3.5 Hsdpa

Published on 5 days ago | Categories: Documents | Downloads: 0 | Comments: 0 | Views: 98
of x
Download PDF   Embed   Report

Comments

Content

 

HSDPA: S Sh hifting Gears Into 3.5G

http://archive.evaluationengineering.com/archive/articles/0205/0205hs...

EE February 2005 Feature Article  

HSDPA: Shifting Gears Into 3.5G by Marta Iglesias, Agilent Technologies

Incorporating HSDPA in Release 5 of the 3GPP W-CDMA specification is the most significant change on the RF side since Release 99 five years ago.

Just when the quest for bandwidth is accelerating competition competition among wireless technologies, W-CDMA appears to have hit a speed bump. W-CDMA technology, which provides the radio interface in the 3G UMTS mobile mobi le system defined by the 3GPP, theoretically can deliver peak data rates up to 2.4 Mb/s. In actual ac tual networks, though, the average data throughput rate reportedly doesn’t go much beyond 384 kb/s. Release 5 of the 3GPP W-CDMA specification adds HSDPA technology in an effort to make the system more efficient efficient for bandwidth-i bandwidth-intensive ntensive data applications. A W-CDMA network upgraded to HSDPA will support downlink downli nk data rates well over 2 Mb/s, up to t o a theoretical t heoretical 14 Mb/s. Because the backwardsfor compatibl compatible e with Release andnew data technology applications appli cationsisdeveloped W-CDMA still3GPP can be run on 99, t hevoice the upgraded networks, and the same radio channel will support support W-CDMA and HSDPA services simultaneously. Although industry predictions regarding the ultimate performance of  Although HSDPA vary, it likely will increase W-CDMA downlink speeds by a factor  of five, double the network capacity, and support a greater number of  users on the network. With these sig significant nificant improvements improvements for data, data , W-CDMA systems will be able to shift gears and move ahead to 3.5G—the enhanced performance enabled by this latest inter-generational mobile communication technology. Mobile carriers committed to W-CDMA are pushing for quick  development developm ent of the HSDPA network and UE to keep kee p them competitive with 1×EV-DO- and 1×EV-DV-based rivals. That means design and test engineers must gain a thorough understanding of the changes in channel coding and physical parameters introduced by HSDPA and the dynamic nature of the technology.

1 di 8 HSDPA: S Sh hifting Gears Into 3.5G

19/05/2009 14.36 http://archive.evaluationengineering.com/archive/articles/0205/0205hs...

 

  What Is New With HSDPA?

To improve W-CDMA system performance, HSDPA makes a number of  changes to the radio interface that mainly affect the physical and transport layers: • Shorter radio frame. • New high-speed downlink channels. • Use of 16 QAM in addition to QPSK modulation. • Code multiplexing combined with time multiplexing. • A new uplink control channel. • Fast link adaptation using AMC. • Use of HARQ. • MAC scheduling function moved to Node-B. The HSDPA radio frame, actually a subframe in the W-CDMA architecture, is 2 ms in length, equivalent equivalent to three of the currently defined  W-CDMA slots. There are five HSDPA subframes in a 10-ms W-CDMA frame. User data transmissions can be assigned to one or more physical channels for a shorter duration, allowing allowing the network to readjust its resource allocation in time as well as in the code domain. HSDPA introduces new physical channels and a new transport channel. HSDPA Two new physical channel types are added to the downlink: the HS-PDSCH HS -PDSCH that handles the payload data and the HS HS-SCC -SCCH H that carries the UE identity and channel parameters of the associated HS-PDSCH. HS-PDSCH. Also added in the downlink is a new transport channel, the HS-DSCH. HSDPA adds one uplink physical channel, the HS-DPCCH, for carrying the HARQ ACK and CQI information. With these enhancements, layer 2 (the MAC layer) can map existing existing

2 di 8 HSDPA: S Sh hifting Gears Into 3.5G

19/05/2009 14.36 http://archive.evaluationengineering.com/archive/articles/0205/0205hs...

 

logical channels DCCH and DTCH onto the HS-DSCH. Layer 1, in turn, maps the transport channel, HS-DS HS-DSCH, onto up to 15 channels (HS-PDSCH). The physical layer then creates the HS-SCCH and  HS-DPCCH to control and assist with HS-DSCH transmission. Downlink Transport Channel Coding The HS-DSCH is evolved from the DSCH introduced in W-CDMA Release 99 to enable time-multiplexing different user transmissions. To obtain higher higher data rates ra tes and greater spectral efficiency, the fast power  control and variable spreading factor of the DS DSCH CH are replaced in Release 5 by short packet pa cket size, multicode multicode operation, and techniques such as AMC

and HARQ on the HS-DSCH. The channel coding always always is 1/3 rate (for every bit that goes into the coder, three bits come out), based on the Release 99 1/3 turbo encoder. The effective code rate varies, va ries, however, depending depending on the parameters applied during the two-stage HARQ rate-matching process. During this process, the number of bits at the output of the channel coder  is matched to the total number of bits of the t he HS-PDSCH HS-PDSCH set to which the HS-DSCH is mapped. The HARQ functionality is controlled by the RV  parameters. The exact set of bits at the output depends on the number of  input bits, the number of output bits, and the RV parameters. Physical channel segmentation divides the bits among the different  physical channels when more more than one HS-PDSCH HS-PDSCH is used. used. Interleaving is done separately for each e ach physical channel. HSDPA uses QPSK modulation and, when radio conditions are good, 16QAM. Constellation rearrangement applies only to 16QAM modulation, in which two of the four bits in a symbol have a higher probability of error  than the other two t wo bits. The rearrangement occurs during retransmission retransmission and disperses the error probability equally among all the bits in the average, after the retransmission combining. An example HS-DSCH channel coding is shown in Figure 1. The coding corresponds to a fixed reference channel, cha nnel, FRC H-Set H-Set 4, that is used for  testing the UE receiver. The first rate-matching r ate-matching stage stage matches the t he number  of input bits to a virtual IR buffer. The second rate-matching rat e-matching stage stage matches the resulting number number of bits to the number of physical channel bits bits available in the HS-PDSCH set during the TTI. This stage is controlled by the RV parameters.

Figure 1. Example of Channel Coding for an HS-DSCH

The number of HS-PDSCHs and the modulation format define the number  of physical channel bits after RV selection (960 bits for QPSK modulation × 5 = 4,800 bits). The turbo-encoding code rate is fixed at 1/3, but the effective code rate is the combination combination of turbo-encoding and the rate

3 di 8 HSDPA: S Sh hifting Gears Into 3.5G  

19/05/2009 14.36 http://archive.evaluationengineering.com/archive/articles/0205/0205hs...

matching stages. The effective code rate for any HS-DSCH configuration can be calculated if the transport block size, the number of HS-PDSCHs, HS-PDSCHs, and the modulation format format are known. In this t his case, the effective code rate is 0.67 or [(3,202 + 24) bits/(960 bits × 5)]. Downlink Physical Channel Structure Of the three t hree slots within a 2-ms subframe, subframe, the first f irst slot carries critical critical information for HS-PDSCH reception, such as the channelization code set and the modulation scheme. After receiving the first slot, the UE has just one time slot for decoding the information and preparing to receive the HS-PDSCH.

The number of HS-PDSCHs or code channels that map onto a single HS-DSCH can vary dynamically between 1 and 15. OVSF codes are used. The number of multicodes multicodes and the corresponding offset for the t he HS-PDSCHs mapped from a given HS-DSCH are signaled on the HS-SCCH. The multicodes (P) at offset (O) are allocated as follows: Cch, 16, O … Cch, 16, O+P-1 The second and third slots carry the HS-DSCH channel coding information, such as the transport block size, HARQ information, the RV and constellation versions, versions, and the new data indi indicator. cator. The data of the three slots is covered with the 16-bit UE identity. identity. Uplink Physical Channel Structure The HS-DPCCH carries uplink feedback signalling related to the downlink  HS-DSCH transmission. This signalling consists of HARQ-ACK and CQI shown in Figure 2. Each 2-ms subframe, like like those of the downli downlink  nk   physical channels, consists consists of three slots, slots, each with 2,560 2,560 chips. The HARQ-ACK is carried in the first slot of the HS-DPCCH subframe and the CQI in the second and third slots.

Figure Figu re 2. High-Sp High-Spee ee d Dedicated Physical Control Channel Channel That Carr Carries ies the Uplink

There is at most one HS-DPCCH on each radio link. The HS-DPCCH can exist only in association with a W-CDMA uplink DPCCH. Two different paths are used for HARQ-ACK and CQI coding. The The HARQ-ACK information is coded to 10 bits, with ACK coded as 1 and   NACK coded as 0. The CQI CQI information information is coded using a 20,5 code. The coded bits are mapped directly to the HS-DPCCH. The feedback cycle of the CQI can c an be set as a network parameter in  predefined steps from 2 ms to infinity. infinity. An active HS-DPCCH HS-DPCCH may may have slots in which no HARQ-ACK or CQI information is transmitted. The HS-DPCCH HS -DPCCH can be a bursted channel. cha nnel.

4 di 8 HSDPA: S Sh hifting Gears Into 3.5G  

19/05/2009 14.36 http://archive.evaluationengineering.com/archive/articles/0205/0205hs...

Adaptive Modulation and Coding Link adaptation is one important way in which HSDPA improves data throughput. The technique used, AMC, matches the t he system’s modulationcoding scheme scheme to the average channel c hannel conditions during each user  transmission. The power of the transmitted signal is held constant over a subframe interval, and the modulation and coding format format are changed to match the current received r eceived signal signal quality or channel conditions.

In this scenario, users close to the BTS typically are assigned higher order  modulation modul ation with higher higher code rates, such as 16QAM with an effective effec tive code rate of 0.89, but the modulation modulation order and code rate will decrease as the distance from the BTS increases. increases. The 1/3 rate rat e turbo coding c oding is is used, and  different effective code rates are a re obtained through various rate-matching  parameters. In HSDPA, the UE reports the channel conditions to the BTS via the uplink channel CQI field in the HS-DPCCH. The CQI value can be 0 to 30, with a value of 0 indicating out of range. ra nge. Each CQI value corresponds to a certain c ertain transport block size, the number of HS-PDSCHs, HS-PDSCHs, and the modulation modul ation format for a certain UE category. These parameters are used   by the BTS in combination combination with with other parameters to determine determine the appropriate TF and effective code rate. For example, the largest transport block size is 27,952 bits, which corresponds to the hig highest hest data rate of 13.976 Mb/s (27,952 bits/2 ms = 13.976 Mb/s). This data is obtained by using 16QAM, 16QAM, an effective code rate of 0.9714, and rate 15 HS-PDSCHs. HS-PDS CHs. Hybrid ARQ HARQ is a technique combining FEC and ARQ methods that save information inform ation from previous failed attempts to be used in future decoding dec oding.. HARQ is an implicit link-adaptation technique. AMC uses explicit C/I or  similar measurements to set the modulation and coding format; HARQ utilizes link-layer acknowledgements (ACK/NACK) for retransmission decisions. decisi ons. Put another way, AMC provides the coarse data-rate data-ra te selection; HARQ accommodates accommodates fine data-rate data-rat e adjustment based on channel conditions.

For a retransmi re transmission ssion,, HARQ uses the same transport-block set and  consequently the same number of information bits as were used in the initial transmission. However, it may use a different modulation scheme  —channelization-code  —channeliz ation-code set including including the size size of the channelization-code channelization-code set—or transmission transmission power. As a result, the number of channel bits available for a retransmission may differ from that of the initial transmission. transmi ssion. Channel bits are the bits actually transmitted transmitted over the air. Even if the number of channel bits is the same, the channel-bit c hannel-bit set may be different. To minimize the number of additional retransmission requests, HARQ uses one of two soft-combining schemes schemes to ensure proper messag messagee decoding dec oding.. CC involves sending sending an identical version of an erroneously detected   packet. Received copies are combined by by the decoder prior to decoding. decoding. IR involves sending a different set of bits incrementally to be combined  with the original original set, increasing the amount of redundant data dat a and the t he likelihood lik elihood of recovering from errors introduced on the air. Use of Incremental Redundancy Figure 3 illustrates how the IR scheme works. For simplicity, an IR buffer  size of 10 bits/process and a single process are assumed in this example. The original original data (4 bits) corresponds to the data block after the CRC has  been added. The data is is encoded at the 1/3 rate, and it then gets gets punctured 

5 di 8 HSDPA: S Sh hifting Gears Into 3.5G  

19/05/2009 14.36 http://archive.evaluationengineering.com/archive/articles/0205/0205hs...

as part of the first rate-matching rate- matching stage. At this stage, the number of output  bits is is matched to the IR buffer size, size, 10 bits bits in this this case.

Figure 3. Demodulation of the HS-DSCH Using Incremental Redundancy

The second rate-matching stage stage punctures the t he data agai again. n. The data can be  punctured into different different data sets, each corresponding to a different RV. RV. These are indicated here as three different colors: red, green, green, and orange. ora nge. Only one of these data sets will be sent in any given transmission. The five red bits are sent OTA, resulting resulting in in an effective eff ective code rate of 4/5. That is, for every original data bit, 1 + 1/4 1/ 4 bits are transmi tr ansmitted tted OTA. The data arrives at the UE and is demodulated, padded with dummy bits, bits, and  stuffed into the IR I R buffer. The data then is decoded, with some possibility possibility of error, to t o provide the four blue bits. This This block is checked against the CRC, and if found to be in error, is stored, and a NACK requests a retransmission. When the retransmis retransmission sion is sent, sent, it uses a different RV or puncture scheme and sends the five green bits OTA. At the UE, the green bits are recombined with the original transmission’s red bits to provide an effective code rate of 2/5. Now for every data bit there are 2½ bits available available for  decoding,, which increases the decoding t he likelihood likelihood of success. However, when the t he results are checked against the CRC, if the block is still still in error, the retransmission process begins again. Yet another RV or puncture scheme is used, appearing now as the orange  bits that are sent OTA OTA and recombined recombined at the UE with with the red and green green  bits from the first and second transmi transmission. ssion. The new RV provides additional redundant data even e ven if some or all of the encoded bits are repetitions of encoded bits sent earlier. After the third transmission, transmission, the effective code rate is 4/15; for every data  bit, there now are 3 + 3/4 bits. bits. At last, last, the data is correctly decoded, and  an ACK is sent back. If the block were still in error, a NACK would be sent, and still more RVs could be transmitted, depending on the maximum number of transmissions allowed for a block. In the case of 16QAM formats, the different RVs not only may correspond  to different puncturing schemes, schemes, but also to different constellati constellation on versions or rearrangem rea rrangements. ents. HARQ Processes The HSDPA system does not retransmit a data block until it receives an ACK or NACK for that data. To avoid wasting time between transmission of the data block and reception re ception of the ACK/NAC ACK/NACK K response, which would  result in wasted throughput, multiple independent HARQ processes can be run in parallel.

6 di 8 HSDPA: S Sh hifting Gears Into 3.5G

19/05/2009 14.36 http://archive.evaluationengineering.com/archive/articles/0205/0205hs...

 

Five subframes nee ded to receive rec eive the ACK/NACK for a transmi t ransmitted  tted  subframes are needed data block. Since the ACK/NACK is required before data transmission for  a specific process can continue, the minimum interval between TTIs must  be at least six six for a singl singlee HARQ process. process. Six Six HARQ HARQ processes running simultaneously will completely fill every subframe with data to specific UE. UE must support a minimum inter-TTI interval of receiving data every subframe, every other subframe, or every third subframe. The minimum minimum interval value they support depends on the HS HS-DS -DSCH CH category. Packet Scheduling Functionality In addition to the channel coding and physical physical and transport layer changes, HSDPA HS DPA implements implements another change to support fast packet transfer. It relocates the packet scheduling functionality functionality from the network controller  to the MAC layer in the Node B.

The packet-scheduling algori algorithm thm takes into account the radio channel conditions,, based on the conditions t he CQI from all the UE involved, and the amount of  data to be transmitted to the t he different users. Throughput gains gains can be maximized by serving the UE that is experiencing the best radio channel conditions, but obviously some degree of fairness in scheduling is required. In addition, there are other factors that the scheduli scheduling ng algorithm algorithm could take into account, such as quality of service. The actual throughput also will depend heavily on the packet-scheduling algorithm used. Scheduling, modulation and coding adaptation, and HARQ retransmissions in HSDPA HSDPA are fast because they t hey are performed as close to the air interface as possible and because a short frame length is used. Fast scheduling makes it possible possible to track the fast-channel variations. What Comes Next? HSDPA technology is incorporated in W-CDMA Release 5 to increase data throughput and improve the efficiency e fficiency of the system for downlink  data traffic. tr affic. The main changes introduced by HSDPA are new hig high-speed  h-speed  data channels, the combination of time-division multiplexing with code-division multiplexing, the use of AMC and HARQ techniques, and  the relocation of MAC layer scheduling to the Node-B. With a thorough understanding of these changes, design and test engineers can begin to successfully implement HSDPA into network and UE.

Looking forward to Release 6, the content of which is being finalized at this time, time, the most significant significant feature targeted for the radio interface is the EUDCH. This feature will introduce techniques similar to HSDPA to improve im prove coverage, increase throughput, and reduce delay on the upli uplink  nk  this time. Release 7 will likely include MIMO antennas, which support higher hig her data rates and are ar e considered an enhancement enhance ment to HSDPA.  About the Author

 Marta Iglesias is a wireless wirele ss industry marketing marketi ng engineer at Agilent  Technologies. She holds a B.S.E.E. from the Universitat Politecnica de Catalunya, Spain. Agilent Technologies, 395 Page Mill Rd., Palo Alto, CA 94303, 800-829-4444, e-mail: [email protected]

FOR MORE INFORMATION on the 3GPP Project Plan www.rsleads.com/502ee-186

  BACK TO TOP

ARTICLE ARCHIVE

7 di 8 HSDPA: S Sh hifting Gears Into 3.5G  

19/05/2009 14.36 http://archive.evaluationengineering.com/archive/articles/0205/0205hs...

Published by EE-Evaluation EE-Evaluation Engineer ing  All con te nts © 2009 Nelson Publ ish ing Inc. No reprint , distribution, or reu se in any medium is permitted without the express written consent of the publisher.

 

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close