Training Objectives Understanding the LTE architecture Understanding OFDM, the key technology of LTE and the LTE physical layer structure Understanding the function of LTE protocol stack L2 and the scheduling algorithm Mastering the quick calculation of peak throughput Understanding L3
Page1
Outline LTE Architecture E-UTRAN and SAE Architecture E-UTRAN Protocol Stack Structure
LTE Physical Layer Introduction to OFDM Concept of LTE Physical Layer
Introduction to LTE L2 Introduction to Downlink and Uplink Scheduling Algorithms Downlink and Uplink Scheduling Procedures Throughput Calculation
Introduction to LTE L3 and Its Service Procedure Page 2
LTE Architecture: Logical Architecture of E-UTRAN
Page 3
LTE Architecture: SAE Architecture UTRAN SGSN HSS
GERAN S3 S1-MME
S6a
MME
PCRF S11 S10
"LTE-Uu"
S4 Serving Gateway
E-UTRAN
UE
S12
S1-U
S5
Rx
Gx PDN Gateway
SGi
Operator's IP Services (e.g. IMS, PSS etc.)
The control plane and the user plane are separate The S1 interface uses (GTP-U). The QoS architecture remains unchanged The interface between the MME and Serving GW is standardized Page 4
LTE Protocol Stack–S1 Interface User Plane
Control Plane Radio Network Layer
S1-AP
Radio Network Layer
GTP-U
SCTP UDP
Transport NetworkL ayer
IP Data link layer Physical layer
Transport Network Layer
IP Data link layer Physical layer
S1AP: The S1 Application Protocol is the application layer protocol between eNodeB and MME. SCTP: The Stream Control Transmission Protocol ensures the delivery of signaling messages on the S1 interface between the MME and the eNodeB. For details about SCTP, see RFC2960. GTP-U: The GPRS Tunneling Protocol–User plane is used for user data transmission between the eNdoeB and S-GW. Page 5 User Datagram Protocol is used for the user data transmission. For details UDP:
LTE Protocol Stack–X2 Interface Radio Network Layer
Control Plane X2-AP
Transport Network Layer
Transport Network User Plane
User Plane User Plane PDUs Transport Network User Plane
GTP-U SCTP IP (IPv6 and/or IPv4) Data link layer Physical layer
UDP IP (IPv6 and/or IPv4) Data link layer Physical layer
The X2 interface is also divided into the user plane (X2-U) and control plane (X2-C). The X2-U interface is required to be the same as the S1-U, and the X2-C is required to be the same as S1-C.
The X2 interface data link layer can use layer 2 technologies, such as PPP and Ethernet.
Page 6
LTE Protocol Stack–Uu Interface UE
eNB
MME
NAS
L3 & NAS
L2
L1
UE
eNB
RRC
PDCP
PDCP
PDCP
RLC
RLC
RLC
RLC
MAC
MAC
MAC
MAC
PHY
PHY
PHY
PHY
RRC protocol layer UE has 2 RRC states. PDCP protocol layer RLC protocol layer MAC protocol layer PHY protocol layer
Page 7
RRC
PDCP
User Plane
NAS
Introduce later
Control Plane
RRC_IDLE: A UE is in RRC_IDLE state when the UE
does not have an RRC connection. RRC_CONNECTED: A UE is in RRC_CONNECTED state when at least one RRC connection is established for the UE.
Outline LTE Architecture E-UTRAN and SAE Architecture E-UTRAN Protocol Stack Structure LTE Physical Layer Key Technology of LTE: Introduction to OFDM Concept of LTE Physical Layer Introduction to LTE L2 Introduction to Downlink and Uplink Scheduling Algorithms Downlink and Uplink Scheduling Procedures Throughput Calculation Introduction to LTE L3 and Its Service Procedure
Page 8
LTE Physical Layer
OFDM
Page 9
The Reason for Using Orthogonal Frequency Division Multiplexing (OFDM) Main Motivation: to Increase Spectral Efficiency The higher-rate data transmission, shorter transmission period. The multipath effect exists results in performance deterioration.. OFDM achieves parallel transmission by using orthogonal subcarrier group to strengthen the robustness against inter-bit interference.
Core Principle OFDM transforms serial transmission into parallel transmission, modulate data to orthogonal subcarriers.
Advantages of the Technology
High spectral efficiency Resistance to multipath interference Resistance to frequency-selective fading Simple channel estimation and equalization
Page 10
Feature of OFDM: Multi-carrier
OFDM is a type of multi-carrier modulation, with small spacings between carriers: delta-f=15 KHz
OFDM subcarriers
Page 11
Advantages of OFDM: High Spectral Efficiency
High Spectral Efficiency OFDM needs less guard bands to protect subcarriers. the spectral efficiency is increased.
The subcarriers in the OFDM system are overlapped and orthogonal. The each subcarrier is a SINC function.
140
120
Linear Amplitude
100
80
60
40
20
0 -10
-8
Page 12
-6
-4
-2
0
2
4
6
spectrum of
Advantages of OFDM: Resistance to Multipath Interference
Resistance to Multipath Interference A guard interval is inserted in front of or behind each OFDM symbol to reduce the inter-symbol interference caused by multipath delay.
In the receiver window, the multipath interference of the signal in blue can be reduced. The signal in red is longer than the guard CP, so the multipath interference exists.
Page 13
Advantages of OFDM: Resistance to Frequency-Selective Fading Resistance to Frequency-Selective Fading
The following figure shows the frequency-selective fading features of the multipath radio channel. The fading is caused by multipath.
The OFDM system is resistant to frequency-selective fading by using dynamic subcarrier allocation. No data is transmitted on fading subcarriers or low-level modulation scheme is used. (advantages and disadvantages of the measuring/scheduling algorithm of different manufacturers)
Channel fading for user A is bigger than user B. The spectral resources can be allocated to user B.
10 5
Frequency Slective Fading
0 -5 -10 -15 -20 -25 -30
Page 14
0
100
200
300 Frequency
400
500
600
Waveform of OFDM Signals in Frequency and Time Features of OFDM Signals in Frequency and Time
The smallest granularity in the time domain is the symbol. The smallest granularity in the frequency domain is the subcarrier.
Page 15
The LTE frame structure is based on symbols and subcarriers.
Outline LTE Architecture E-UTRAN and SAE Architecture E-UTRAN Protocol Stack Structure Key Technology of LTE Key technology of LTE: Introduction to OFDM Concept of LTE Physical Layer Introduction to LTE Power Calculation Introduction to LTE L2 Introduction to Downlink and Uplink Scheduling Algorithms Downlink and Uplink Scheduling Procedures Throughput Calculation Introduction to LTE L3 and Its Service Procedure Page 16
LTE Frame Structure: Focus on FDD Focus on FDD Frame Structure
A radio frame is 10 ms long and consists of 10 subframes in both downlink and uplink in FDD system. A subframe is 1 ms long and is the smallest transmission time interval (TTI) in scheduling algorithm. A slot is 0.5 ms long and is generally mentioned in frequency hopping.
Frame structure type 1 for FDD One radio frame, Tf = 307200Ts=10 ms
One slot, Tslot = 15360Ts = 0.5 ms #0
#1
One subframe
Page 17
#2
#3
#18
#19
FDD LTE Frame Structure: the Smallest Resource Unit (RE) Focus on FDD Frame Structure
1 RE = 1 symbol × 1 subcarrier (See waveform of OFDM signals in frequency and time.) 1 RB = 1 slot × 12 subcarriers – –
One RB occupies 0.5 ms in the time domain. The general definition of RB is 1 PRB = 2 slots × 12 subcarriers. If we do not take frequency hopping into consideration, slot 0 and slot 1 are allocated together, generally 1 PRB = 1 RB pair. Usually, the difference between PRB and RB in the time domain is ignored.
The following table shows the number of symbols in 1 slot. It is related to the CP length and is configured by the cell.
1slot Nsymb
Configuration
N scRB f 15 kHz
12
Normal Cyclic Prefix Extended Cyclic Prefix
12
Frame Structure Type 1
Frame Structure Type 2
7
9
6
8
Extended CP is generally used in cells with extended coverage. Page 18
FDD LTE Frame Structure: the Smallest Resource Unit (RE)
Page 19
LTE Physical Layer
Major Physical Channels
Page 20
Brief Introduction of Physical Channels Downlink Channels：
Physical Broadcast Channel (PBCH): Carries system information for cell search, such as cell ID. Physical Downlink Control Channel (PDCCH) : Carries the resource allocation of PCH and DL-SCH, and Hybrid ARQ information. Physical Downlink Shared Channel (PDSCH) : Carries the downlink user data. Physical Control Format Indicator Channel (PCFICH) : Carriers information of the OFDM symbols number used for the PDCCH. Physical Hybrid ARQ Indicator Channel (PHICH) : Carries Hybrid ARQ ACK/NACK in response to uplink transmissions. Physical Multicast Channel (PMCH) : Carries the multicast information.
BCH
MCH
PCH
DL-SCH
MAC Layer Physical Layer
PBCH
PMCH
PDSCH
RACH
Uplink Transport channels
Uplink Channels：
Page 21
PDCCH
Downlink Physical channels
Mapping between downlink transport channels and downlink physical channels
UL-SCH
Physical Random Access Channel (PRACH) : Carries the random access preamble. Physical Uplink Shared Channel (PUSCH) : Carries the uplink user data. Physical Uplink Control Channel (PUCCH) : Carries the HARQ ACK/NACK, Scheduling Request (SR) and Channel Quality Indicator (CQI), etc.
Downlink Transport channels
MAC Layer Physical Layer
PUSCH
PRACH
PUCCH
Uplink Physical channels
Mapping between uplink transport channels and downlink physical channels
LTE Physical Channels Physical channels are divided into uplink and downlink physical channels.
Downlink physical channels include: – PDSCH: physical downlink shared channel – PDCCH: physical downlink control channel – PBCH: physical broadcast channel – PHICH: physical hybrid ARQ indicator channel – PCFICH: physical control format indicator channel Uplink physical channels include: – – –
PRACH: physical random access channel PUSCH: physical uplink shared channel PUCCH: physical uplink control channel
Logical channels indicate the type of information transferred. Transport channels describe what typical configuration the physical layer uses to provide transport services on the air interface.
Page 22
Physical channels describe the physical features of signals,
Introduction to Main LTE Physical Channels Uplink physical channels include: – – –
PRACH：physical random access channel. This channel transmits uplink random access preambles. PUSCH: physical uplink shared channel. This channel transmits user data. PUCCH：physical uplink control channel. This channel transmits information about downlink data demodulation performance (ACK/NACK), channel quality measurement results and scheduling requests.
Amplification of 1 RB
PUCCH locates on the two sides of the frequency, related to the number of users in the cell , expands dynamically. PUCCH number is limited by the product specification. PRACH can be configured by using MML The remaining resources can all be used as PUSCH. PUSCH transmits uplink modulated at the UE side. the PUSCH RBs allocated to a user must be continuous in the frequency domain (RBs on two sides of PRACH cannot be allocated to the same user).
Page 23
Introduction to Main LTE Physical Channels
Downlink physical channels include: – PDSCH: physical downlink shared channel, transmits user data. – PDCCH: physical downlink control channel, indicates user scheduling information (uplink and downlink). • The eNodeB controls the time to schedule UEs, the RB allocation, and the MCS selection. • The eNodeB informs the UE of the information on the PDCCH. – –
–
PBCH: physical broadcast channel, broadcasts system information. • The location is defined by the protocol, which is known to both the UE and eNodeB. PHICH: physical HARQ indicator channel, indicates the demodulation status on the PUSCH (ACK/NACK). • Uplink data is sent by the UE and demodulated by the eNodeB. If an error has occurred in data transmission, the eNodeB needs to provide a feedback on the error. • After the UE receives ACK/NACK, it uses the feedback information to further estimate the channel quality. • The eNodeB sends ACK/NACK to the UE on the PHICH. PCFICH: physical control format indicator channel, indicates the number of symbols PDCCH resources occupy in the time domain. • PDCCH occupies all the resources in the frequency domain. • PCFICH is always at the first symbol in the time domain, occupying some of the REs of the first symbol.
Page 24
Introduction to Main LTE Physical Channels
Amplification of 1 RB
Every RB has SRS REs (the figure shows the port 0 of 2T antenna). Features of SRS REs include: 1. SRS REs is related to antennas. 2. The SRS is unique to each cell(PCI). 3. The power of SRS is configured by using MML and broadcast in system information. 4. SRSs are used to estimate and measure the downlink channel quality.
Control channels include PCFICH (at the first symbol), PHICH, and PDCCH. The number of symbols occupied by control channels is changeable (1–4). Four is only suitable for small bandwidths such as 3 MHz, 1.4 MHz.
Page 25
LTE Physical Layer
Cell Search and Random Access
Page 26
Synchronization Channel: Cell Search and Downlink Synchronization Primary-synchronization channel (P-SCH) and secondary-synchronization channel (S-SCH) frequency domain: The PSS and SSS time domain: The period is 5 ms.
are transmitted in the central six RBs.
504
physical cell IDs (PCIs) are divided into 168 groups. Each group has three PCIs. SSS mapping to PCI group. (PCI = Group ID × 3 + Cell Id in group) The UE gets synchronized in the time domain and frequency domain Reference Signal
The UE obtains the cell RSRP by measuring downlink reference signals.
Slot
Slot
0
1
2
PCFICH PHICH PDCCH
Page 27
PBCH
PSS
SSS
3
4
5
6
7
8
9
Broadcast Channel: PBCH and System Information Acquisition
System information scheduling
MIB: The scheduling period is 40 ms. The MIB is resent every 10 ms at subframe 0.
SIB1: The scheduling period is 80 ms. The SIB1 is resent every 20 ms at subframe 5.
Other SIBs: The scheduling period depends on SIBx period(x=2,3…8) and the scheduling period is broadcast in SIB1.
SIBs with the same scheduling period can be sent in the same SI. Each SI window can send only one SI. SI can be resent for multiple times in the SI window to improve reliability.
20 ms
80 ms
MIB
SI1 period = 80 ms SI2 period = 80 ms SI3 period = 160 ms SI Window = 20 ms
SIB1
SI1
SI2
SI3
SI1
SI2
Slot
Slot
0
PCFICH PHICH PDCCH
1
PBCH
2
PSS
3
4
5
6
7
SSS
Page 28
8
9
Contents of System Information MIB: An
MIB contains SFN (8 bits), cell bandwidth, and PHICH
configuration parameters. SIB1:
PLMN ID, Cell ID, TAC, Cell barred, cell selection
parameters, SI scheduling information SIB2: Access
parameters, UE timer and common channel
parameter configuration (RACH, PRACH, BCCH, PCCH, The first three PDSCH, PUCCH, PUSCH, SRS…) are key SIBs. SIB3: cell reselection information
Page 29
SIB4:
intra-frequency neighboring cell information
SIB5:
inter-frequency neighboring cell information
SIB6:
UMTS neighboring cell information
SIB7:
GSM neighboring cell information
SIB8:
CDMA neighboring cell information
PRACH Time-and-Frequency Resources
The CP length varies with the cell radius. For details, see 3GPP TS 36.211.
PRACH configuration principle
The period and preamble format of PRACH resources depend on cell radius and bandwidth. The configuration compromises between PRACH resources and access delay and handover delay.
PRACH configuration
System frame number
0
Even
1
1
Even
4
2
Even
7
3
Any
1
4
Any
4
5
Any
7
6
Any
1, 6
7
Any
2 ,7
8
Any
3, 8
9
Any
1, 4, 7
10
Any
2, 5, 8
11
Any
3, 6, 9
12
Any
0, 2, 4, 6, 8
13
Any
1, 3, 5, 7, 9
14
Any
0, 1, 2, 3, 4, 5, 6, 7, 8, 9
15
Even
9
Configurations that Huawei support:
Preamble format: 0-3 (The CP length varies with the cell radius.) PRACH period: 10 ms, 5 ms (Different bandwidths have different default configurations. For example, the PRACH period for the 20 MHz bandwidth is 5 ms and that for the 10 MHz bandwidth is 10 ms.) Random access procedure: contention-based (handover, resynchronization initiated by eNodeB), contention-free
Page 30
Subframe number
Time resources (The subframe configuration in every radio frame for random access is automatically calculated by the eNodeB based on the cell bandwidth.)
PRACH: Random Access
Purpose: – –
A user accesses the network and acquires a UE ID used by the eNodeB to identify the UE. The UE is time-synchronized in the uplink.
Procedure: – – – –
The eNodeB detects the random access preamble sent from the UE. Based on the selection of preamble, the access comes in two forms: contention-free and contention-based. In contention-free random access, preamble is allocated by the network. In contention-based random access, it is chosen randomly by the UE. In contention-free random access, the network ensures that no random access collision exists in a period of time. In contention-based random access, the risk of collision exists and the eNodeB needs to provide a contention resolution.
Random Access Scenarios Initial access: The UE is not timesynchronized yet and no information about the UE exists in the network, for example, when the UE is just switched on. The UE must be timeand-frequency synchronized before sending any data. Handover between cells The UE sends a TMSI.
The UE is out of synchronization in the uplink. An error has occurred in the downlink and the UE re-establishes the link.
The UE determines whether it is chosen based on its own TMSI.
Page 31
The eNodeB transmits the TMSI to the UEs.
If two UEs send their TMSIs simultaneously, the eNodeB needs to choose a UE to connect.
Uplink Synchronization in Random Access –
Uplink synchronization means the time when data reaches the eNodeB is the same as that when the eNodeB receives it.
–
LTE uses orthogonal subcarriers. If UE 1 and UE 2 are not time-synchronized with the eNodeB, interference exists.
–
Uplink time deviation exists because of transmission latency. The distances between the UEs and the eNodeB are different.
Resolutions:
–
Generally, the eNodeB obtains the timing information by detecting the uplink reference signal (periodic SRS or DMRS) sent by the UE in the uplink.
–
The eNodeB transmits the TA (Time Alignment) to the UE on the PDSCH.
–
•
In random access, the eNodeB obtains the uplink timing information by measuring preamble signals.
•
In random access, the eNodeB sends the uplink timing information to the UE on the RAR channel.
The eNodeB and UE maintain the same timer to update TA to ensure that the UE keeps uplink synchronized when it is in the connected state.
Page 32
Outline LTE Architecture E-UTRAN and SAE Architecture E-UTRAN Protocol Stack Structure Key Technology of LTE Key technology of LTE: Introduction to OFDM Concept of LTE Physical Layer Introduction to LTE Power Calculation Introduction to LTE L2 Introduction to Downlink and Uplink Scheduling Algorithms Downlink and Uplink Scheduling Procedures Throughput Calculation Introduction to LTE L3 and Its Service Procedure Page 33
Power of Symbol A/B and Total Power Calculation in LTE
Ri is the reference signal (RS), used to estimate downlink channel quality.
R0
Not used for transmission on this anten 1 R0
R is the reference signal, used to estimate and measure the downlink channel quality of the UE.
R0
R0
R1
R0
Reference symbols on this antenna port
R0 l 0
R0
R1
R0 l 6
l 0
Not used for transmission on this anten
R1 1
R1
R1
l 6 l 0
Antenna port 0
Symbol B contains RS.
R1
R1
Reference symbols on this antenna port
R1 l 6 l 0
l 6
Antenna port 1
Symbol A does not contain RS.
Downlink RS power can be configured by using MML: PRS, PA and PB. These parameters can be used to calculate the cell power. The power of each RE of Symbol A and Symbol B is defined as PPDSCH_A and PPDSCH_B. PPDSCH_A = PRS + A (note: PA = A ) PPDSCH_A = PRS + B The ratio between PPDSCH_B and PPDSCH_A is acquired by checking the PB index. There are two RS REs and eight non-RS REs in Symbol B. The power of each RB is: 2 × PRS + 8 × PPDSCH_B There are no RS RE and 12 non-RS REs in Symbol A. The power of each RB is: 12 × PPDSCH_A Multiply the RB power and the number of RBs and you get the symbol power, or the cell power. The maximum cell power is the greater one of the powers of symbol A and symbol B.
Number of Antenna Ports: 1
Number of Antenna Ports: 1
For example, PRS = 18.2 dBm, PA = -3, PB = 1, with the bandwidth of 20 MHz and 100 RBs, The total power of Symbol B is calculated as follows: 18.2 - 3 + 10log(100 × 12) = 15.2 + 30.8 = 46dBm = 40 W The powers of Symbol A and Symbol B are equal.
Page 34
Outline LTE Architecture E-UTRAN and SAE Architecture E-UTRAN Protocol Stack Structure Key Technology of LTE Key technology of LTE: Introduction to OFDM Concept of LTE Physical Layer: Synchronization and Random Access Introduction to LTE L2 Introduction to LTE L2 Downlink and Uplink Scheduling Procedures Throughput Calculation and FAQ Introduction to LTE L3 and Its Service Procedure
Page 35
Location of L2 in the LTE Protocol Stack UE
eNB
MME
NAS
L3 & NAS UE L2 L1
eNB
PDCP
PDCP
RLC
RLC
RLC
RLC
MAC
MAC
MAC
MAC
PHY
PHY
PHY
PHY
RLC protocol layer Solve the size matching problem from SDU to PDU Support TM, UM and AM UM and AM RLC support SDU segmentation and concatenation. AM RLC supports ARQ. AM RLC supports the resegmentation of RLC PDU. AM RLC supports the re-detection of SDU. UM and AM RLC support in-sequence delivery. UM and AM RLC support SDU discarding based on the indication of PDCP. UM and AM RLC support the reestablishment of the RLC entity. ……
Page 36
RRC
PDCP
User Plane PDCP protocol layer
RRC
PDCP
Data transmission in the user plane, data transmission in the control plane Encryption and integrity protection (in the control plane) Header compression (in the user plane) Duplicate detection and in-sequence delivery when the RB is mapped on AM RLC to perform handover PDCP SN maintenance Time-based data discarding
NAS
Control Plane
MAC protocol layer Mapping between logical channels and transport channels Multiplexing/demultiplexing of logical channel data in the transport channel. Service amount measurement Data scheduling (between UEs and between RBs in one UE) HARQ Transport format Selection
The scheduling performance depends on the appropriate resource allocation and MCS selection that are performed on the MAC layer.
LTE Scheduling Procedure-Location of MAC in L2 Radio Bearers ROHC
ROHC
ROHC
ROHC
Security
Security
Security
Security
Segm. ARQ etc
Segm. ARQ etc
PDCP
RLC
Segm. ARQ etc
...
...
Segm. ARQ etc
Logical Channels Scheduling / Priority Handling
MAC
Multiplexing UE1
Multiplexing UEn
HARQ
HARQ Transport Channels
The main function of MAC is scheduling based on service priorities. Page 37
CCCH BCCH
PCCH
Outline LTE Architecture E-UTRAN and SAE Architecture E-UTRAN Protocol Stack Structure Key Technology of LTE Key technology of LTE: Introduction to OFDM Concept of LTE Physical Layer: Synchronization and Random Access Introduction to LTE L2 Introduction to LTE L2 Downlink and Uplink Scheduling Procedures Throughput Calculation and FAQ Introduction to LTE L3 and Its Service Procedure
Page 38
LTE L2
LTE Service Processing
Page 39
Definition of LTE Services
The service and radio bearer have a one-to-one relationship. The bearer information can be obtained by checking the bearer setup signaling on the S1 interface. Default bearer: Initial Context Setup Request Dedicated bearer: E-RAB Setup Request
Voice service (VoIP) Guaranteed bit rate (GBR) services use dedicated bearers. For example, as for a 10 Mbps GBR service based on the UGW rules, the eNodeB should guarantee the rate of 10 Mbit/s.
IMS is generally not used to bear data services. Non-GBR services have different priorities. They generally use default bearers, as well as dedicated bearers. Different QCIs have different priority factors. The priority has a positive correlation with the number of allocated RBs. For example, a service with QCI 7 and a service with QCI 8 have different priorities.
Page 40
Scheduling Algorithm Functions of Scheduling Algorithms:
Ensure the coverage (access/call drop). Satisfy QoS requirements by esuring fairness of the same services and differentiation of different services. Maximize the system throughput by making full use of channel status information, that is, allocate appropriate timeand-frequency resources to users.
Algorithm
Max-C/I Only the user with the best channel quality is scheduled by using this algorithm. Therefore it has ideal throughput but cannot ensure fairness or satisfy QoS requirements. Round Robin This algorithm lays an emphasis on fairness and allocates the transmission chance to each user in turn. PF/EPF Proportional fair (PF) scheduling is a compromise between fairness and throughput. Enhanced proportional fair (EPF) classifies services (not users) into GBR services and non-GBR One user Theservices. scheduling opportunity has a may have multiple services.
The scheduling opportunity has a positive correlation with the CQI. A user is scheduled when the channel quality for the user is the best.
Page 41
UE historical scheduled bit rates UE latest channel quality CQI
UE QOS info.: QOS requirement on delay、 packet loss、AMBR UE Capability
EPF algorithm
negative correlation with the amount of historical data. This ensures that every UE has an opportunity to be scheduled. Output scheduling result, sending scheduling Grant to UE
Principle of the EPF Algorithm
PF is a compromise between fairness and throughput. EPF divides services (not users) into: GBR and non-GBR. One user may have multiple services GBR services have a higher priority to use the cell resources than non-GBR services. As for the same non-GBR services (assume all the QCIs are nine), if there are N users in the cell, the resources are allocated to them equally. (The number of RBs is related to the cell bandwidth. For details, see 3GPP TS 36.211.) As for different non-GBR services (assume QCI 6 and QCI 9 have different priorities and the priority can be configured by using MML), the number of RBs has a positive correlation with the priority.
Pr i It has a positive correlation with the CQI. A user is scheduled when the channel quality is the best.
Effi Effi 1 * QCI X * QCI X * QCI X R(t) Effi * RB RB UE historical scheduled bit rates
UE latest channel quality CQI
UE QOS info.: QOS requirement on delay、 packet loss、AMBR
UE Capability
Page 42
EPF algorithm
The scheduling opportunity has a negative correlation with the amount of historical data. This ensures that every UE has an opportunity to be scheduled. Output scheduling result, sending scheduling Grant to UE
Outline LTE Architecture E-UTRAN and SAE Architecture E-UTRAN Protocol Stack Structure Key Technology of LTE Key technology of LTE: Introduction to OFDM Concept of LTE Physical Layer: Synchronization and Random Access Introduction to LTE L2 Introduction to LTE L2 Downlink and Uplink Scheduling Procedure Throughput Calculation and FAQ Introduction to LTE L3 and Its Service Procedure
Page 43
LTE L2
LTE Throughput Calculation
Page 44
Theoretical Calculation of LTE Throughput The basis for LTE throughput calculation is the TBS specified in 3GPP TS 36.213. (The horizontal axis is the number of RBs, and the vertical axis is TBS index, or MCS.) TBS indicates the amount of data that can be transmitted at a time. (The table applies to both downlink and uplink.) 29, 30, and 31 correspond to the retransmissions of different modulation modes.
I TBS 0
91
92
93
2536
2536
2600
1
3368
3368
2
4136
3
94 N
95
96
97
98
99
100
2600
2664
2664
2728
2728
2728
2792
3368
3496
3496
3496
3496
3624
3624
3624
4136
4136
4264
4264
4264
4392
4392
4392
4584
5352
5352
5352
5544
5544
5544
5736
5736
5736
5736
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
.
.
.
.
.
.
.
.
.
.
.
19
39232
39232
40576
40576
40576
40576
42368
42368
42368
43816
20
42368
42368
43816
43816
43816
45352
45352
45352
46888
46888
21
45352
46888
46888
46888
46888
48936
48936
48936
48936
51024
22
48936
48936
51024
51024
51024
51024
52752
52752
52752
55056
23
52752
52752
52752
55056
55056
55056
55056
57336
57336
57336
24
55056
57336
57336
57336
57336
59256
59256
59256
61664
61664
25
57336
59256
59256
59256
61664
61664
61664
61664
63776
63776
26
66592
68808
68808
68808
71112
71112
71112
73712
73712
75376
PRB
MCS Index
Modulation Order
TBS Index
0 1 2 。 . . 19 20 21 22 23 24 25 26 27 28 29 30 31
2 2 2 . . . 6 6 6 6 6 6 6 6 6 6 2 4 6
0 1 2 . . . 17 18 19 20 21 22 23 24 25 26 reserved
Frequently asked question: Why can multiple MCSs correspond to the same modulation mode (for example, QPSK) when the number of RBs remains unchanged?
The modulation mode is related to the physical layer. The number of bits that can be transmitted on the physical layer = Modulation mode × Number of REs. As long as the number of RBs and the modulation mode remain unchanged, the number of bits also remains unchanged. But the MCS determines how many information bits (transport block size, TBS) can be transmitted. Even though the modulation mode remains unchanged, with different MCSs, the number of information bits may be different.
Page 45
Theoretical LTE Throughput It depends on cell bandwidth, MCS, MIMO mode, UE capability, and subscription information stored on the Home Subscriber Server (HSS). 1. The cell bandwidth determines the largest available frequency resources (number of RBs). 2. The MCS determines the spectral efficiency.
Cell bandwidth
Available RBs
Scheduling times
The number of scheduling times is related to the number of UEs in the cell. If there are multiple users, the number of scheduling times for each user are reduced. It is also related to the amount of data. If there is no data, there is no scheduling.
MCS
Transport block size
UE capability and QoS information
MIMO mode
Throughput (Mbit/s)
Page 46
UE Capability and Subscription Information Different UEs have different capabilities. Generally, commercial UEs are of Category 3. The UE capability limits the TBS. DL
UE Category
Maximum number
Maximum
of DL-SCH
number of bits
transport block bits
of a DL-SCH
received within a
transport block
block transmitted
TTI
received within
within a TTI
UL
UE Category
Maximum number
Support for
of bits of an UL-
64QAM in UL
SCH transport
a TTI Category 1
5160
No
Category 2
25456
No
Category 3
51024
No
75376
Category 4
51024
No
149776
Category 5
75376
Yes
Category 1
10296
10296
Category 2
51024
51024
Category 3
102048
75376
Category 4
150752
Category 5
299552
Page 47
Querying th capability on interface
UE Capability and Subscription Information According to the protocol, the total amount of data of non-GBR services cannot exceed the AMBR. If the AMBR is too small, the peak rate is limited. Or if the AMBR is set to 0, the transmission cannot be performed.
QoS and AMBR are queried on the S1 interface.
Page 48
FAQs in LTE Throughput Calculation How to calculate the throughput?
The eNodeB estimates the channel quality: The uplink channel quality is determined by measuring the SINR, and the downlink channel quality is determined by the CQI sent from the UE.
The SINR in the uplink or CQI in the downlink need to be adjusted due to the measurement inaccuracy, the difference between the channels, and the deep fading. (The eNodeB adjusts the SINR and CQI based on ACK/NACK and the distance from the target value of BLER.) The eNodeB selects the MCS based on the channel quality (The spectral efficiency depends on the MCS).
The eNodeB allocates the RBs based on the amount of data the UE needs to transmit.
Based on the MCS, number of RBs, and the index in 3GPP TS 36.213, the eNodeB determines the TBS. The throughput is the result of multiplying the TBS and the number of scheduling times (based on the PDCCH indication).
Probe and other UE log tools output data every second. Theoretically, the TBS of every millisecond should be calculated and then added together. What if it cannot be done? Actually, if the measurement is performed at a specified time point, using the average MCS and number of RBs (scheduling times are taken into consideration) provided by Probe to calculate the throughput is also accurate. Take BLER into consideration when calculating the actual throughput.
Is the downlink SINR related to CQI? Is the SINR measured by the UE related to the throughput?
It depends on the UE performance. Different UEs have different algorithms (the algorithms have experienced sufficient tests by using test UEs). The protocol does not define the mapping from the unadjusted SINR to CQI.
The throughput-versus-SINR curve is different in different networks. It depends on the network interference. The curves of SINR and THP with a large number of samples are considered as referential. There may be much difference between the actual throughput and the calculation result by measuring SINR once. The SINR is obtained by only measuring SRSs. SRSs may not reflect the channel quality accurately. For example, if the intra-eNodeB neighboring cell is open, the THP decreases quickly.
A good method is to check the IBLER. If the IBLER is less than 10%, the process of the eNodeB is correct. The low throughput is caused by the poor channel quality.
Page 49
FAQs in LTE Throughput Calculation What is the principle for multi-user scheduling? How to calculate the throughput?
The principle is differentiating between GBR services and non-GBR services. The GBR services are guaranteed bit rate services and should be given a higher priority. The non-GBR services are scheduled based on the QCI priority.
Fairness between the same non-GBR services: The same number of RBs are allocated to same services. The eNodeB estimates the number of RBs that can be allocated to each user based on the QoS information of each user and the service distribution in the cell. If the number of RBs matches the configuration using MML, the fairness is ensured.
Differentiation for non-GBR services: The RB resources allocated to different services match the configured proportions. The eNodeB estimates the number of RBs that can be allocated to each user based on the QoS message of each user and the service distribution in the cell. If the number of RBs matches the configured proportion using MML, the fairness is guaranteed.
For example, the cell QCI is configured as follows：
Assume two UEs access the cell. UE 1 has two services, QCI 6 and QCI 9. UE 2 has one service with QCI 9. With the bandwidth of 20 MHz, 98 RBs are available because the system information occupies some RBs. QCI 6 weight : QCI 9 weight = 1000 : 700 = 10 : 7 The theoretical number of RBs allocated to each UE is calculated as follows: – –
Switching from multi If the number of RBs is known, the throughput can be calculated based on the MCS of the UE channel. user to single to user Note: The number of RBs is statistical. For example, if the number of RBs in one second is 69.4, the number of RBs allocated each Number of RBs for UE 1: (QCI 6 + QCI 9)/(QCI 6 + QCI 9 + QCI 9) × 98 = 69.4 Number of RBs for UE 2: QCI 9/(QCI 6 + QCI 9 + QCI 9) × 98 = 28.6
TTI is the result 69.4 divided by the number of scheduling times. Throughput = TBS (MCS, statistical number of RBs/scheduling times) × Number of scheduling times
Page 50
Outline LTE Architecture E-UTRAN and SAE Architecture E-UTRAN Protocol Stack Structure Key Technology of LTE Key technology of LTE: Introduction to OFDM Concept of LTE Physical Layer: Synchronization and Random Access Introduction to LTE L2 Introduction to LTE L2 Downlink and Uplink Scheduling Procedure Throughput Calculation Introduction to LTE L3 and Its Service Procedure Not emphasized here because many specialized lectures will be held later.
Page 51
Introduction to LTE L3 and Interfaces Functions of the RRC interface (signaling over the air interface between the UE and eNodeB)
System information broadcast(important, including cell resource management)
RRC connection management: paging, setup/reconfiguration/release of the RRC connections
Functions of the X2 interface (signaling between the eNodeBs)
(important, including resource allocation)
AS (access layer/air interface) security management: encryption/integrity
protection configuration
Air interface bearer management: setup/reconfiguration/release of the user
QoS parameter mapping
Radio link failure management
Measurement control and report
Handover (intra-frequency/inter-frequency/inter-RAT)
X2 mobility management: handover signaling forwarding over the X2 interface Load management: signaling exchange between cells to balance cell loads, for example, information exchange in the inter-cell interference coordination (ICIC) X2 interface management
Functions of S1 interface (S1 signaling between the SAE and eNodeB)
Bearer management: The MME initiates the setup, modification and release of an SAE bearer. The eNodeB can also initiate the release of an SAE bearer. Handover signaling forwarding over the S1 interface Status forwarding: If in-sequence delivery and duplicate detection are supported when intra-RAT handover happens,the PDCP SN is forwarded from the source eNodeB to the target eNodeB. S1 paging: enabling the EPC to page a UE. S1 interface management: This function includes S1 interface resetting (ensuring the initialization of the S1 interface) and error indication (if no message is available for error indication, it reports/deals with the error). Load management: This function includes overload indication (indicating the load status in the control plane of S1 interface), load balancing (ensuring the load balancing in the MME pool), and S1 setup (initializing the S1 interface and providing configuration information). NAS signaling transmission between the UE and MME Releasing the UE context: This function manages the UE context release between the eNodeB and MME Indicating UE capacity: This function provides the UE capacity information
Page 52
Network Access Procedure for a Calling UE UE
E-NODEB
MME
RRC CONN SETUP REQ
RRC CONN SETUP RRC CONN SETUP CMP INITIAL UE MESSAGE
Direct transmission (authentication, negotiation) INITIAL UE CONTEXT SETUP REQ RRC SECURITY MODE CMD RRC SECURITY MODE CMP
RRC CONN RECFG RRC CONN RECFG CMP INITIAL UE CONTEXT SETUP RSP
Direct transmission (negotiation, procedure information) SAEB SETUP REQ RRC CONN RECFG RRC CONN RECFG CMP SAEB SETUP RSP
Page 53
Network Access Procedure for a Called UE UE
E-NODEB
MME PAGING
RRC PAGING RRC CONN SETUP REQ
RRC CONN SETUP RRC CONN SETUP CMP INITIAL UE MESSAGE
Direct transmission (authentication, negotiation) INITIAL UE CONTEXT SETUP REQ RRC SECURITY MODE CMD RRC SECURITY MODE CMP
RRC CONN RECFG RRC CONN RECFG CMP INITIAL UE CONTEXT SETUP RSP
Direct transmission (negotiation, procedure information) SAEB SETUP REQ RRC CONN RECFG RRC CONN RECFG CMP SAEB SETUP RSP
Page 54
Features of LTE Architecture Flat, less NEs and layers
In terms of the access network, the number of NEs is reduced, with all the functions concentrated on the eNodeB.
In terms of the EPC, the MME performs mobility management and the user plane entity (UPE) performs data management in the user plane.
Spectral efficiency is increased due to the broadband
E-UTRAN uses OFDM in the downlink and SC-FDMA in the uplink. Compared with single carrier systems, multicarrier technology has a higher spectral efficiency because multiple users occupy orthogonal subcarriers and guard-bands are not needed.
The multi-antenna technique can improve the system performance. Parallel transmission in the downlink is possible due to the MIMO spatial multiplexing.
Application of LTE
Physical layer and scheduling: OFDM, SC-FDMA, MIMO, HARQ, and scheduling algorithms can improve the sensitivity of a receiver and the capacity and coverage area of a system.
Compared with WCDMA, the bandwidth and spectral efficiency of LTE is increased and the platform techniques are increased.
Operators may have multiple networks when deploying E-UTRAN in the future. Then a question arises that how to deploy EUTRAN with multiple modes and multi-carriers: The solution lies in idle mode management and mobility management in connected mode, for example, camping and reselection principle, handover principle, and load balance principle.
LTE has the self-organizing network (SON) function. SON can perform automatic configuration and optimization. For example, neighboring relationship management, automatic handover optimization and load optimization....
Page 56
Core of OFDM: Analog Baseband Realization (For Reference) Core Principle
The OFDM transforms serial transmission into parallel transmission to get multiple parallel data streams and modulates data to orthogonal subcarriers. The spectra of subcarriers overlap.
Features of baseband signals
The following formula represents a transmit signal. N 1
X (t ) (an cos nt bn sin nt ) n 0
When the signals are demodulated at the receiving side, the orthogonality of subcarriers are guaranteed according to the following formulas.
1 T0 Ts cos(2f i t ) sin(2f j t )dt 0, i, j T 0 Ts 1 T0 Ts cos(2f i t ) cos(2f j t )dt 0, i j Ts T0 Note: They are baseband signals, not modulated to the RF signals yet.
Page 57
Core of OFDM: Digital Baseband (For Reference) Core Principle
The OFDM transforms serial transmission into parallel transmission to get multiple parallel data streams and modulates data to orthogonal subcarriers. The spectra of subcarriers are overlapped.
Digital Signal Realization
The following formula represents that sending signals are transformed into real numbers and modulated to complex carriers.
N 1 X ( t ) ( a n cos n t bn sin n t ) n 0
N 1 Re ( an jbn ) exp( j n t ) n 0
The following formula represents that the sending signals are sampled based on Ts/N.
N 1 n mTs X (m) Re (an jbn ) exp( j 2 ) T N s n 0 mn N 1 Re ( an jbn ) exp( j 2 ) N n 0 Re IDFT ( Z n )
It becomes IFFT format after conversion.
Note: Digital baseband signals are mentioned here. We can perform the quick calculation by using digits to realize the OFDM. Page 58
OFDM: Figure of Digital Baseband Realization (For Reference) Core Principle
The OFDM transforms serial transmission into parallel transmission to get multiple parallel data streams and modulates data to orthogonal subcarriers. The spectra of subcarriers are overlapped.
The following figure shows the process of superposing subcarriers using IFFT at the transmit side in an OFDM system: According to the previous formulas, the result of IFFT of transmit sequence is just the data sampling sequence when the analog baseband is realized and multicarriers are superposed. After D/A conversion, the analog baseband transmit signals can be obtained. Demodulation is a procedure in a reverse direction.
sn , 0
sn , 0
e j 0 t
sn, N 1
sn, N 1
e j N 1t
Page 59
S
IDFT
P/S
D/A
OFDM: Figure of Digital Baseband Realization (For Reference) Core Principle
The OFDM transforms serial transmission into parallel transmission to get multiple parallel data streams and modulates data to orthogonal subcarriers. The spectra of subcarriers are overlapped.
The following figure shows the rough process from data transmission to data demodulation. Demodulation is a reverse process of modulation and can be deduced based on the same method (FFT).
s(t)
S/P
IFFT
Add Cyclic Prefix
P/S
Transmitter Channel n(t)
Receiver r(t)
Page 60
P/S
FFT
Remove Cyclic Prefix
S/P
OFDM: Waveform of Signals in Frequency and Time (For Reference) Core Principle
The OFDM transforms serial transmission into parallel transmission to get multiple parallel data streams and modulates data to orthogonal subcarriers. The spectra of subcarriers are overlapped.
Signals in the time domain: Multiple carrier signals overlap in the time domain. Spectrum: Spectral intervals are the same and they overlap.
Page 61
Relationship Between the Protocols in the Physical Layer
To/From Higher Layers
36.212 Multiplexing and channel coding
Page 62
36.211
36.213
36.214
Physical Channels and Modulation
Physical layer procedures
Physical layer – Measurements
LTE Scheduling Process - Resource Allocation on the PDCCH The PDCCH indicates the downlink scheduling information, RB resources and MCS.
The PDCCH actually indicates the PDSCH or PUSCH resource information and MCS.
No other channels are used to indicate the PDCCH resource information for a UE.
How does the UE get to know its PDCCH resource information? Blind demodulation
Indications seem to be endless? PDCCH has to perform blind demodulation.
Conditions of blind demodulation: The UE attempts to decode its PDCCH message on the PDCCH (the maximum number of blind decoding attempts is 44).
1 CCE =9REG =36RE We consider the CCE as logical resources and do not discuss the mapping from the CCE to RE.
Features of PDCCH:
PDCCH occupies all the resources in the cell in the frequency domain. In the time domain, it occupies 1 to 3 (large bandwidth) or 2 to 4 (small bandwidth, 1.4 MHz) symbols. The granularity of PDCCH resource allocation is CCE. Only 1, 2, 4, or 8 CCEs can be allocated to PDCCH.
Page 63
LTE Scheduling Procedure - Blind Demodulation of The PDCCH Physical number of resources are limited.
Blind demodulation of PDCCH
attempts is limited.
The aggregation level of the UE is limited (only 1, 2, 4, or 8 according to the protocol).
According to the protocol, the number of demodulation times (or possible resource starting positions) is limited.
PDCCH resources are physical resources (CCE or RE resources) and they bear digital information (bit stream or DCI). The UE can know its RB resources and MCS only by reading the digital information on the PDCCH. The bit streams are unchangeable according to the protocol.
What does the bit information of DCI contain? (DCI 2A is used as an example.) Field Resource allocation: type 0 or type 1
Number of Bits
Allocation granularity of CCE resources
Bearer information streams are unchangeable.
Remark
1/0
Resource allocation indication: RGB allocation of type 0 or RGB subset allocation of type 1. RBG subset is left justified or right justified in type 1. HARQ process
3
TPC command, controlled by the PUCCH
2
DAI TDD parameters Codeword exchange information in dual1 codeword transmission MCS
5
NDI
1
Redundancy version Precoding information (only for 4 antennas)
2
Page 64
0 (2 antennas)
Same for the two codewords bit×2
Public space is the PDCCH that needs to be monitored by a type of UEs. The UEs receive RAR and information about dynamic scheduling of SI and paging messages. The aggregation level is limited to 4 or 8.
UE-specific space is the PDCCH that needs to be monitored by one UE, such as downlink and uplink scheduling information.
LTE Scheduling - HARQ Retransmission Procedure The purpose of hybrid automatic repeat request (HARQ) is to ensure the correct data transmission in L1. ACK/NACK determines whether retransmission is required. Forward error correction (Redundant bits are added in the procedure of encoding.) Combination of initally transmitted data and retransmitted data (If the information is not demodulated after one attempt, more bits are transmitted. The demodulation may be successful when initially transmitted data and retransmitted data are combined together.)
Features of LTE HARQ: N-channel stop and wait (N = 8) The unit of initial transmission and retransmission is TB. (The size of TBs and amount of information in initial transmission and retransmission are the same. But the bit streams may be different.) Asynchronous adaptive HARQ is used in the downlink. –
The UE sends ACK/NACK on the PUCCH or PUSCH. ACK indicates retransmission is not needed and NACK indicates retransmission is needed.
–
The initial transmission/retransmission HARQ process number must be indicated. Retransmission must use PDCCH scheduling. Same or different bit streams can be transmitted at a time (version number). There are four version numbers in LTE system but they have the same amount of information.
In the uplink –
The higher layer configures the maximum number of retransmissions for each UE.
–
The eNodeB sends ACK/NACK to the UE on the PHICH. ACK indicates retransmission is not required and NACK indicates retransmission is required.
–
There are two modes of HARQ: non-adaptive (The eNodeB does not need to send PDCCH to UE. It only sends NACK to UE and UE retransmits the data at the initial transmission position by default.) and adaptive. (The eNodeB needs to send PDCCH to UE. The position of retransmission may be different from the initial one.)
Page 65
Stop-And-Wait HARQ If N is 1, it takes a lot of time to wait. NAC K
Retransmission By using N-channel stop and wait, more information is transmitted. If an error occurred, retransmission is performed.
Each FDD UE has eight HARQ channels. Generally, HARQ channel T0 transmits data. T0+4 receives ACK/NACK and T0+8 sends other data (or retransmits data) .
Page 66
Adaptive HARQ and Non-adaptive HARQ Data is retransmitted at the position indicated by HARQ process ID 0 without the indication from the eNodeB.
The eNodeB needs to tell the UE the HARQ process number.
N-channel stop and wait (in the figure, N=8, T=0-7)
The unit of initial transmission and retransmission is TB. (The size of TBs and amount of information in initial transmission and retransmission are the same. So the initial transmitted data and retransmitted data can be combined together.)
Asynchronous adaptive HARQ is used in the downlink.
–
The UE sends ACK/NACK on the PUCCH or PUSCH. ACK indicates retransmission is not required and NACK indicates retransmission is required.
–
The eNodeB indicates initial transmission or retransmission HARQ process number. Retransmission must use PDCCH scheduling. Same or different bit streams can be transmitted at a time (version number). There are four version numbers in LTE system but they have the same amount of information.
Adaptive HARQ or non-adaptive HARQ can both be used in the uplink. –
The higher layer configures the maximum numbers of retransmissions for each UE. The eNodeB sends ACK/NACK to the UE on PHICH. ACK indicates retransmission is not needed and NACK indicates retransmission is needed.
–
Non-adaptive: The eNodeB does not need to send PDCCH to the UE. It only sends NACK to the UE and the UE retransmits data at the initial transmission position by default after 8 TTIs.
–
Adaptive: The eNodeB needs to send PDCCH to the UE. The position of retransmission may be different from the initial
Page 67 one.
