HSDPA Principles

Page 1Page 1

HUAWEI TECHNOLOGIES Co., Ltd. Huawei Confidential

The Overview of HSDPA(High Speed Downlink Packet Access)

Page 2Page 2


Outline

� Introduction

� HSDPA New Techniques

� Transport and Physical Channels

� Spreading, Modulation and Coding

� Protocol Architecture

� Terminal Capabilities

Page 3Page 3


HSDPA Bit Rate Advantage

0

200

400

600

800

1000

1200

1400

1600

1800

2000

GPRS EDGE WCDMA EV-DO EV-DV HSDPA

1.5 Mbps

700 kbps

400 kbps

350 kbps

150 kbps

30 kbps

kbps

� Typical average bit rate for different technologies in a medium loaded Macro cell.

Page 4Page 4


HSDPA Latency Advantage

0

100

200

300

400

500

600

700

GPRS/EDGE R99 WCDMA R99 HSDPA R5

ms

650 ms

200 ms

100 ms

� Typical average round trip time for different technologies. Smaller round trip times would benefit interactive applications.

Page 5Page 5


HSDPA Considerations

� The same carrier can be shared between WCDMA and HSPDA. It’s the DL power which should be intelligently divided between two services.

� Unlike 3GPP2 standards, EV-DO and IS-95/1xRTT, which can not share a carrier.

� An evolutionary rather than a revolutionary philosophy.

� WCDMA networks can be upgraded with HSDPA hardware/software on Node-B by Node-B basis.

� Even HSDPA features can be added gradually, if required.

� Priority to urban environments and indoor deployments.

� Support full mobility but should be optimized for low and medium speed users.

� Focus on streaming, interactive and background services.

� HSDPA new features should show significant incremental gains over existing R’99 performance.

� Consider value added to the user, cost to the operators, increased revenue for operators, etc., in adding any new feature.

Page 6Page 6


Outline

� Introduction






Page 7Page 7


Power Utilization in HSDPA

� Efficient use of power: the unused power by dedicated/common channels is exploited by HS-DSCH with use of dynamic power allocation.

Page 8Page 8


Channel Sharing in HSDPA

� Efficient use of time: several packet data streams are time multiplexed and sent over High Speed Downlink Shared Channel (HS-DSCH).

Page 9Page 9


Major New Techniques

� Fast Scheduling

� Fast Hybrid ARQ

� Fast Link Adaptation

Page 10Page 10


Fast Scheduling

Page 11Page 11


Packet Scheduler

� The packet scheduler’s task is to maximize the network throughput while satisfying the QoS requirements of the users.

� The packet scheduling method has significant impact on the cell throughput and on the user-perceived quality of service.

� The scheduler is located in the Node B and can respond quickly to channel conditions, since the Iub and the RNC are not involved in the process.

� Multi-user diversity:

� Selection of the best users in the cell in terms of the UE’s received signal strength is known as multi-user diversity.

� The scheduler may select for transmission in each TTI (Transmission Time Interval) users that have good signal to noise ratio and therefore ensure better data reception and fewer retransmissions.

� Multi-user diversity will increase the average cell throughput by using the network resources more efficiently.

TTI intervals

ChannelQuality

UE1UE2

UE1

UE2

Node B

Page 12Page 12


Scheduling Methods

� Round Robin

� Users are served in cyclic order ignoring channel conditions. It is simple and ensures a fair resource distribution among the users at the cost of cell capacity.

� Maximum C/I (example shown in next page)

� The cell serves in every TTI the user with the largest instantaneous supportable data rate. Users with lower average radio conditions receive less resources but due to large fading dynamics, these users are still able to receive service.

� Average C/I

� The cell serves in every TTI the user with the largest average C/I that has data to be transmitted. Averaging windows can be as large as 50 TTI’s. This tends to average the short term fading conditions for users.

� Proportional Fairness

� The cell serves the user with the largest relative channel quality, based on the short term data rate of the user relative to its average data throughput. Users with better short term channel conditions will have higher priority than users that are temporarily located in a fade.

� Fair Throughput

� Modifies the proportional fair algorithm to increase the priority of users that receive lower average throughput, in an attempt to equalize the throughput to all users. A variant that does not use instantaneous channel quality information and serves in every TTI the user with the lowest average throughput.

Page 13Page 13


Example of Max C/I Scheduling Method

� Example of Max C/I Scheduling

Page 14Page 14


Fast Hybrid ARQ

Page 15Page 15


Fast Retransmission

� All Release 99 (pre-HSDPA) transport channels are terminated at the RNC.

� Retransmission procedure is located in the serving RNC.

� The serving RNC (SRNC) may not be the controlling RNC (or drift RNC), and it may be several hops away from the controlling RNC, increasing response times.

� For high speed data, this potential delay is not acceptable.

� The new high speed channel, the HS-DSCH, terminates at Node B.

� A new MAC layer, the MAC-hs, is introduced in the Node B in order to control all retransmissions in the high speed data channel and provide a quick response to channel errors.

Page 16Page 16


Hybrid ARQ Types

� HARQ is an implicit link adaptation technique

� It uses ACK/NACKS to produce correct packets (implicit adaptation to channel conditions).

� It uses samples weighted by the signal to noise ratio to combine received versions of the packets, which provides time-diversity.

� There are three types of ARQ:

� Type I ARQ - a pure repetition mode, the original data block is retransmitted.

� Variants are called Chase Combining, when the data block is soft-combined with the original block and Optimum Combining, where each block is weighted by the signal to noise ratio and then combined.

� Type II ARQ - This is called Full Incremental Redundancy (FIR) combining.

� A non-self decodable retransmission is sent. This retransmission must be combined with the original block in order to decode. It consists of parity bits and does not include the original data.

� Type III ARQ - This is called Partial Incremental Redundancy (PIR)

� The retransmitted block must be self-decodable, that is, it must include the original version of the data in addition to any other redundant information.

Page 17Page 17


Fast Link Adaptation

Page 18Page 18


Link Adaptation by Transport Format

� The HS-DSCH does not use fast power control.

� The transmitted power in the HS-DSCH is substantially constant, depending on power sharing for other downlink channels in the Node-B.

� The adaptation to channel conditions is done by selection of the transport format, such as modulation and coding rate.

� The transport format may be changed every TTI (2 ms).

UE Node B

Transport Format: Modulation and Coding

Power measurement, CQI selection

CQI report every 1 to 80 TTI’s

Transport Format selection, new modulation and coding

Page 19Page 19


Link Adaptation

� The transmitter receives information on the channel conditions from the UE and selects an appropriate transport format for transmission.

� Selects QPSK or 16QAM modulation.

� Selects a specific coding rate that works well in those conditions (approximately a 10% block error rate).

� Lets the HARQ process fine-tune the coding rate by use of retransmissions to bring down the error rate.

� Link adaptation is fast, since it all happens in the Physical layer between the UE and the Node-B.

� Example of Adaptive Modulation and Coding:

Page 20Page 20


Outline

� Introduction






Page 21Page 21


Channels and Layers

Radio Link Control (RLC)

Medium Access Control (MAC)

Physical Layer

Layer 1

Layer 2

Transport Channels: How data is sent

Logical Channels: Type of data sent

Physical Channels: Media used for data

Page 22Page 22


Transport Channels

Transport Channels

BroadcastChannel (BCH)(Downlink)

Common ChannelsDedicated Channels

Downlink SharedChannel(DSCH)(Downlink)

Common PacketChannel (CPCH)(Uplink)

Forward-AccessChannel (FACH)(Downlink)

PagingChannel (PCH)(Downlink)

Random-AccessChannel (RACH)(Uplink)

Dedicated Channel (DCH)(Down & uplink)

High Speed DownlinkShared Channel(HS-DSCH)

Page 23Page 23


Downlink Data Transport Channels

� DCH- Dedicated Channel� Voice, data up to 2 Mbps� Long channel set-up times, long channel release time, not good utilization for bursty traffic� Fixed Spreading Factor (SF), code tree is reserved for maximum data rate� Fast power control� Soft handoffs

� FACH- Forward Access Channel� Common channel, no channel set-up time required� Suitable for small IP packets, infrequent packets for interactive gaming� High power channel - no feedback from UE� No fast power control, fixed SF� No soft handoffs - so cell reselection process takes time

� DSCH- Downlink Shared Channel� Paired with a DCH that carries control information� Suitable for bursty traffic, channel is shared by UE’s� Dynamic SF� Fast power control� No soft handoffs

� HS-DSCH- High Speed- Downlink Shared Channel� Supports HSDPA, associated with HS-SCCH and HS-DPCCH

Page 24Page 24


HS-DSCH Attributes (Summary)

� High Speed downlink channel to support high data rates

� Shared among UE’s

� No fast power control

� No soft handoffs

� Fixed SF (SF=16)

� Fixed CRC size (24 bits)

� Uses Adaptive Modulation and Coding (AMC) to support different data rates

� Uses Hybrid ARQ (HARQ) to provide error-free operation

� Support multi-code operation (up to 15 codes) for data rates up to 10 Mbps

� Uses Turbo code R1/3, with rate adaptation to support other coding rates

� Uses a short Transmission Time Interval (TTI) 2 msec long, consisting of 3 timeslots (TSs) in order to achieve a short round trip delay

� Possible to use beam-forming for broadcast to part of a cell, or to the entire cell

� Possible to vary transmit power

� Always associated with a DPCH and one or more DSCHs

Page 25Page 25


HS-SCCH (High Speed Shared Control Channel)

� A DL channel which carries the information required to demodulate the HS-DSCH.

� Every UE using the HS-DSCH is assigned an HS-SCCH by the Node-B.

� The HS-SCCH has a fixed rate of 60 kbps, fixed Spreading Factor (SF) of 128 and uses rate 1/3 convolutional coding.

� The HS-SCCH block consists of three Time Slots (1 TS = 667 microseconds) and is transmitted two time slots before the start of the HS-DSCH transmission in order to allow the UE to demodulate the data.

� An HS-SCCH block carries necessary information in two separate coding chains:

� Part 1 information:

� Coding information, according to the UE’s capabilities

� Modulation used for the HS-DSCH, i.e., QPSK or 16 QAM

� Part 2 information:

� Redundancy information, to allow the UE to combine with prior information

� HARQ process information

� New date indicator, whether the transmission is a first transmission or a re-transmission

� Transport block size

Page 26Page 26


HS-DPCCH (HS Dedicated Physical Control Channel)

� An uplink control channel which uses SF=256 code multiplexed with other uplink control channels.

� Channel used to provide constant feedback to the Node-B on signal quality and packet errors.

� Transmitted 5 msec after the reception of the HS-DSCH frame.

� This packet is divided into two parts:

� ACK/NACK information part - One bit long transmitted in one timeslot, provides the result of the CRC check after packet decoding and indicates whether a retransmission is required or not.

� Downlink Channel Quality Indicator (CQI) - 5 bits long transmitted in two timeslots, to indicate what block size, modulation and number of parallel codes could be received with reasonable error rate based on measured channel conditions. This information is used by the Node-B to determine the modulation and coding to be used in the next HS-DSCH transmission.

Page 27Page 27


Outline

� Introduction



� Spreading and Coding



Page 28Page 28


Channelisation & Scrambling

� Channelisation spreads the input bits or symbols into a chip sequence by a factor equal to the spreading factor. It separates the transmissions within a single source (i.e., a sector or a UE). It uses Orthogonal Variable Spreading Factor (OVSF) codes which are built by Walsh codes.

� Scrambling multiplies the already spread sequence by a second code that separates the sources of the data (i.e., different sectors in DL or different UE’s in UL). It uses Gold Codes or shorter codes.

Cell 1 Cell 2 Cell nCell 3 Cell 4

ScramblingChannelisation & Spreading

1 23

n432

1

Page 29Page 29


DL Channelization and Modulation

� Channelization codes are OVSF (Walsh) codes with SF=16 for HS-DSCH and SF=128 for HS-SCCH.

� Modulation mapper maps symbols to the I and Q branches

� For QPSK, symbols are sent alternatively to the I and Q branches

� For 16QAM, two pairs of symbols are first mapped to a constellation value and then sent to the I and Q branches alternatively

I

Downlink physical channel

Serial to Parallel

Converter

Channelization code, SF value

j

Scrambling code

Q

I+jQ S Modulation Mapper

Page 30Page 30


Outline

� Introduction






Page 31Page 31


HSDPA Protocol Architecture

UE Node B SRNC

UuAir Interface

IUb

Layer 1

MAC-hs

MAC-d

RLC

Transport

Frame Protocol

MAC-d

RLC

Layer 1

MAC-hs

Transport

Frame Protocol

Page 32Page 32


MAC Layer and MAC-hs

� The Medium Access Control (MAC) layer maps the logical channels coming from the Radio Link Control (RLC) function to the transport channels connecting it to the Physical layer.

� The MAC layer is responsible for selecting an appropriate transport format for each transport channel based on the requirements of the logical channels.

� The MAC-hs is a new MAC entity located in the Node-B (in order to be closer to the air interface) and the UE, to support the HSDPA high speed channel.

� MAC-hs is responsible for:

� Handling the new high speed shared channel, HS-DSCH.

� Managing physical resources allocated to HSDPA.

� Other MAC entities:

� MAC-d is responsible for handling the dedicated channels (DCH transport channel). It is located in the Serving RNC (SRNC).

� MAC-b handles the broadcast channel (BCH). It is located in the Node-B.

� MAC-c/sh handles the common and the shared channels, including the DSCH. It is located in the SRNC.

Page 33Page 33


MAC Architecture

MAC-d

MAC Control

MAC-c/sh MAC-hs

HS-DSCH

PCCH BCCH CCCH CTCH DCCH DTCH

PCH FACH RACH CPCH DSCH DCH Associated Downlink Associated Uplink Signaling Signaling

Page 34Page 34


Outline

� Introduction






Page 35Page 35


HSDPA Terminal Capabilities

� UE Power Class

� Power Class III is 24 dBm and Class IV is 21 dBm. Initially, most UEs fall in Class IV.

� There are UE Categories defined for HSDPA:

14.0QPSK, 16-QAM11510

0.9QPSK2511

QPSK

QPSK, 16-QAM

QPSK, 16-QAM

QPSK, 16-QAM

QPSK, 16-QAM

QPSK, 16-QAM

QPSK, 16-QAM

QPSK, 16-QAM

QPSK, 16-QAM

QPSK, 16-QAM

Modulation Scheme

1.81512

10.01159

7.21108

7.21107

3.6156

3.6155

1.8254

1.8253

1.2352

1.2351

Max data rate (Mbps)

Min. Inter-TTI interval

Max number of parallel codes

Category

Reference UE capability combinations:1.2 Mbps class 3.6 Mbps class 7.2 Mbps class 10 Mbps class

Page 36Page 36


UE Capability Classes

� Examples of UE capability classes proposed in 3GPP are listed in Table below. Note that more combinations are still possible.

1.2 Mbps class

3.6 Mbps class

7.2 Mbps class

10 Mbps class

Page 37Page 37


Thank you!

HSDPA Principles

Documents

HSDPA Principles