IEEE COMMUNICATIONS LETTERS, VOL. 10, NO. 6, JUNE 2006 459

Channel Adaptive CQI Reporting Schemes forHSDPA Systems

Soo-Yong Jeon, Student Member, IEEE, and Dong-Ho Cho, Senior Member, IEEE

Abstract HSDPA (High-Speed Downlink Packet Access) is anevolved UMTS packet scheme that delivers increased user peakdata rates and quality of service. A key technique supportingHSDPA is adaptive modulation and coding (AMC), in which themodulation scheme and the coding rate are changed adaptivelyaccording to the downlink channel quality reported by the userequipment (UE). Therefore, the channel quality indication (CQI)reporting scheme is directly related to the accuracy of AMCand the performance of HSDPA. This letter proposes channeladaptive CQI reporting schemes in which UEs report the CQIvalue intelligently by using information about channel quality.With the proposed schemes, the battery capacity of UE can beconserved and the uplink interference can be lowered by filteringoff redundant CQI reports or the transmission error rate can belowered by fast CQI reports.

Index Terms UMTS, HSDPA, AMC, CQI.

I. INTRODUCTION

HSDPA is a packet-based data service in W-CDMA andit was specified in the 3rd generation partnership project(3GPP). The main target of high-speed downlink packet access(HSDPA) is to increase user peak data rates and qualityof service, and to improve spectral efficiency for downlinkasymmetrical and bursty packet data services. HSDPA offerstheoretical peak data rates on the order of 10 Mbps over a5MHz bandwidth in WCDMA downlink. HSDPA implemen-tations include Adaptive Modulation and Coding (AMC) aswell as Hybrid Automatic Request (HARQ). With the AMCtechnique, the modulation scheme and the coding rate arechanged adaptively according to the CQI reported by the UE.So, exact channel quality estimation through a proper CQIreporting scheme and a channel quality prediction methodare essential in AMC. In [1], the channel quality predictionmethod was proposed for the reliable adaptive modulationwith limited channel quality feedbacks in a correlated fadingchannel.

In the current specification of HSDPA, each UE mustperiodically report to Node B the CQI indicating the downlinkchannel quality. The CQI value is transmitted by using a High-Speed Dedicated Physical Control Channel (HS-DPCCH). IfUEs report the CQI periodically, they must report the sameCQI value repeatedly when the channel quality is unchanged.Obviously, this is the waste of the UEs battery power. To

Manuscript received January 13, 2006. The associate editor coordinatingthe review of this letter and approving it for publication was Dr. ChristosDouligeris. This work was supported in part by the university IT researchcenter program of the government of Korea.

The authors are with the Dept. of Electrical Engineering and ComputerScience, Korea Advanced Institute of Science and Technology, Daejeon 305-701, Republic of Korea (e-mail: syjeon@comis.kaist.ac.kr).

Digital Object Identifier 10.1109/LCOMM.2006.06021.

solve this problem, a new CQI reporting scheme was proposedin [2]. In this scheme, the UE reports the CQI value to theNode B only when the value difference between new measuredchannel quality and last reported one becomes larger than deltavalue which is a certain threshold decided by higher layer.With this scheme, the uplink signalling overhead is reducedefficiently. However, if the Node B receives the CQI valuewith error, it has incorrect channel quality information for along time until it receives next CQI value.

In this paper, we propose channel adaptive CQI reportingschemes motivated by the same idea as above scheme. Ourschemes behave intelligently to prevent the Node B fromkeeping incorrect channel quality information for a long timeby using a timer and an efficient CQI report timing criterion.Thus, with the proposed schemes, either the number of redun-dant CQI reports is decreased, maintaining the performance ofa periodic CQI reporting scheme, or the accuracy of AMC isincreased by reducing the probability of transmission error.

The rest of this letter is organized as follows: Section II in-troduces the CQI reporting schemes in the 3GPP specificationand section III explains the proposed CQI reporting schemes.In section IV, we describe the simulation environments andevaluate the performance of the proposed schemes. The lastsection makes conclusions.

II. 3GPP-SPECIFIED CQI REPORTING SCHEMESIn the specification Release 5 of HSDPA, a periodic CQI

reporting scheme is included. The definition of the CQIand UE procedure for reporting CQI are described in [3].According to the specification, each UE measures downlinkchannel quality and selects suitable CQI values to indicatetransport block size, number of High-speed Physical DownlinkShared Channel (HS-PDSCH) codes and modulation schemes.The selected CQI values must be such that the transport blockerror probability would not exceed 0.1 under the measureddownlink channel condition. After the selection of suitableCQI values, the UE reports the selected CQI value to NodeB at its own CQI report timing. The CQI report timing isdetermined by each UEs connection frame number (CFN)given by the radio link establishment procedure. According tothe specification [4], the report cycle of CQI is defined as 0, 1,2, 4, 5, 10, 20, 40, 80 subframes, where the subframe lengthis 2ms. The report cycle of CQI is informed to each UE byhigher-layer signalling.

Clearly, the short report cycle of CQI gives better through-put performance than the long report cycle of CQI, sincethe short report cycle of CQI enables Node B to estimatethe downlink channel condition precisely. However, the short

1089-7798/06$20.00 c 2006 IEEE

460 IEEE COMMUNICATIONS LETTERS, VOL. 10, NO. 6, JUNE 2006

report cycle of CQI results in a larger CQI report overheadand uplink interference compared to the long report cycle ofCQI. Conversely, the longer is the CQI report cycle of CQI,the smaller are the CQI overhead and uplink interference, butthe throughput performance is poorer.

In addition to the periodic CQI reporting scheme, anactivity-based CQI reporting scheme and NACK-based CQIreporting scheme are proposed in [5][6]. In the activity-basedCQI reporting scheme, additional CQI reports are sent withevery ACK or NACK of HS-PDSCH data. In the NACK-based CQI reporting scheme, additional CQI reports are sentwith every NACK of HS-PDSCH data. Both schemes enable amore exact downlink channel estimation of Node B by usingthe additional CQI reports. However, both schemes use morepower due to the additional CQI reports.

After receiving the CQI value from each UE, Node B selectsa suitable modulation scheme, coding rate and number ofcodes for HS-PDSCH. In order to provide control informationto each UE in time, Node B uses High-speed Shared ControlChannel (HS-SCCH). Each UE must monitor a maximum offour HS-SCCHs simultaneously. By monitoring HS-SCCHs,each UE is informed of the upcoming data transmission, andcan prepare to receive data.

III. PROPOSED CQI REPORTING SCHEMESWith the periodic, activity-based and NACK-based CQI

reporting schemes, UEs must report the CQI value periodi-cally, although the channel quality is unchanged and hencethe current CQI value is the same as the last reported CQIvalue. Obviously, reporting the same CQI value consecu-tively wastes the UEs battery power. To solve this problemand to allow effective scheduling and MCS selection fortransmissions on the HS-PDSCH, while at the same timeminimizing the uplink interference and battery consumptiondue to CQI transmissions on HS-DPCCH, we propose twochannel adaptive CQI reporting schemes (CAS1, CAS2). Theproposed CQI reporting schemes are based on the Release5 specifications. The specification related to the method ofderiving an individual CQI value by the UE is not changed inthe proposed CQI reporting schemes.

The basic behavior of CAS1 and CAS2 is as follows. If thechannel is unchanged, and hence the current CQI value is thesame as the last reported CQI value, UEs do not report theCQI value although they are at their own CQI report timing.They send the CQI value only when the current CQI value isdifferent from the last reported value or the timer has expired.The timer is initialized whenever the CQI value is sent by theUE.

The timing of reporting the CQI value of CAS1 and CAS2is different. In the CAS1, UEs wait until their own CQI reporttiming and report the CQI value at that time, when the currentCQI value is different from the last reported value or thetimer has expired. So, with CAS1, battery capacity of UEsis conserved by preventing consecutive reports of the sameCQI value, maintaining the performance of the periodic CQIreporting scheme. Fig. 1(a) illustrates CAS1.

In the CAS2, UEs report the CQI value immediately whenthe current CQI value is different from the last reported value

Node B

(a)

(b)

UE

Timer hasexpired

CQIchanges

CQIchanges

Node B

UE

Timer has expired

CQIchanges

CQIchanges

Periodic CQI report cycle

Fig. 1. Diagram illustrating (a) CAS1 and (b) CAS2.

or the timer has expired. So, with CAS2, more CQI reportsmay be sent than with the periodic CQI report scheme if thechannel quality changes quickly, resulting in frequent changesof the current CQI value. However, the CAS2 enables moreexact downlink channel estimation of Node B by fast CQIinformation update, and hence the probability of transmissionerror is smaller in case of CAS2. If the channel quality isstable, the battery power of UEs also can be conserved withCAS2. Fig. 1(b) illustrates CAS2.

IV. SIMULATION CONFIGURATION AND RESULTS

The default simulation parameters are as follows. We em-ploy a no-sector, 19-hexagonal cell model, where the site-to-site distance is 2800m. At initialization, the number of UEsin each cell is 20, and the location and moving direction ofeach UE are assigned randomly with a uniform distributionwithin each cell. Every 5 seconds, the UEs either changetheir moving direction randomly with a probability of 0.3or do not change with a probability of 0.7 [7]. The numberof codes for HS-PDSCH is 15, and the carrier frequency is2.19GHz. We consider a maximum of four retransmission andfive MCS levels (QPSK with R=1/4, 1/2, 3/4, and 16QAMwith R=1/2, 3/4). We assume that the channel estimation isideal and that HS-SCCH, HS-DPCCH transmissions are errorfree. In addition, we assume that the roundtrip delay is sixsubframes. We use an open-loop traffic model for HSDPA[8] in which the distribution of the packet size is a Paretodistribution with cut-off, =1.1, k=4.5kbytes and m=2Mbytes.The channel model is one path rayleigh fading, and pass lossis proportional to 1/d4, where d is distance. Meanwhile, thelink layer simulation is executed using the FER data in [9].The timers of CAS1, CAS2 are set to expire after 160 ms,which is the maximum CQI report interval of the periodicCQI reporting scheme.

Fig. 2 shows the normalized transmission error ratio accord-ing the various speeds of UE and the CQI report intervals of

JEON and CHO: CHANNEL ADAPTIVE CQI REPORTING SCHEMES FOR HSDPA SYSTEMS 461

102 40 800.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

10 km/h20 km/h30 km/h40 km/h50 km/h

Dashed lines : CAS1Solid lines : CAS2

204 8 160CQI Report Interval [ms]

Nor

mal

ized

Tran

smis

sion

Err

orR

atio

Fig. 2. Normalized transmission error ratio of CAS1 and CAS2 accordingto various UE speeds and CQI report intervals.

102 40 80

10 km/h20 km/h30 km/h40 km/h50 km/h

Dashed lines : CAS1Solid lines : CAS2

204 8 160CQI Report Interval [ms]

0.65

0.7

0.75

0.8

0.85

0.9

0.95

1.05

1.1

1.15

1.0

Nor

mal

ized

Con

sum

edB

atte

ryP

owe r

Rat

io

Fig. 3. Normalized consumed battery power ratio of CAS1 and CAS2according to various UE speeds and CQI report intervals.

the periodic CQI reporting scheme. The normalized transmis-sion error ratio is defined as the probability of transmissionerror with the proposed scheme divided by the probability oftransmission error with the periodic CQI reporting scheme.Since CAS1 has no fast update of CQI information, itsperformance is the same as that of the periodic CQI reportingscheme with respect to the probability of transmission error.However, with CAS2, the probability of transmission errorcan be much smaller than that of the periodic CQI reportingscheme. As the periodic CQI report interval is longer and thespeed of UEs decreases, the fast update of CQI informationhas more effectiveness in enhancing the accuracy of AMC, andhence CAS2 performs better than the periodic CQI reportingscheme with respect to the probability of transmission error.

Fig. 3 shows the normalized consumed battery power ratioaccording the various speeds of UE and the CQI report inter-vals of the periodic CQI reporting scheme. The normalizedconsumed battery power ratio is defined as the consumedbattery power of UEs with the proposed scheme dividedby the consumed battery power of UEs with the periodicCQI reporting scheme. In the simulation, we assume thateach UE always spends a power of 1 due to uplink datatransmission and it spends a power of 0.5 additionally whenit reports the CQI value. Since the redundant CQI reportsof the periodic CQI scheme are filtered off with CAS1, theconsumed battery power of UEs for CAS1 is smaller than thatfor the periodic CQI reporting scheme. Hence, the normalizedconsumed battery power ratio of CAS1 is always less than 1.However, if the speed of UEs increases and the CQI reportinterval is very long, the normalized consumed battery powerratio of CAS2 can be greater than 1, because the CQI valuechanges more frequently as the speed of UEs increases, andthe number of CQI reports during one periodic CQI reportinginterval increases as the periodic CQI report interval is longer.

V. CONCLUSIONSWe have proposed two channel adaptive CQI reporting

schemes (CAS1, CAS2). With the proposed schemes, UEs donot report the same CQI consecutively, even when it is time forthem to report. They report the CQI value only if the currentCQI value is different from the last reported value or thetimer has expired; at their own report timing with CAS1 andimmediately with CAS2. As a result, UEs can conserve theirbattery power with CAS1, maintaining the performance of theperiodic CQI reporting scheme. With CAS2, the probabilityof transmission error is greatly reduced, although there is apossibility that the UEs consume more battery power than theperiodic CQI reporting scheme.

REFERENCES[1] Yang et al., Reliable adaptive modulation aided by observations of

another fading channel, IEEE Trans. Commun., vol. 52, pp. 605-611,Apr. 2004.

[2] 3GPP, Downlink channel quality report scheme, 3GPP TSG R1-01-1055.

[3] 3GPP, Physical layer procedures (FDD), 3GPP TS25.214 V6.7.1.[4] 3GPP, Radio resource control (RRC): protocol specificaition, 3GPP

TS25.331 V6.8.0.[5] 3GPP, High speed download packet access (HSDPA) enhancements,

3GPP TR25.899 V6.1.0.[6] N. Fukui, Study of channel quality feedback in UMTS HSDPA, in

Proc. 14ht IEEE PIMRC 2003, vol. 1, pp. 336340.[7] W. S. Jeon, D. G. Jeong, and B. Kim, Design of packet transmission

scheduler for high speed downlink packet access systems, in Proc. 55thIEEE VTC Spring 2002, vol. 3, pp. 11251129.

[8] 3GPP, Physical layer aspects of UTRA high speed downlink packetaccess, 3GPP TR25.848 V4.0.0.

[9] M. Dottling, J. Michel, and B. Raaf, Hybrid ARQ and adaptivemodulation and coding schemes for high speed downlink packet access,in Proc. 13th IEEE International Symposium on Personal, Indoor andMobile Radio Communications, vol. 3, pp. 10731077.