INSIGHT ON HSPA
Wireless evolution: Competing 3.9G systems
10/20/11
2
HSDPA Overview
15 Code Shared transmission
16QAM Modulation
TTI = 2 ms
Hybrid ARQ with incr. redundancy
Benefit Higher Downlink Peak rates: 14 Mbps Higher Capacity: +100-200% Reduced Latency: ~75 ms
Fast Link Adaptation
Advanced Scheduling
HSUPA Overview
1-4 Code Multi-Code transmission
TTI = 10 ms
Hybrid ARQ with incr. redundancy
Benefit Higher Uplink Peak rates: 2.0 Mbps Higher Capacity: +50-100% Reduced Latency: ~50-75 ms
Fast Power Control
NodeB Controlled Scheduling
HSxPA Motivation and General Principle –
Improved performance and spectral efficiency in DL and UL by introducing a shared channel principle: •
Significant enchancement with peak rates up to 14.4 Mbps (28 Mbps in Rel7) in DL, and 2 Mbps (11.5 Mbps with 16QAM) in UL
•
Huge capacity increase per site; no site pre-planning necessary
•
Improved end user experience: reduced delay/latency, high response time -A H C D H -B C DC H DC
Rel. 99 Dedicated pipe for every UE
u ed h Sc
g lin
A,
C B,
HSDPA (3GPP Rel5) Fast pipe is shared among UEs
H DC E
-A
H DC E
-B
H DC E
-C
HSUPA (3GPP Rel6)
Dedicated pipe for every UE in UL Pipe (codes and grants) changing with time E-DCH scheduling
UL DCH vs HSDPA vs HSUPA Concepts HSUPA is like “reversed HSDPA”, except HSDPA
HSUPA
Modulation
QPSK and 16-QAM
BPSK and DualBPSK
Soft handover
No
Yes
Fast power control
No
Yes
Scheduling
Point to multipoint
Non-scheduled transmission
No
Multipoint to point Yes, for minimum/ guaranteed bit rate
Efficient UE power amplifier Required for near-far avoidance Scheduling cannot be as fast as in HSDPA Similar to R99 DCH but with HARQ
HSUPA could be better described as Enhanced DCH in the uplink than “reversed HSDPA” HSUPA (E-DCH) is an uplink DCH with BTS-based HARQ and scheduling and true multicode support Feature
DCH
HSUPA
HSDPA
Variable spreading factor
Yes
Yes
No
Multicode transmission
Yes
Yes
Yes
Fast power control
Yes
Yes
No
Soft handover
Yes
Yes
Adaptive modulation
No
No
Yes
BTS based scheduling
No
Yes
Yes
Fast L1 HARQ
No
Yes
Yes
(No in practice)
No
(associated DCH only)
HSDPA – New Channels for HSDPA per UE High Speed Dedicated Physical NB
•
Control Channel
Uplink
Down link
•
SF=256
•
Carries H-ARQ ACK/NACK, Channel Quality Information (CQI)
•
It can only exist with UL DPCH
High Speed Physical Downlink Shared Channel
NB
– – – –
Data bearer: Peak data rate 14.4Mbit/s QPSK and 16 QAM can be applied SF=16 Up to 15, always associated with a DCH
High Speed Shared Control Channel – –
SF=128 Carries H-ARQ information, channelization code set, modulation scheme. Up to 4 logical channels per UE
HSDPA – Protocol Stack PS-RAB
PS-RAB
PDCP
PDCP
Iu-UP
Iu-UP
RLC
RLC
GTP-U
GTP-U
MAC-d
MAC-d
UDP
UDP
HS-DSCH -FP
IP
IP
MAC-hs
MAC-hs
PHY
UE
PHY
Uu
HS-DSCH -FP
HS-DSCH HS-DSCH -FP -FP
AAL2
AAL2
AAL2
AAL2
AAL5
AAL5
ATM
ATM
ATM
ATM
ATM
ATM
PHY
PHY
PHY
PHY
PHY
PHY
Node B
Iub
CRNC
Iur
SRNC
Iu
SGSN
HS-PDSCH Transmit power The Packet Scheduler is responsible for determining the transmission power on the HS-PDSCH channels • Dynamic HSDPA power allocation is always used in BTS – HSDPA power can be limited with PtxMaxHSDPA • HSDPA Dynamic Resource Allocation feature is activated with RNC parameter HSDPADynamicResourceAllocation – Disabled: PtxMaxHSDPA sent to BTS and used to limit the maximum HSDPA power – Enabled: No power limitation sent to BTS, all available power allocated to HSDPA Ptx Maximum HSDPA power (PtxMaxHSDPA )
Cell maximum TX power
Ptx
HSDPA HSDPA
NonHSDPA power
HSDPA power is limited by the PtxMaxHSDPA parameter
Non-HSDPA power Common chs
Common chs •
Cell maximum TX power
Time
Time
HSDPA power is not limited, all available power allocated to HSDPA Still PtxMaxHSDPA can be used to limit
–
–
HS-PDSCH Transmit Power
HSDPA dynamic power control enables use of higher HSDPA power, when DCH traffic load is low •
Cell throughput increases as the cell resources are divided more efficiently between varying mix of Rel-99 and HSDPA traffic
•
Higher average downlink power also means interference rise to DCH users
HSDPA power is reduced dynamically in the case DCH load is increasing •
–
Packet scheduler measures the unused transmission power in the cell every 2ms
The maximum limit for the HSDPA power allocation is produced with equation: PHSDPA = PtxMax – PtxnonHSDPA •
PtxMax is the maximum transmission power of the cell (PtxCellMax parameter in RNC and BTS capability/license)
•
PtxnonHSDPA is the measured transmission power of all codes not used for HS-PDSCH or HS-SCCH transmission Total HSDPA Power
Allocated HSDPA power is shared by HS-PDSCH and HS-SCCH HS-SCCH PtxHSDPA = P + P – HS − PDSCH _ tx HS − SCCH _ tx All HS-PDSCH codes (sent to one UE) have equal power (max 15 codes per UE) –
HS-PDSCH
•
E-DPDCH (E-DCH Dedicated Physical Data Channel) –
•
HSUPA – New Channels for HSUPA per UE
Data channel, for transmitting MAC-PDUs: •
Total E-DCH Buffer Status (TEBS)
•
UE Power Headroom (UPH)
NB
Uplink
E-DPCCH (E-DCH Dedicated Physical Control Channel) –
Transmitted in parallel with E-DPDCH – contains L1 control information for HARQ and scheduling: •
E-DCH Transport Format Combination Indicator (E-TFCI)
Down link
E-AGCH (E-DCH Absolute Grant Channel) – – – –
•
Retransmission Sequence Number (RSN) control HARQ
Time multiplexed shared channel using explicit addressing (E-RNTI) of UEs • Primary E-RNTI for Happy Bitsingle UE UE could use more resources or not Secondary E-RNTI for group of UEs
Grants E-DPDCH/DPCCH power ratio
E-RGCH (E-DCH Relative Grant Channel) – (Common or dedicated)
NB
– Implicit addressing by assignment of physical channel (DL code + signature) – Common channel by assignment of same physical channel to UEs for E-RGCH purposes
– Up, Down, Hold
E-HICH (E-DCH Hybrid ARQ Indication Channel) • Used for signalling ACK/NACK (Dedicated)
E-RNTI: E-DCH RNTI RNTI: Radio Network Temporary Identity
HSUPA – Protocol Stack DTCH DCCH
DCCH DTCH
MAC-d
MAC-d
MAC-es
MAC-es / MAC-e MAC-e
MAC-e EDCH FP
PHY
UE
PHY
Uu
EDCH FP
TNL
NodeB
TNL
Iub
TNL
TNL
DRNC
Iur
SRNC
HSUPA – Node B controlled Scheduling
The UE provides the BTS scheduler with (in MAC-e header): –
–
–
UE buffer occupancy: how much data is in RLC buffers
NB
Information about the priority of the data in the buffer Available transmission power resource
E-DPCCH (L1)
MAC-e PDU on E-DPDCH (L2)
NodeB Controlled Scheduling
In physical layer (E-DPCCH) the UE provides to BTS: –
E-TFI, indicating what is transmitted in the E-DPDCH
–
Information of the HARQ redundancy version for the packet •
Timing is known, thus BTS knows which ARQ channel to expect
– Happy bit:reduces Is the current data rate satisfactory Faster scheduling noise rise variations point can • UE would not be happy of the data rate if it could transmit with higherOperation rate be increased because Less headroom needed variance is reduced Cell capacity andhave userenough data rates • I.e. dataare in its buffers and would have sufficient power resource to transmit with a increased higher power than currently
A first estimated increase in the cell capacity is 15-20% for the same noise rise outage
• • • • •
HSUPA – Node B controlled Scheduling
NodeB Controlled Scheduling
Node B scheduler shares resources between UEs with HSUPA connections RNC scheduler continues to manage R99 DCH connections Similar to HSDPA scheduler in MAC-hs, HSUPA scheduler in MAC-e is faster than an RNC scheduler Both absolute and relative grants are used (E-AGCH and E-RGCH) Scheduling period is 10ms
RNC 384 256 12 864 32 16 8 Zero Grant
1.
Node B 384 256 12 864 32 16 8 Zero Grant
2.
UE 384 256 12 864 32 16 8 Zero Grant
1. 2. 3. 4.
3. 4.
RNC limits the E-TFCI based upon UE capability and QoS profile NodeB limits E-TFCI based upon packet scheduling principles UE limits E-TFCI based upon transmit power capability UE selects E-TFCI based upon data to be transferred
HSUPA – Fast Scheduling Reduces Noise Variance • Scheduler uses absolute and relative grants to maximise the utilisation of every user and minimise the difference between the requested and allocated bit rates • Scheduling decisions are based upon the – Uplink interference margin – Physical layer feedback (happy bit) – Iub capacity allocation – Available baseband processing capacity
• HSUPA scheduler combines
throughput and load based algorithms • Throughput based scheduling is applied for lower loads • Power based scheduling is applied for higher loads
Increase in interference floor (dB)
Power based scheduling Throughput based scheduling Cell Load (%)
Minimum Baseband and 16 Users per Cell Minimum baseband
4 users
6 users
6 users
1 users
–
9 users
6 users
One WSPC/FSPA supports 1 - 3 HSDPA cells in cell group
–
QPSK/16-QAM supported
–
Max 5 codes per cell
–
Max. 16 HSDPA users per WSPC/FSPA: •
16 users
Up to 3.6 Mbps is supported per WSPC/FSPA 16 users per cell –
32 CE from FSMx allocated to HSDPA scheduler One WSPC/FSPA unit per HSDPA cell is earmarked for HSDPA – Cells can be on different frequencies Max 16 HSDPA users per cell is supported Max– number percells BTS with is not limited for HSDPA Up toof 12users HSDPA 4 WSPC – Number of HSDPA cells * 16 Up to 3.6 Mbps with 16QAM is supported per cell Max 5 codes per cell Each HSDPA cell requires 32 CE from FSMx allocated to HSDPA scheduler –
16 users
16 users
HSDPA users can be divided freely between three cells in cell group
• • • • • •
Shared HSDPA Scheduler for baseband efficiency Shared HSDPA Scheduler for baseband efficiency and 48 Users per cell 6 users •
6 users
36 users
–
One WSPC/FSPA supports 1 - 3 HSDPA cells •
–
Max 48 HSDPA users per WSPC/FSPA •
48 users
The whole card is reserved for HSDPA
HSDPA users can be divided freely between three cells
–
Up to 10.8 Mbps with 16QAM is supported per WSPC/FSPA
–
Max 15 codes per cell; 45 codes per BTS
–
Possible to transmit to 3 users simultaneously
• cell Users can be from any cell mapped on the same Mac48 users per
hs entity (cell group)
48 users
48 users
• – –
One WSPC/FSPA per HSDPA cell scheduler (per BTS) 80 CEs from FSMxunit allocated to HSDPA •
–
–
–
–
With code multiplexing in same cell The whole card is reserved for HSDPA
Max 48 HSDPA users per cell Up to 14.4 Mbps with 16QAM per cell (with code multiplexing) Higher number of users allows users to be kept on cell_DCH state for longer periods Iub and BTS baseband dimensioning is possible only if the 16 kbit/s Return Channel DCH Data Rate Support for
HSDPA – Theoretical Throughput 1 code, QPSK
3,84 Mchip / s ⋅ 2bit / symb = 480kbit / s 16chip / symb
5 codes, QPSK 3,84 Mchip / s ⋅ 2bit / symb ⋅ 5codes = 2 ,4 Mbit / s 16chip / symb
15 codes, QPSK 3,84 Mchip / s ⋅ 2bit / symb ⋅15codes = 7 ,2Mbit / s 16chip / symb
15 codes, 16QAM 3,84Mchip / s ⋅ 4bit / symb ⋅15codes = 14,4Mbit / s 16chip / symb
HSUPA – Theoretical Max UE throughput calculation
Example UE CAT 3
10ms TTI & 2 E-DCH codes in parallel Minimum SF = 4 SF = 4
3,84 Mchips/s/4 = 960000 chips/s
2x E-DPDCH channels BPSK
1920000 chips/s
Symbol/Bit = 1
E-DCH physical channel has 1,92 Mbps With UE CAT 3 we have approximately net rate of 0.75 so this results in 0.75*1,92 = 1,44Mbps.
HSUPA – Theoretical Max UE throughput calculation
Example UE CAT 6
2ms TTI & 4 E-DCH codes in parallel 2 x SF = 4 and 2 x SF = 4 SF = 4
3,84 Mchips/s/4 = 960000 chips/s
2x E-DPDCH channels 2x E-DPDCH channels BPSK
1920000 chips/s with SF = 4 3840000 chips/s with SF = 2
Symbol/Bit = 1
E-DCH physical channel has (1,92 + 3,84 =)5,76 Mbps With UE CAT 6 we have approximately net rate of 1 so this results in 1*5,76 = 5,76 Mbps.
LTE enabling technologies •
Multicarrier-based radio air interface –
OFDMA and SC-FDMA
•
IP-based flat network architecture
•
Multi-input multi-output (MIMO)
•
Active interference avoidance and coordination –
•
Fractional frequency re-use (FFR)
Fast frequency-selective resource
10/20/11
21
LTE beauties
10/20/11
22
LTE Advanced Features: OFDMA/SC-FDMA 10/20/11
23
LTE PHY: Modulation / Multiple Access •
Downlink: OFDMA –
–
–
–
•
Allows simple receivers in the terminal in case of large bandwidth #subcarriers scales with bandwidth Frequency selective scheduling in DL (i.e. OFDMA) Adaptive modulation and coding (up to 64QAM)
Uplink: SC-FDMA (Single Carrier -
10/20/11
24
Time-Frequency Representation of OFDM Signal
10/20/11
25
Difference between OFDM and OFDMA
10/20/11
26
OFDM/SC-FDMA Transmitter and Receiver OFDM Transmitter and Receiver
SC-FDMA Transmitter and Receiver
10/20/11
27
SC-FDMA time/frequency domain orthogonality •
Time Division Multiple Access (TDMA): –
–
Only time-domain orthogonality Entire bandwidth assigned to one user at a time High peak data rates
Potentially in-efficient 10/20/11 for small available –
28
LTE Advanced Features: Adaptive Modulation in Time and Frequency 10/20/11 29
Constellations for BPSK, QPSK, 16-QAM and 64-QAM
10/20/11
30
Adaptive Modulation and Coding (AMC)
10/20/11
31
Adaptive modulation in frequency
10/20/11
32
LTE Advanced Features: Efficient Scheduling in Time and Frequency 10/20/11 33
LTE RRM: Scheduling 1/4 Packet scheduling in time/frequency and spatial domain Frequency
•
•
Frequency diverse scheduling at34higher
10/20/11
LTE RRM: Scheduling 2/4 •
Frequency selective scheduling at low/medium mobility –
Each UE is allocated its individual best part of the spectrum Multiuser diversity gain
Best use of the spectrum Adaptive 10/20/11 modulation & coding –
35
LTE RRM: Scheduling 3/4
10/20/11
36
LTE RRM: Scheduling 4/4 •
Trigger for UL Scheduling –
–
–
•
Buffer Status report (BSR): PUSCH scheduled user Scheduling request (SR): Synchronous user (PUCCH allocated) PRACH : Asynchronous users
Frequency Selective Scheduling –
DL: Based on CQI (Channel Quality Indicator) reported by UE
10/20/11
37
LTE PHY: Feedback channel concept •
UE: Reports the finest possible granularity –
The reporting scheme and granularity depend on the radio channel quality variation
eNB: Receives
10/20/11 •
38
LTE Advanced Features: Advanced Multi-Antenna Techniques
10/20/11
39
Multi-antenna techniques
10/20/11
40
General MIMO Principles
10/20/11
41
MIMO operation requires a priori knowledge of all channel responses 1/2
10/20/11
42
MIMO operation requires a priori knowledge of all channel responses 2/2
10/20/11
43