) which are transmitted to the UE by Measurement Configuration, the UE performs measurements and sends event based or periodic measurement results to the eNodeB. Depending on the RSRP of the serving cell different objects are measured as shown in Table 3:

Table 3 - Measurement activities in UE based on the RSRP of serving cell

Th1 and Th2 are thresholds set via O&M parameters.

Note: In 3GPP, RSRQ is also allowed to be the basis for measurement and trigger, but in RL20 only RSRP is considered.

As indicated in the table above, if RSRP of the serving cell is higher than Th1 then only the serving cell is measured. If RSRP gets worse and is between Th1 and Th2, then additionally all other cells on the same frequency band (intra-frequency) are measured.

RL20 add inter-RAT measurements if the RSRP of the serving cell is worse than threshold Th2. This is only done if there is no proper intra-frequency neighbour cell to which a handover can be triggered.

WCDMA measurements are started only if the feature HO to WCDMA is activated and licensed and if the UE capabilities indicate that UE is able to measure WCDMA.

When the RSRP of the serving cell becomes better than Th2a, then Inter-RAT measurements are de-activated. Th2a is similar to Th2, but in order to prevent ping-pong effects it is necessary that the value of Th2a is higher than Th2.

The measurement reporting events are:

Event A1: Serving becomes better than threshold

Event A2: Serving becomes worse than threshold

Event A3: Neighbour becomes amount of offset better than serving

Event A4: Neighbour becomes better than absolute threshold

Event A5: Serving becomes worse than absolute threshold1 and inter RAT neighbour becomes better than threshold2

The events are fully described in 3GPP TS 36.331, chapter 5.5.

Two events are considered in RL09: Event A5 (coverage handover) and Event A3 (Better Cell Handover). Relese RL10, 20 contains also event A2 in addition.

2.4.6 MIMO Mode ControlOne of the design requirements for LTE was to be able to achieve peak data rates of 100Mbps in downlink and 50 Mbps in uplink. Those thresholds assume two receive antennas in the terminal for the downlink capability and one transmit antenna in the terminal for the uplink capability. Therefore, in order to be able to achieve those peak throughputs in downlink it is necessary to use multiple antennas, which in case of LTE is supported at the eNodeB and at the terminal as a basic part of the specifications.

Note: NSNs eNodeB is hardware ready for MIMO although the usage of MIMO is a software feature (optional). As per the terminals, MIMO (2Tx- 2Rx) is mandatory for all LTE UE classes (except class1).

The term MIMO (Multiple Input Multiple Output) is generally used to indicate the use of multiple antennas, e.g. the maximum (marketing) peak throughput figures for LTE are 172.8 Mbps downlink and 56.8 Mbps uplink for a 20 MHz bandwidth using 2x2 MIMO in downlink (2 transmit antennas in the eNodeB and 2 receive antennas at the UE) and single stream in uplink (1 transmit antenna at the UE). It is possible to achieve higher peak data rates in downlink by using 4x4 MIMO (4 transmit antennas in the eNodeB and 4 receive antennas in the terminal). Currently 4x4 MIMO is part of RL30 (NSN implementation).

There is little consistency between different sources of information about what is meant by the term MIMO. As previously mentioned, strictly speaking, the term refers to all cases of multiple transmit and multiple receive antennas. However, in this document and being in line with NSN features, MIMO refers to the spatial multiplexing case. Transmit diversity (although using 2 transmit antennas) is not referred as MIMO but simply as transmit diversity.

NSN implementation (based on LTE specifications) supports the following multi-antenna transmission schemes in addition to single antenna transmission (refer to Section 3.3):

Transmit diversity Open loop spatial multiplexing (Open Loop MIMO) Dynamic Open Loop MIMO: PDSCH are transmitted using either Single Stream Downlink Transmit Diversity or Dual Stream MIMO with spatial multiplexing depending on radio conditions

Closed Loop spatial multiplexing (Closed Loop MIMO) in RL20.

All these cases use the Downlink Reference Signals for channel estimation. There is one downlink reference signal for each transmit antenna or, more exactly for each antenna port as defined by the 3GPP meaning that what is referred to does not necessarily correspond to a single physical antenna. An antenna port is defined by the presence of an antenna-port-specific reference signal. If identical reference signals are transmitted from several physical antennas the UE does not differentiate between them and all the antennas can be seen as a single antenna port.2.4.6.1 Transmit DiversityIn the case of two transmit antennas (antenna ports) the LTE transmit diversity is based on Space Frequency Block Coding (SFBC) which means that two consecutive modulation symbols S1 and S2 are mapped directly onto adjacent subcarriers on the first antenna port. The second antenna port transmits the swapped and transformed symbols S2* and S1*.

Figure 26 - Two antennas Transmit Diversity

As it can be seen in Figure 26 with transmit diversity, only a single data stream is transmitted through the two antenna ports. Each symbol is transmitted twice: once per antenna. The diversity is achieved by mapping the symbols into different subcarriers (frequency domain).

Transmit diversity is the only multi-antenna transmission scheme that can be applied to all physical channels (see Figure 29). However, synchronization signals are transmitted only on the 1st antenna.2.4.6.2 Spatial Multiplexing (MIMO)With spatial multiplexing, multiple streams or layers are transmitted in parallel which implies higher data rates for a given bandwidth. In general, LTE spatial multiplexing allows for the transmission of a variable number of layers up to a maximum that equals the number of antenna ports. As mentioned before, it is also known as MIMO. Since only MIMO 2x2 is part of RL10, we will only refer to the case of 2 antenna ports and 2 layers. In this case, the peak rates are double compared to 1 Tx antenna. Four antenna ports and up to four layers is also possible (MIMO 4x4) but not covered in these guidelines.

Figure 27 - Two antennas Spatial Multiplexing (MIMO)

LTE Spatial Multiplexing may operate in two different modes: open loop spatial multiplexing (open loop MIMO) or closed loop spatial multiplexing (closed loop MIMO). The difference between both modes is that in the case of closed loop MIMO there is more feedback from the UE than in the case of open loop MIMO. Closed Loop MIMO (since RL20)LTE spatial multiplexing uses codebook-based pre-coding, so for each combination of antennas (ports) and number of layers a set of pre-coder matrices are defined in the 3GPP standard. The UE decides on a suitable rank (number of layers) and the corresponding pre-coder matrix. This decision is based on measurements on the downlink reference signals of the different antennas.

First step is the mapping of the symbols of the code-words (transport blocks) into layers. In case of MIMO 2x2, two code-words are mapped into 2 antenna layers that is also the number of antenna ports.

After layer mapping, one symbol from each layer (S1 and S2 in Figure 27) is linearly combined and mapped to the antenna ports. The combining/mapping is done using a pre-coder matrix. For the case of 2 antenna ports and 2 layers there are two possible pre-coding matrices:

Figure 28 - Pre- coding matrices for two antenna ports and two layers

The terminal may report a recommended number of layers (Rank Indication, RI) and the recommended pre-coder matrix (Pre-Coder Matrix Indication, PMI) corresponding to the number of layers based on the estimates of the downlink channel condition (downlink reference signal measurements). In this way, the terminal is assisting the network in selecting a suitable pre-coder matrix and that is why it is called closed loop MIMO. If the network does not follow the terminal recommendation when transmitting to the UE then it needs to inform the terminal about the pre-coding matrix used in the next downlink transmission.

Open Loop MIMOThe open loop spatial multiplexing does not consider any pre-coder recommended fed back from the UE (PMI). Pre-coding is still done but because the pre-coding matrices vary in a pre-determined way it is not necessary for the pre-coding matrix information to be provided to the terminal. Open loop MIMO is suitable for high-mobility scenarios and cases where the additional overhead associated with closed-loop MIMO is not justifiable.

Another name for open loop MIMO is large- delay CDD. This is due to the fact that the overall pre-coding matrix introduces Cyclic Delay Diversity (CDD) in the layer domain. In the case of two layers, the first layer works as with the single antenna transmission but the CDD operation introduces a delay of certain number of samples (linear phase shift) on the second layer. The signals of both antennas are sent at the same time, so at the receiver, the frequency selectivity increases. It is a way of increasing diversity against fading.

In this case, it is called large-delay because the cyclic delay is very large (for two layers the delay is equal to half of the block length).

Spatial multiplexing (open and closed loop MIMO) only applies to the downlink data channel (see Figure 29).2.4.6.3 MIMO Mode ControlThe MIMO Mode Control Functionality, part of RRM, is provided by the eNodeB. It works only in downlink (uplink is not affected) and it switches between the above transmission modes plus the single Tx mode (1x1 Single Input Single output - SISO - or 1x2 Single Input Multiple Output - SIMO -).

Figure below shows that only the data channel can be configured to use Open or closed loop MIMO. Synch signals cannot use MIMO or Tx Diversity and the control channels can only use transmit diversity.

Figure 29 - Multiple antenna transmission modes vs. DL channels In RL09 the MIMO Mode Control affects only the DL direction. UL is always a SIMO 1x2 with Rx diversity by MRC (Maximum Ratio Combining) at the eNodeB receiver.

In downlink, the antenna mode can be:

A semi-static scheme: The selection between open loop MIMO (spatial multiplexing), transmit diversity and single Tx mode is done via an O&M parameter. Dynamic Downlink Open Loop MIMO: The selection between transmit diversity or spatial multiplexing is done dynamically for each UE depending on the radio conditions and based on O&M thresholds. The CQI and rank measurements from the UE are taken into account. The UE capabilities are also taken into account, i.e. UEs that are non MIMO capable should only use transmit diversity. This is also applicable to the previous semi-static scheme, i.e. if spatial multiplexing is commanded for the whole cell, non MIMO capable UEs are still served with transmit diversity.

Other RRM functions like Scheduler, DL PC, DL AMC, OLQC, etc. are dynamically informed (event driven) about the MIMO mode decision. The Dynamic Switching feature can be enabled/disabled by means of O&M.

In RL10 the MIMO Mode Control still affects only the DL direction. Aditionally, the antenna mode in downlink can be: Dynamic Downlink Closed Loop MIMO (RL20): The selection between transmit diversity or spatial multiplexing is done dynamically for each UE depending on the radio conditions and the reported Precoding Matrix Indicator (PMI) and based on O&M thresholds. The CQI, rank and PMI measurements from the UE are taken into account. As per the Open Loop MIMO, the UE capabilities (if UE supports MIMO or not) are taken into account.

Other RRM functions like Scheduler, DL AMC, OLQC etc. are dynamically informed (event driven) about the MIMO mode decision and the selected DL PMI. Algorithm decisions are controlled via O&M configurable thresholds. These thresholds can be configured differently for closed loop adaptive MIMO and open Loop MIMO Switch.2.4.7 Outer Link Quality Control (OLQC)OLQC is another function of the RRM only for the downlink direction. It is used to compensate for CQI measurement errors.

It controls the link adaptation in such a way that a certain target block error ratio (BLER) is achieved for the 1st transmission of a transport block. The function is needed to compensate to a certain degree for errors in the link adaptation. Some of those errors can be:

CQI estimation error of the UE

CQI reporting error

Time delay between CQI measurement and the reception of the subsequent data block

Errors due to CQI averaging of PRBs

Outer Link Quality control adds a CQI offset to the CQI reports that are provided by the UE. The corrected CQI report is provided them to the link adaptation for further processing. The CQI offset is controlled by the ACK/NACK responses for the initial transmission of each transport block provided by the DL HARQ.

2.5 Comparison with UMTS

The scope for LTE, as defined in the 3GPP standard, is to become a high-data rate, low-latency and packet-optimized radio access technology. Additionally, other aspects like the cost of installing and operating the network need to be taken into account as part of the evolution. All this implies certain changes when compared with previous UMTS:

New air interface transmission techniques (OFDMA and SC-FDMA):As seen in Section 2.3, the OFDM techniques split the available spectrum into hundreds and even thousands of orthogonal narrowband carriers. The high spectral efficiency of OFDM is increased further in LTE thanks to the possibility of using higher order modulation schemes such as 64QAM, sophisticated FEC (Forward Error Correction) schemes such as convolutional coding and turbo coding and additional ratio techniques like MIMO.

The fact that LTE has higher spectral efficiency allows operators to increase the number of customers with less cost of delivery per bit.

The radio performance improves significantly with respect to UMTS allowing for higher data rates. The (theoretical) peak data rates are 300 and 75 Mbps (20 MHz bandwidth) in downlink and uplink respectively.

HSPA is evolving by also using 64QAM and MIMO in 3GPP Release 7 (RU20: HSPA +). However, the improvements in overall performance for LTE are higher due to the use of OFDM instead of WCDMA and through more complex MIMO antenna configurations. The nature of OFDM makes it simpler to implement MIMO techniques than with WCDMA.

Frequency Domain Scheduling:In UMTS where a single wideband carrier is transmitted it is not possible to assign resources only to certain frequency parts. Therefore, every fading gap within the channel will affect the data. In LTE, thanks to the nature of OFDMA, it is possible to avoid allocating resources to those parts of the channel where the conditions are bad (fading). Within LTE, in addition to the time domain scheduling it is possible to schedule resources in the frequency domain which translates into capacity gains of up to 40%. Note that in HSPA only time domain scheduling is possible.

Bandwidth and frequency flexibility: OFDM makes it simpler to provide different channel bandwidths (ranging from 1.4 MHz to 20 MHz) versus the 5 MHz bandwidth in WCDMA based system. Higher bandwidths allow for higher cell throughputs and facilitate the spectrum re-farming so migration is easy. For the operators, LTE presents the possibility of more efficient use of the subcarriers.

Same multiple access techniques for FDD and TDD in LTE:As part of the requirements, LTE can be deployed in paired or unpaired spectrum. Therefore, it supports FDD and TDD deployments. The same requirements were part of the 3GPP specifications for UMTS but in that case it was done by using two different radio technologies: WCDMA/HSPA for FDD and TD-CDMA/TD-SCDMA for TDD. As a consequence, terminals capable of both FDD and TDD operations in UMTS are uncommon.

In case of LTE, both operations use the same radio access technology (OFDMA in downlink and SC-FDMA in uplink) and because of it, there is a considerable harmonization between FDD-LTE and TDD-LTE. Flat and packet based architecture:The Evolved Packet Core (EPC) is based on TCP/IP protocols like the majority of fixed data networks and provides DSL-like services including voice, video, rich media and messaging. The evolution to the packet based architecture allows for better interworking with other fixed and wireless networks. The flat architecture concept was explained in Section 2.2

No soft / softer HO:In case of LTE there is no need or requirement for soft HO (or macro diversity). Soft handover is applied to dedicated channels but not in shared channels (like HSDPA). Additionally, it was defined as an RNC feature in UMTS. In LTE there is no RNC, no need for Iur and no need for additional Iub capacity for multiple transmissions. These are significant changes regarding the network definition and operation.

Mobility in LTE is handled via the S1 and the X2 interfaces. Handover via X2 interface is triggered by default unless there is no X2 interface established or the eNodeB is configured to use the S1-handover. No need for neighbour planning:If the quality of the serving cell is below a configured threshold the UE starts searching periodically for candidate neighbour cells. Cells can be any combination of intra-frequency, inter-frequency and inter-RAT cells and the search is done following a defined priority. Once the candidate neighbour cells are identified they are periodically measured over a certain period of time. Note that although the UE measures and reports neighbours it is the eNodeB which makes the handover decision .

Based on the above procedure, it is clear that the UE is responsible for identifying the neighbouring cells so in LTE, mobility does not rely upon neighbour lists. Therefore, there is no need for neighbour list planning although in NSNs RL10 it will still be necessary to plan neighbours as the UE measurements are not taken into account. RL20 take UE measurement into account, but due to fact that UE GCID reporting problem, either neighbor list is needed or NetAct based colution is necessary.

Although eventually no neighbour planning will be necessary, the UE can still be provided with neighbour cell specific measurement offsets to make a neighbour appear more attractive. The UE can also be provided with RF carriers upon which to search for neighbours. More information about neighbour planning can be found in Section 6.7 Reduced UE power consumption: One of the reasons why SC-FDMA was selected in UL instead of OFDMA was its reduced Peak to Average Power Ratio (PAPR) when compared to OFDMA. This improves the power amplifier efficiency of the terminal and reduces the HW requirements. SC-FDMA technology was not available when UMTS multiple access selection was selected although first articles were just being published at the time (around 1997).3 NSN LTE Solution

3.1 Flexi eNodeB

NSNs eNodeB is the Flexi Multiradio BTS. It is called Multiradio because the same BTS HW can run different radio technologies (GSM/EDGE and/or WCDMA/HSPA and LTE) depending on the SW package in use. Additionally, based on the software, the Flexi Multiradio BTS can work in: Dedicated mode: When it operates in one radio mode at a time but it can be changed to other operational mode by software. Concurrent mode: The same Flexi BTS module operates simultaneously in two or more radio access modes.

Note: Some documentation may still refer to the Flexi eNodeB as Flexi Multimode BTS. The naming convention was changed in January 2009 from Flexi Multimode BTS to Flexi Multiradio BTS to avoid confusion: Flexi Multimode BTS term covered WCDMA and LTE technologies. Once the Flexi BTS would have also covered GSM technology in addition to WCDMA and LTE (estimated for 2010) the name would have changed to Flexi Multiradio BTS. Therefore, the term Flexi Multimode BTS is not used anymore and the generic name for a base station equipped with at least one multimode or Multiradio capable module is Flexi Multiradio BTS.

The actual Flexi Multiradio BTS HW can run two separate SW applications: LTE or WCDMA. Only the Flexi Multiradio in LTE mode is of interest in these guidelines. It is compliant with the 3GPP LTE Release 8 air interface and S1 and X2 network interfaces and it consists of: System Module, called Flexi Multimode System Module (note that for the system modules the term multimode is still used) supporting WCDMA and LTE. RF modules: The Flexi 3-sector Radio Module or Remote Radio Heads (RRH) In both cases the sites might be designed as feederless with RF units close to antenna. Such design is optimal in sense of minimum noise figure and nimumum losses of RF power as well.

Other possible optional modules are:

AC/DC Power sub-module: To provide BTS modules with 48 V DC power from AC current.

Battery backup sub-module (BBU): Used together with the AC/DC power sub-module.

Figure 30 illustrates examples of Multiradio configurations. The Multimode system module can be WCDMA or LTE depending on the software package. In the case where multiple RF modules are required, because each RF module supports a specific range of frequencies, it would also be a Multiband configuration.

Figure 30 - Example eNodeB illustration Multiradio Configurations(*) Any other Flexi 3-Sector RF Module, HW Rel. 2, on different bands (e.g. 700, 2600 MHz) is also part of the Multiradio ConfigurationFlexi Multiradio BTS can have two different applications: Feederless: The 3-sector RF module or the whole Flexi BTS is installed close to the antennas. Long antenna feeders and Mast Head Amplifiers are not needed optimising the performance in uplink and downlink. Distributed: The 3-sector RF module is located up to 20 km away from the main Flexi system module.

Receiver SensitivitiesThe minimum uplink reference sensitivity power level specified for LTE eNodeB receiver for planning purposes is given by the combination of the eNodeB noise figure (2.2 dB), the Modulation and Coding Scheme the diversity mode and the bandwidth used. Sensitivity of eNB is then visible on the power budget as per 4.10.2 and document 1] and power budget tool 25]. Table 4 shows the tentative sensitivity values for 5 to 20 MHz bandwidths under different conditions i.e. 1Rx or 2-way RX and MHA/no MHA usage. Note the sensitivity is normalized per 25 Resource Blocks bandwidth.

Table 4 - Flexi Multiradio BTS tentative sensitivity values (for 5MHz bandwidths and higher)

3.2 eNodeB Configurations

eNodeB configurations are based upon a combination of the Flexi System Module and Flexi RF Module. The Flexi System Module includes the Transport Module providing support for the X2 and S1 interfaces.

Flexi System Modules FSMD and FSME are LTE capable. FSME is supported in RL10 but FSMD is part of RL20. These system modules are marketed as multimode meaning that they can be used for either UMTS or LTE. The software loaded onto the system module determines which technology is used. Multimode system modules cannot be used for both UMTS and LTE simultaneously. The use of both technologies simultaneously is under study. The baseband processing capabilities of FSMD and FSME are summarised in Table 5.Flexi System ModuleMax. Number of Active Users

FSMD, 420 (RL20)

FSME400 (RL10),600 (RL20)

Table 5 - Summary of Flexi System Modules for LTE (BW = 10 MHz)

Flexi RF modules are designed to support specific operating bands, e.g. Flexi RF module FRGP supports the 2100 MHz band. If an operator wishes to provide coverage using multiple operating bands then multiple RF modules are required, e.g. a site providing both 2100 and 900 MHz coverage would require both FRGP and FXDA RF modules. The set of RF modules applicable to LTE are summarised in Table 6. All of these RF modules include 3 power amplifiers so only a single unit is required for a 3 sector configuration without MIMO. In addition, these RF modules are able to support each of the relevant technologies simultaneously.RF ModuleNameRelease

Flexi 3-sector RF Module 2600FRHARL09

Flexi 3-sector RF Module 800EUFRMARL10

Flexi 3-sector RF Module 2100FRGPRL10

Flexi 3- sector RF Module1.7/2.1FRIERL10

Flexi RRH 2Tx 800EUFRMBRL10

Flexi 3-sector RF Module 1800FXEARL20

Flexi 3-sector RF Module 1600FRNARL20

Flexi RRH 2Tx 1800FHEARL20

Table 6 - Summary of Flexi RF Modules for LTE (as of July 2010)

Identifying an appropriate eNodeB configuration for LTE involves considering a range of different options. These options are listed in Table 7. The choice and number of system modules depends upon the dimensioning of the baseband processing requirement. The choice of RF module depends upon the target operating band. A subsequent decision can then be taken regarding the choice of whether or not to share the LTE RF module with another technology.

Type of Flexi System ModuleFSME

Number of Flexi System Modules1, 2

Flexi RF Module (depends upon target operating band)FRIE, FRGP, FRMA, FRHA, FRMB

RF Carriers for LTE1

Number of Sectors1, 2, 3

MIMONone, 2+2

Receive DiversityNone, dual,

Channel Bandwidth 5, 10, 20 MHz

Downlink Transmit Power8, 20, 40, 60 W

Feederless DesignYes (then, no MHA required), No

MHAYes, No

Table 7 - eNodeB configuration options for LTE (RL20)

LTE is expected to be deployed using single RF carrier configurations, e.g. a single 20 MHz channel would be deployed rather than 2 10 MHz channels. This simplifies the network design in terms of frequency planning and also the requirement for inter-frequency handovers and inter-frequency cell re-selections.

It is expected that the majority of LTE macrocells will have 3 sectors and consequently can be configured using a single Flexi RF module. The basic 3 sector configuration is illustrated in Figure 31.The single RF module provides 3 transmit powers and 6 receive ports. The 6 receive ports are paired into receive and receive diversity ports for 3 sectors.

Figure 31 - Flexi RF Module configuration for a 1+1+1 eNodeB without MIMO

The configuration illustrated in Figure 31 allows a single transmit path per sector. Introducing 2+2 MIMO requires 2 transmit paths per sector and so requires a second RF module. The 3 sector configuration with 2+2 MIMO is illustrated in Figure 32. This configuration fully utilises the set of 6 transmit ports but leaves half of the receive ports unused. The remaining receive ports could be used for additional receive diversity if sufficient antenna elements are available.

Figure 32 - Flexi RF Module configuration for a 1+1+1 eNodeB with 2+2 MIMO

Selecting an appropriate channel bandwidth depends upon the spectrum available and the target maximum throughput. MIMO can be used to increase the maximum throughput if the spectrum available is limited. The basic LTE software provides support for the 1.4 MHz channel bandwidth. Larger bandwidths are optional and must be licensed.

The downlink transmit power requirement depends upon the deployment scenario. The basic LTE software provides support for 20 W of transmit power for a single sector. Additional sectors and transmit powers of 8, 40 and 60W must be licensed.

Similar to UMTS and as mentioned previously, the Flexi eNodeB site design can avoid the use of long RF feeders if the RF module can be mounted close to the antennas. This represents the feederless design and helps to avoid the requirement for Mast Head Amplifiers (MHA).

3.3 6-sector Sites The usage of higher number of sectors enables efficient spectrum utilization and commonly reduces the amount of sites necessary to reach the coverage targets.

The introduction of 6-sectors solution (i.e. six dedicated sectors all of them with their own power amplifier) implies clear benefits over the typical 3-sector case:

Higher coverage due to the usage of narrow beam antennas with higher gain than the typical 3-sector antennas with a wider beam. Tests have shown that there can be up to a 40% improvement in area coverage

Higher capacity due to more dedicated physical resources (e.g. six power amplifiers). Again, tests have shown up to 80% capacity increase per carrier although more attention should be put to control interference.

Although studies have been carried out for WCDMA the gains are related to antenna gains and site layouts so the results should also be valid for LTE. Simulations even show an 88% capacity gain for the case of LTE.

SHAPE

Figure 33 - Example of 6 sector site

The solution of 6-sector sites can be a valid for operators:

With little bandwidth available or where the spectrum licenses are expensive

With challenges to find site locations

Interested in deploying a faster coverage solution (less sites needed)

Interested in deploying a future proof network

With 6-sector sites already deployed in their existing 3G networks

6-sector sites can be deployed on network, area or cluster level. This versatility means they can be an additional solution for specific cases/areas to the traditional 3-sectors one. A network with a mix of 6 and 3-sector sites is possible. Appropriate tilting and sharp antenna usage is critical especially in the case of network upgrade when 6 sector site need to fit into existing network grid.

Although the reduction in the amount of required sites can be considerable (20-30% on network level) it is necessary to bear in mind that the cost per site is higher (around 50%) than a 3 sector site because of the necessity of more hardware and software licenses. Six sector site represents combination of two 3 sectors sites co-located together..

It is possible to build 6-sector sites with MIMO by using four 3-sector RF modules per site (Feederless solution) and one system module providing very high capacity. Support of 6 sectors and six LTE cells with optional 2Tx MIMO Flexi LTE BTS configuration is planned for RL30 (Focal Point Feature 106). Initial commercial networks are not expected to be capacity limited so, it should be possible to deploy initially 6-sector sites with no MIMO and upgrade to MIMO later on based on capacity requirements.

Figure 34 - HW requirements for 3 and 6-sector sites (w and w/o MIMO)3.4 RL09 RL10, and RL20 FeaturesFeatures are organised in groups based on the areas covered:

Radio Resource Management (RRM) and Telecom

Performance and Capacity

Transmission

BTS Site Solution

Operability

Performance Monitoring

Out of these, the RRM and Telecom are the most interesting but not the only. They are also organised according to the release in which they are/will be implemented. However, features often move between releases that is why it is always convenient to check these features availability in the latest roadmap or in Focal Point (access rights required): https://inside.nokiasiemensnetworks.com/global/MyServices/IT/Applications/Nokia_cross-usage_applications/NokiaProductProcess/Pages/Focal_Point_application.aspxIn this section, the most relevant features are presented. The actual release the feature belongs to and feature ID (in brackets) are indicated for information: The Chapter however, cannot be considered as a detailed description of the features. Reader is recommended to check LTE Enabling RL20 materials to see details under links [38] and namely the latest info in 39].

3.4.1 RRM and Telecom Fair scheduler. RL 09 [LTE45]As mentioned in the Scheduler section, the fair scheduling provides efficient usage of HW and radio resources.

Flexi Multiradio BTS supports time and frequency domain multiplexing in packet scheduling. Both are applied in uplink and in downlink. The packet scheduler applies a derivate of the proportional fair algorithm.

In downlink, the packet scheduler is channel-aware so decisions are based on the quality of the radio channel. The Flexi Multiradio supports both periodic wideband CQI (Channel Quality Indicator) reporting and aperiodic sub-band CQI reporting from each connected UE. Reporting rates are configurable via O&M parameter.

In uplink, the packet scheduler applies a random frequency allocation. Channel-aware scheduling in uplink, based on the sounding reference signal and measurements on other allocated PRBs, is planned for RL40 and it will be a SW licensed enhancement

The scheduling is a mixture between service fair (i.e. bit rate) and resource fair. The scheduler operates on resource fair manner when there are a low number of simultaneous communicating UEs and changes to service fair when the number of active data users sharing the cell capacity increases. Maximum and minimum bit rates for downlink and uplink and maximum number of UEs scheduled per TTI can be configured on cell basis. The latest, is constrained (downlink and uplink) by the bandwidth.

Common channels (e.g. system information, random access and paging) are scheduled with higher priority than dedicated channels. Open loop UL power control and DL power setting RL 09 [LTE27]As mentioned in the Power Control section, power control is necessary for the efficient air interface operation. The open loop power control is the basic power control mode in uplink. The UE adjusts its Tx power based on downlink pathloss estimate, on broadcasted parameters and on uplink transmission configuration. Uplink power control can be further enhanced with the closed loop mode (see next feature). In downlink, there is a semi-static flat power spectral density.

Uplink open loop power control is applied on PUSCH, PUCCH and on PRACH. The random access procedure applies power ramp-up when multiple access attempts are needed.

Open loop power control configuration is broadcast by the eNodeB in the system information and the related parameters can be adjusted semi-statically through O&M parameters.

In downlink, it is possible to configure a flat spectral power density to the active and scheduled radio resource. It is also possible to reduce the cell power via CELL_PWR_RED with 010dB attenuation in 0.1dB steps along with MIMOcompensation parameter. Closed loop UL power control since RL 10 [LTE28]Slow closed loop function is added to the basic open loop UL power control. It is based on received signal level and quality measures. The closed loop power control feature adapts the open loop control to operate at an optimum reception point under varying propagation and interference conditions. It enhances the UL performance and reduces the UE power consumption.

Flexi Multiradio BTS measures the received signal level and the quality in terms of SINR. For both measures, different upper and lower thresholds can be set by O&M parameters. The closed loop command is generated based on the PMI matrix. If the lower threshold for quality or signal level is not fulfilled then the command is to power up.

The closed loop commands are sent via PDCCH (Physical Downlink Control Channel) to the UE. Closed loop mode can be enabled/disabled via O&M parameter. Admission Control since RL09 [LTE20]The radio admission control (RAC) is based on the number of RRC connections and the number of connected users per cell. It keeps the eNodeB in stable operation.

RAC applies different thresholds for the maximum number of RRC connections and the maximum number of connected users per cell.

An RRC connection is established when the signalling radio bearers (SRB) have been successfully configured. A UE is considered as connected when a Radio bearer is established. In RL09, all RRC connection establishments are admitted by default. The number of RRC connections includes the connected users.

Handover related cell access can be prioritised because separate thresholds are applied by the RAC in case of handover. All RAC thresholds can be configured via O&M parameter. The maximum number of supported connections depends on the baseband hardware configuration of Flexi Multiradio BTS and on the cell configuration (e.g. cell bandwidth). RL20 admission control have been modified to copy with the fact of Dedicated Bearers. Admission Control will check the number of active bearers against threshold "maxNumActDrb". Voice is expected to take GBR with QCI=1 and such bearers are also compared against specific thereshold maxNumQci1Drb per cell.

RAC is triggered only once when the UE is accessing into a new cell (because initial access or handover). Possible later resource congestion is considered by the packet scheduler. Link adaptation by AMC (UL/DL) RL09 [LTE31]Link adaptation is the most important radio link control function for optimizing the efficiency of the air interface. It uses Adaptive Modulation and Coding (AMC) in uplink and downlink (in both cases the adaptation is done separately).

With the link adaptation the Modulation and Coding Scheme (MCS) is dynamically optimised per each UE and it is based on the quality of the radio link. The packet scheduler further allocates PRB based on the MCS selections from the link adaptation.

For downlink data, the link adaptation uses the Channel Quality Indicator (CQI) that is measured and reported by each UE. The link adaptation rate can be configured (minimum) per TTI. In case of MIMO the UE reports the CQI per Tx antenna. Additionally, the CQI reporting is also used for the Packet scheduler for channel aware and proportional fair scheduling.

In uplink, the link adaptation is based on BLER. For more information, refer to the Uplink link adaptation procedure Link adaptation for PDCCH since RL10 [LTE749]This is the UE specific fast link adaptation for the PDCCH selects of the code rate rsp. the channel control elements (CCE) aggregation level based on wideband CQI reports

The link adaptation works on TTI level and the link adaptation takes the available PDCCH resources into account. The following code rates rsp. CCE aggregation levels are supported: 1/12, 1/6, 1/3 and 2/3 rsp. 8 CCEs, 4 CCEs, 2 CCEs and 1 CCE.The functionality can be enabled/disabled by O&M parameter.

Intra and inter eNodeB handover with X2. since RL09 [LTE53]Flexi Multiradio BTS supports intra-LTE handover via X2. There are two intra-LTE handover scenarios: intra-eNodeB handover (i.e. inter-sector) and inter-eNodeB handover (i.e. inter-site) both of them are intra frequency type handovers. RL20 brings additional possibility to make inter eNodeB handover via S1 interface [LTE_54].The serving eNodeB configures measurements in each RRC connected UE. HO is event triggered and the related parameters can be configured via O&M parameters. The UE initiates the neighbour cell measurements when the serving cell goes below a configurable threshold. These thresholds can be configured for coverage based handover or for best cell based handover. The intra-LTE handover is triggered by event A3 (more details in section 2.4.5.1).

The serving eNodeB evaluates the handover reports and when the handover decision has been taken the selected target cell is prepared for handover execution, either within the current serving eNodeB or over the X2 interface in the adjacent target eNodeB. The handover decision is based on downlink measurements for RSRP (reference signal received power).

The Flexi MultiRadio supports temporary downlink user data forwarding over X2 during the inter-eNodeB handover execution phase. Uplink data is transported directly towards the S-GW.

Whereas the intra-eNodeB is invisible to the Evolved Packet Core (EPC), the inter-eNodeB handover requires the EPC involvement (the target eNodeB commands the EPC for Path Switch from the source eNodeB to the target).

Handover cases may be prioritized over other access cases by the Admission Control functionality (RAC). The priority is based on operator configurable thresholds.

Figure 35 - Inter-eNodeB handover Inter RAT handover to WCDMA. RL30 [LTE56]This feature refers to the coverage based inter-RAT PS handover from LTE to WCDMA that allows for continuous PS coverage in LTE rollout phase. This handover type is only applicable to Multiradio devices supporting both, LTE and WCDMA technologies.

The eNodeB broadcasts Inter RAT cell selection information and signals dedicated measurement control for connected UE. The initiation of inter- RAT measurements is based on RSRP level of the serving cell and they are only activated if there is not intra-frequency neighbour cell available.

The handover decision is based on the signal level threshold for the best WCDMA cell. In case of an unknown target WCDMA cell the next best and known target WCDMA cell may be selected if above the threshold.

The eNodeB initiates the handover preparation via the EPC and sends the handover command to the UE after receiving the handover preparation response which carries radio resource allocation for the target cell. After the handover execution the eNodeB releases the dedicated resources.

Figure 36 - Inter-RAT handover to WCDMANote regarding other inter-RAT HOs are subject of future releses: Inter RAT handover from WCDMA to LTE

Inter RAT handover to/from GSM

Inter RAT handover to HRPD/3GPP2

Inter RAT handover from HRPD/3GPP2 Single Tx path mode. RL09 [LTE187]The Tx signal is transmitted through a single Tx antenna per cell so there is no spatial diversity in the eNodeB.

The reason for transmitting only through a single Tx antenna can be either due to Hardware configuration (only one Tx path per cell) or to a configuration with two Tx paths per cell but with the second path disabled. This second option is intended for trialling purposes so 1 Rx can be compared with MIMO Spatial Multiplexing and/or Tx Diversity.

In uplink, the Flexi Multiradio BTS supports always 2Rx paths per cell. Transmit Diversity for two antennas. RL09 [LTE69]A single data stream is transmitted through two diversity antennas per sector. Space-Frequency Block Code (SFBC) mode is applied. Transmit diversity mode is applicable for most physical downlink channels with the following exceptions:

Synchronization signals are transmitted only via the first Tx antenna

eNodeB shall send different cell-specific reference signals per antenna

Operator can enable the semi-static transmit diversity mode on cell basis by O&M parameters.

More information on transmit diversity can be found in section 2.4.6.1 Transmit Diversity Downlink adaptive open loop MIMO for two antennas. RL09 [LTE70]The feature allows dynamic switching between transmit diversity (SFBC) and open loop spatial multiplexing with two code words and large CDD on UE basis. Spatial multiplexing is applied only for the PDSCH.

The adaptive algorithm provides the gain of high peak rates (when using two code words) and good cell edge performance when using single code word (transmit diversity).

The open loop dynamic MIMO switch functionality can be enabled/ disabled on a cell level by means of O&M parameters. When the dynamic MIMO switch is disabled either static spatial multiplexing or static transmit diversity can be selected for the whole cell (all UEs).

Dynamic switching takes into account the UE specific link quality and rank information. For more information about open loop MIMO refer to 2.4.6.2 Spatial MultiplexingExtensions for RL10:

To consider the UE radio capabilities.

Additionnal offset for CQI reporting compensation (for dynamic MIMO switching functionality).

Support of a performance counter for the transmission mode usage per cell.

Figure 37 - Adaptive open/closed loop MIMO Downlink adaptive closed loop MIMO for two antennas. RL20 [LTE703]The feature allows dynamic switching between transmit diversity (transmission with one code word) and closed loop spatial multiplexing with two code words in downlink on UE basis. Zero delay CDD is supported. Pre-coding is done according to the codebook described in TS36.211.

The adaptive algorithm provides high peak rates (when using two code words) and good cell edge performance when using single code word (transmit diversity).

The switching point between the two transmission modes is operator configurable per cell. The spatial multiplexing and transmit diversity functionalities can be switched on/off. A performance counter for the transmission mode usage is supported. Transmit diversity and CQI adaptation need to be supported/ enabled.Figure 37 also shows the concept for closed loop MIMO. The difference between open and closed loop MIMO is the amount of information reported by the UE. For more information about closed loop MIMO see 2.4.6.2 Spatial Multiplexing CQI adaptation (DL) RL09 [LTE30]CQI adaptation is the way to control the link adaptation behaviour and to remedy UE measurement errors. It is related to the Outer Link Quality Control functional block of the RRM. The CQI reported by the UE is adjusted with an offset before the CQI report is applied to the fast link adaptation in downlink.

The offset is calculated with the outer loop link adaptation function (OLLA) and is based on the BLER measurement for downlink data transport. The target BLER after 1st HARQ transmission can be configured via O&M parameter. The CQI adaptation corrects errors due to the variation of radio conditions or to the UE CQI measurement implementation.

3.4.2 Performance and Capacity Downlink peak data rates since RL10 [LTE 693]RL20 shall support under lab conditions 150.751 Mbps net PHY peak throughput for 20 MHz bandwidth with 2x2 MIMO and a code rate of 0.9. Marketing figure of 172.8 Mbps is theoretical and cannot be demonstrated as it would require code rate 1 which is not defined in 3GPP.

In the case of 10 MHz bandwidth the maximum peak data rate that shall be demonstrated under lab conditions is 72.790 Mbps instead of 86.4 Mbps theoretical (marketing). Reason is the code rate, same as above.

Note that at RLC level, these figures are reduced by the size of RLC/MAC headers. Uplink peak data rates since RL10 [LTE694] RL20 shall support under lab conditions 40.576 Mbps net PHY peak throughput for 20 MHz with single PUCCH pair, no sounding, PRACH every 20ms and code rate 0.83. Marketing, theoretical, figure of 56.8 Mbps cannot be demonstrated as code rate 1 is not defined in 3GPP.

In case of 10 MHz bandwidth, the uplink peak data rate that shall be supported in RL20 is 20.616 Mbps net PHY peak with PUCCH pair, no sounding, PRACH every 20ms and code rate 0.83. Marketing, theoretical figure of 28.4 Mbps can not be demonstrated as code rate 1 is not defined in 3GPP.

Note that at RLC level, these figures are reduced by the size of RLC/MAC headers. User-Plane Latencies (since RL09 and RL10) [LTE696] 3GPP 25.913 requires one-way user plane latencies 360 cells,

Let have 1000 UE/Cell and 5Pages/UE/hour ---> 1.800.000 pages/hour

So Paging load is 1.800.000/3.600 = 500 pages/second which is the number one should start to think how to split/redesign the TA6.10 MME Area PlanningNSN Recommendations MME planning should account for the equipment capability in terms of maximum subscribers, maximum bearers, maximum transaction rate and maximum eNodeB. The initial plan should not fully load each MME but should allow some margin for traffic growth.

MME boundaries should be planned to avoid areas of dense traffic.

MME boundaries should be planned to minimise the requirement for inter-MME mobility.BackgroundMobility Management Engines (MME) are control plane nodes within the Evolved Packet Core (EPC). They are responsible for control plane traffic handling, session management, mobility management and paging. MME planning involves identifying the group of eNodeB to be parented by each MME. The capability of the NSN Flexi Network Server MME is presented in Table 25.Maximum Subscribers2 000 000

Maximum Bearers5 000 000

Maximum Transaction Rate7 400 per second

Maximum eNodeB20 000

Table 25 - MME capability for NSN Flexi Network Server release 1.0 CD4

A relatively large number of eNodeB can be connected to each MME. The MME requirement is likely to be determined by the number of subscribers. This means that MME planning must account for the distribution of subscribers. The number of subscribers within an area should be estimated and then associated with a specific MME. The initial plan should not fully load each MME but should allow some margin for traffic growth.

MME boundaries should be planned to avoid areas of dense traffic. They should also be planned to minimise the requirement for inter-MME mobility, e.g. if possible, avoid planning an MME boundary across a motorway.

6.11 PRACH PlanningNSN Recommendations Unlike in WCDMA where peambles are just randomly taken from a pool of preambles in LTE it is necessary the radio planner selects the preamble format for each cell.

There are four possible preamble formats. Selection of one or the other is based on the maximum estimated cell range.

Typical preamble format will be Preamble Format 0, allowing for cell sizes up to 15km. Other preamble formats allow for larger cell ranges.

Preamble format is broadcasted as part of the system information (SIB2).BackgroundRandom access is a basic procedure within cellular technologies so the terminal establishes uplink synchronization and starts the uplink transmission. In LTE, the random access is used for several purposes, some of them:

Initial access when establishing a radio link (RRC_IDLE to RRC_CONNECTED)

To re-establish a radio link after a radio link failure

During the handover process to establish uplink synchronization to the target cell

Random access process is described in Figure 75. Before starting the random acess process the UE has already adquired downlink synchronization. The UE selects randomly a preamble from a list of parameters broadcasted in BCCH (SIB2) and transmits it in the PRACH with an initial power result of a basic downlink pathloss estimation. If the random access atempt is unsuccesful (there is no answer from the eNB) the UE makes a retry with higher power level. However, in LTE there is no need for several preamble transmissions attempts as in WCDMA since preamble and data transmissions from other terminals in the cell are orthogonal (lower intracell interference) and the initial power can be set such that the first random access is likely to be successful which has a positive impact in the delay.

Upon the reception of the preamble the eNB calculates the power adjustment for initial UE transmission, the delay between eNB and the UE and assigns a UL capacity grant so the UE can send the information.

Figure 75 - Overview of Random Access Procedure

The PRACH only carries preambles (no additional signalling or user data like in WCDMA Rel.99). In each cell there are 64 preamble sequences available. They are grouped in subsets and the set of sequences within each subset is signalled as part of the system information. The terminal randomly selects one sequence in one of the subsets during the random access process. The subset from which the preamble sequence is selected depends on the amount of data the terminal would like to transmit on PUSCH. Therefore, based on the preamble the eNodeB has some orientation about the amount of uplink resources to be granted to the terminal.

The preamble consists of two parts: the preamble sequence and the cyclic prefi