Optical Communication & Device Technologies Edition

ISSN 1345-3041

3G Mobile-Phone Technologies Edition

VOL. 102/JUN. 2003

● Vol. 102/June 2003 Mitsubishi Electric ADVANCE

A Quarterly Survey of New Products, Systems, and Technology

TECHNICAL REPORTSOverview ............................................................................................ 1by Yoshifumi Ito

W-CDMA Mobile Phone Baseband-DemodulationTechnology ......................................................................................... 2by Takahisa Aoyagi and Takahiko Nakamura

W-CDMA Mobile-Phone Transceiver Devices .................................. 5by Eiji Taniguchi and Shintaro Shinjo

W-CDMA Speech and Acoustic Processing Technology ................. 8by Shinya Takahashi, Keiko Yoshida and Hitoshi Maruhashi

W-CDMA Mobile-Phone Video Coding and TransmissionTechnology ........................................................................................ 11by Fuminobu Ogawa and Shin-ichi Kuroda

W-CDMA Mobile-Phone Data-Security Technology ........................ 14by Tetsuo Nakakawaji, Mitsuru Matsui and Takeshi Chikazawa

W-CDMA Mobile-Phone Software Technology ................................ 16by Tomo-oki Ukiana and Yoshiaki Katayama

Mobile-Phone Structural Design Technology .................................. 18by Shiro Takada and Atsushi Musha

W-CDMA Mobile-Phone Imaging Technology .................................. 21by Yoshiko Hatano

W-CDMA Mobile-Phone Design ....................................................... 23by Yoshihito Nakahara and Tomoaki Tsukada

CONTENTS

Mitsubishi Electric Advance is published online quarterly (in March, June, September,and December) by Mitsubishi ElectricCorporation.Copyright © 2003 by Mitsubishi ElectricCorporation; all rights reserved.Printed in Japan.

Cover StoryOur cover shows development concept modelsfor third-generation mobile phones, which areintended to help their owners create a truly 21stcentury lifestyle for themselves.

In our continuously evolving digital-network based society, third-generationmobile-phone technology already enablespeople to make use of knowledge andinformation from all around the world. Thiswill both encourage further businessdevelopment and significantly enhancelifestyles.

Kiyoshi Ide

Chisato KobayashiKoji KuwaharaKeizo HamaKatsuto NakajimaHiroshi HasegawaHiroshi MuramatsuNoriichi TajimaFuminobu HidaniYukio KurohataHiroshi YamakiKiyohide TsutsumiOsamu MatsumotoHiromasa Nakagawa

Kiyokazu Chiba

Keizo HamaCorporate Total Productivity Management& Environmental ProgramsMitsubishi Electric Corporation2-2-3 MarunouchiChiyoda-ku, Tokyo 100-8310, JapanFax 03-3218-2465

Atsuya Kume3G Mobile Terminal Project GroupMobile Terminal CenterMitsubishi Electric Corporation8-1-1 Tsukaguchi-HonmachiAmagasaki, Hyogo, 661-0001, Japan

Editor-in-Chief

Editorial Advisors

3G Mobile-Phone Technologies Edition

Vol. 102 Feature Articles Editor

Editorial Inquiries

Product Inquiries

· 1June 2003

TECHNICAL REPORTS

*Yoshifumi Ito is Executive Vice President and a Member of the Board of Mitsubishi Electric Corporation.

OverviewTechnologies for Third-Generation Mobile Phones

by Yoshifumi Ito*

The 21st century has opened up a new era of third-generation mobile phoneswith mobile-multimedia capability under the new IMT-2000 global standard.Videophones and high-speed data transmission are greatly enhancing the powerof mobile communications and providing new applications for daily life. In Japan,a third-generation mobile-communication service using W-CDMA technology waslaunched by NTT DoCoMo in October 2001. This service is currently being ex-panded as FOMA.

Mitsubishi Electric Corporation recognized the importance of commonly ac-cepted global standards for mobile-communication systems from the beginning ofthe IMT-2000 standardizations, and our research laboratories in Japan, Europe andthe United States have been active participants in this work. Based on the “MISTY”encryption algorithm developed by the corporation in 1996, the standardized“KASUMI” encryption algorithm was developed in ETSI in 2000 for use in W-CDMA mobile phones. This special edition of Advance introduces major tech-nologies developed within Mitsubishi Electric to design and implement W-CDMAmobile phones.

Mobile phones using W-CDMA technology feature many advantages, includinghigh quality and high-speed transmission of information, but they are also bring-ing about new forms of mobile-phone utilization by providing mobile-videophonefunctions and combining conversation with Internet access and other forms ofdata communication. Under the slogan “Better Link to the Future,” we are com-mitted to contributing to richly enhanced lifestyles in the 21st century by ourongoing development of third-generation mobile-phone technologies. ❑

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE2 ·

*Takahisa Aoyagi and Takahiko Nakamura are with the Information Technology R&D Center

by Takahisa Aoyagi and Takahiko Nakamura*

W-CDMA Mobile Phone Baseband-Demodulation Technology

This article describes the structure and featuresof a baseband demodulator developed for the“FOMA” W-CDMA mobile phone (type D2101V).The demodulation function consists of three pri-mary structural blocks: the searcher unit, thefinger unit, and the decoder unit. The size of thecircuitry and power consumption have been re-duced by using interpolation processing to reducethe operating speed, shared hardware and time-division processing, etc.

The International Mobile Telecommunications2000 (IMT -2000) third-generation mobile-phonesystem is expected to comfortably exceed the com-munications capacities provide by second-genera-tion digital systems such as the personal digitalcellular (PDC) system protocol. This makes it pos-sible to handle high-speed communications for, say,video; both low-speed and high-speed communi-cations can be accommodated efficiently. Progressis being made in international standardizationwithin the International TelecommunicationUnion (ITU). W-CDMA is undergoing developmenteven within IMT-2000 as a powerful communica-tions protocol, and in 2001, NTT DoCoMo, Ltd.,was the first in the world to market a third-gen-eration system, known as FOMA.

Mitsubishi Electric Corporation has contributedto the standardization activities through the adop-tion of the KASUMI encryption algorithm in the3rd Generation Partnership Project (3GPP) Standard.Our efforts have focused on the activities of the3GPP, the standardization organization for W-CDMA. Also, since beginning W-CDMA mobile-phone development, we have been active in thedevelopment of demodulation units for W-CDMA[1].These constitute a key technology for reducingpower consumption and improving functionality.The corporation’s proprietary RAKE receptionmethod protocol has been incorporated into the in-tegrated circuits. The direct-conversion method hasalso been adopted for packaging/insertion intoFOMA-enabled video image-capable mobile phone(D2101V) from NTT DoCoMo. Fig. 1 shows a blockdiagram of the demodulation processing unit de-veloped. This article describes the structure andfeatures of the baseband demodulator, which is vitalto the process of demodulation.

A/D

Finger Viterbi/turboDecoder

De-interleaverDe-coder

Searcher

Demodulator block

High-performance RAKE recept ionReduced memory usage throughthe use of a s l id ing window funct ionHigh-speed Vir terbi / turbo decodingSmal ler / low-power matched f i l ter

Hardware controller

Control block

Nyquist filter Interleaver

Encoder blockModulator block

Spread spectrummodulator

Convolutional/turboencoder

Timing/sequencecontrollerRF/IF

block Adapterblock

D/A

Fig. 1 Block diagram of the digital basebandprocessing unit

Structure of the Baseband Demodulator UnitTable 1 shows the specifications of the basebanddemodulator unit devloped. This demodulator isbased on the 3GPP Standard, Release 99[2], andis compatible with a variety of physical channelformats, from 12.2kbps for voice through a maxi-mum of 384kbps. The primary receiver functions

Table 1 Specifications of the BasebandDemodulator

Standard Based on the 3GPP Release 99Standard

Wireless-access methodDirect sequence-code division multipleaccess frequency-division duplex

Chip rate 3.84Mcps

Multicode Up to 3 codes

Information data rate 12.2 kbps to 384 kbps

Forward-error correction Virterbi/turbo decoding

Reception diversity RAKE combing (up to 8 fingers)

are provided by the various blocks shown in Fig.2. The primary blocks are the searcher block,which provides searcher functions (to ensuresystem synchronization) and multipath-detec-tion functions (to measure a delay profile); a fin-ger unit, which performs the RAKE receptionby performing the despread function on the vari-ous multipaths and by performing coherent de-

TECHNICAL REPORTS

· 3June 2003

tection; and the decoder block, which performshigh-speed Virterbi/turbo decoding. The featuresof these functional blocks are described below.

Searcher BlockThe searcher block itself comprises the four sub-blocks shown in Fig. 2. These blocks are inde-pendent of each other, and perform cell searchesand path-timing detection under operationalcontrol from the controller unit. The searcherblock provides, primarily, the following functions: * Detection of receivable frequency bands * Cell searches that detect scrambling codes

and timing * Path-timing detection for RAKE reception * Periodic level measurements for the detected

cells.

The cell-search functions are also able tosearch for cells with a frequency deviation ofgreater than 6kHz prior to exerting automaticfrequency control. In cell searches after auto-matic frequency control, the search time andpower consumption are reduced by the simulta-neous detection of multiple base stations.

In the path-timing measurements for the RAKEreception, multiple digital matched filters areprovided in order to measure the delay profiles ofmultiple cells quickly, enabling stabilized recep-tion even in soft handovers when the mobilephone is traveling at high speeds. The digitalmatched filtering provides a large reduction inpower consumption through low-speed operationsafter checking the degradation due to aliasing andchecking tolerable performance limits.

Finger BlockThis enables RAKE reception by performingdespread processing and coherent detection uti-lizing pilot symbols appropriate to the variousdelay paths detected by the searcher block. The

A/D

Controller block

1st sync code (p-SCH) detector

2nd sync code(s-SCH) detectorScrambling code detector

RAKE path detector

Channel power measurement

Searcher block

Finger 1 2nd de-interleaver

1st de-interleaver

Rate de-matching

Turbodecoder

CRC check

Convolutionaldecoder

Finger 2

Finger n

RAKE receiver unit 1

RAKE receiver unit 2

Finger block Decoder block

Finger 1

Finger 2

Finger nPIL

generator

Fig. 2 Block diagram of the digital baseband demodulator

finger block has multiple fingers that performthe coherent detection processes on each path,enabling RAKE-combining synthesis with a maxi-mum of eight fingers. The finger unit also hastwo sets of RAKE receiver functions for data-channel and control-channel reception, so thatdata and control channels can be received si-multaneously. Each finger can handle differentbase stations, enabling RAKE combining for upto four base stations in receiving data channelsduring soft handover.

The operations and functions of the block de-pend on the various specifications in the 3GPPstandard[2], and RAKE reception is supported notonly for dedicated physical channels but also forthe common control channels such as the pri-mary and secondary common-control physicalchannels, indicator channels such as acquisi-tion indicator channels and paging indicatorchannels, and synchronization channels used indetecting transmission diversity types applied tothe primary common control physical channel.In the dedicated physical channels, data recep-tion up to a maximum of 384kbps is supportedthrough triple multicode reception. For down-link transmit diversity, space-time transmit di-versity, closed-loop mode 1, and closed-loop mode2 are each supported.

The circuit scale and power consumption forthe functions described above are kept to a bareminimum by the effective use of the interpola-tion method and shared circuitry (using time-multiplexed processing).

Channel Decoder BlockThis performs the second de-interleaving andrate matching for each physical channel on thesoft-decision signals input from the finger block.It also performs the first de-interleaving and er-ror correction for each transport channel, anderror detection using cyclic-redundancy checks

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE4 ·

(CRCs). Convolutional decoding and turbo de-coding are provided for error correction, and theerror-correction processing uses the methodspecified by the digital-signal processor. In theconvolutional decoding, a soft-decision Viterbidecoding process is performed, and turbo decod-ing is performed by iterative decoding using asoft-input/soft-output algorithm that uses amaximum a-posteriori probability (max-logMAP)algorithm. The benefits of this approach includethe following:1. Although, when performing turbo decoding

using the max-logMAP algorithm, the forwardor backward path-metric values must all berecorded at the point in time for the codelength, the sliding window function makes itpossible to reduce to one-third the memorycapacity that must be provided in the turbo-decoder unit, reducing the memory region bystoring only a portion of the path-metric val-ues and decoding a part at a time.

2. A blind transport-format detection (BTFD)function has been added to infer the trans-port format that is transmitted, doing so basedon error-detection results using Viterbi decod-ing that correspond to the variable rates whendecoding the adaptive multi-rate (AMR) au-dio signal used in W-CDMA.

3. The following structures are used so as to havepacket signals with transmission-time inter-vals (TTIs) equal to 20ms at 384kbps, AMRaudio signals at 12.2kbps, and dedicated con-trol channels (DCCH) processing at no morethan 10ms:

* The primary functions are hardwired, and theparameters such as the calculation settingsare set by the DSP, providing both high speedand flexibility.

* The turbo-decoding block and convolutionalblock can operate in parallel, making it pos-sible to process multiple transport channelssimultaneously.

4. Although performance is improved when thenumber of decoding iterations for turbo en-coding is large, increasing the number of it-erations increases power consumption, andso the number of decoding iterations is set bythe digital-signal processor, increasing flex-ibility.

CharacteristicsFig. 3 shows typical performance characteristicsobtained using a functional prototype based onthe demodulation method described above. Thefigure shows the block-error rate characteristicsgiven in the Case 1[3] environment defined in the3GPP standard (i.e., a two-path model corre-sponding to 3km/h), and the results essentially

match those from a computer simulation. Themethod developed in this research project ob-tained characteristics that satisfy the 3GPP stan-dard under other conditions as well, and an ICcontaining the modulator and controller unitshas been created based on these prototyping re-sults. This is used in the FOMA mobile phone(D2101V) made by NTT DoCoMo.

The W-CDMA protocols are implemented by aconfluence of multiple individual technologies,where most of these technologies do not, bythemselves, provide optimal solutions for thetransmission capacity of the system as a whole.Thus, in the development of modulation/de-modulation technology in the future, it seemscertain that system performance design, in par-ticular, will become increasingly important. Inaddition, we anticipate the development ofmodulation/demodulation units for high-speedcommunications, such as high-speed downlinkpacket access (HSDPA), in addition to technolo-gies for further miniaturization and power re-ductions, that are superior to those used inmobile phones for PDC systems. ❑

References[1]Takahisa Aoyagi, Michiaki Takano, Yasuhiro Yano, Kenro Hirata,

Hideshi Murai and Makoto Miyake, “Demodulator IC for W-CDMA,”Institute of Electronics, Information, and Communication EngineerGeneral Conference Proceedings, B-5-120 (1999)

[2]3rd Generation Partnership Project: Technical Specification GroupRadio Access Networks; TS25.211 to TS25.215 V3.5.0 (Release99) (Dec. 2000)

[3]3rd Generation Partnership Project: Technical Specification GroupRadio Access Networks; TS25.101 V3.5.0 (Release 99) (Dec. 2000)

simulation_CASE1_BLERexperiment_CASE1_BLER

Eb/No[dB]

1.E-040.0 5.0 10.0 15.0 20.0

1.E-03

1.E-02

1.E-01

1.E+00

BL

ER

Eb/No vsBLER(UDI54kbps,CASE1)

Fig. 3 Block error rate (BLER) of the prototypeunit

TECHNICAL REPORTS

· 5June 2003

*Eiji Taniguchi and Shintaro Shinjo are with the Information Technology R&D Center

by Eiji Taniguchi and Shintaro Shinjo*

W-CDMA Mobile-PhoneTransceiver Devices

Mitsubishi Electric Corporation has developedtransceiver amplifiers for the broad dynamicrange required by W-CDMA applications. AnSiGe process has been used in low-noise ampli-fiers for receivers and driver amplifiers for trans-mitters with internal high-performance biascircuits, and a GaAs-pseudomorphic high elec-tron-mobility transistor process has been usedin the design of interstage matching circuits forhigh-power transmitter amplifiers that provideoptimized source/load impedance. Both devicesprovide excellent performance.

Radio Frequency Component CircuitStructuresThe radio frequency (RF) component circuit struc-tures for a W-CDMA mobile phone are shown inFig. 1. The receiver unit has a two-stage low-noiseamplifier, integrated with a direct conversion mixeractive element (a diode) in a SiGe monolithicmicrowave integrated circuit. In the transmitterunit, the driver amplifier is integrated with RF andintermediate-frequency (IF) variable-gain amplifi-ers, an up mixer, and a quadruple modulator in anSiGe monolithic microwave integrated circuit. Thehigh-power amplifier uses a GaAs pseudomorphichigh electron-mobility transistor to provide high-efficiency/low-distortion performance. Variable-gain amplifiers are used in the RF units and the IFunits in order to provide a broad dynamic range.The low-noise amplifier, the driver amplifier, and

EHMIX

RX

ISO Balun RF-SAW IF-SAW

IF-VGARF-VGAHPA

LHA1

Drv-Amp

LHA2

UPMIX

QMOD

LO2 LO2

ANT

DUP

BRF1 BRF2

EHMIX

LO

0οDIV90ο

HYB

LO1

Q

I

Q

I

TX

Fig. 1 Block diagram of RF section

the high-power amplifier were developed asdescribed in this article. The circuit structures andperformance of these amplifiers will be explainedbelow.

LOW-NOISE AMPLIFIER. The low-noise amplifierused in W-CDMA generally requires not only alow noise figure, low DC current operation, andsmall size, but also high-saturation and low-dis-tortion characteristics for transmit signal leak-age. In this development project, high-saturationcharacteristics and low distortion were achievedwhile suppressing the quiescent current andminimizing degradation in the basic perfor-mance (i.e., the noise figure and the gain)through the application of a low-noise amplifier.This is based on an innovative dual-bias feed cir-cuit in which the bias circuit is switched de-pending on the input signal level[1]. Fig. 2 showsthe circuit structure of this innovative dual-biasfeed-type low-noise amplifier. This amplifier hastwo different base-bias circuits for the resistorfeed and the diode feed, where, when a smallsignal is inputted, the diode bias-feed circuit isessentially in an open state; on the other hand,when there is a large input signal, the diode bias-feed circuit operates as a constant-voltagesource, producing low-noise, high-saturation,and low-distortion performance.

The results of prototyping using an SiGe pro-cess showed excellent performance in the dual

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE6 ·

fier. In the figure, (a) is an equivalent circuitdiagram of a conventional driver amplifier (TypeA) with an internal constant-voltage bias circuit.The Type A driver amplifier does not supply anadequate base current when there is a high out-put power if the quiescent current is reduced,leading to poor performance in terms of distor-tion. Fig. 3 (b) is an equivalent circuit diagramof the innovative driver amplifier (Type B) witha self base bias-control circuit.[2] This latter cir-cuit is structured from a combination of a p-metal oxide semiconductor field-effect transistor(FET) current-mirror circuit, in a constant-volt-age bias circuit, and a constant-current source.The Type B driver amplifier makes it possible tosupply automatically a current that is beyondthe increase in the base current produced by theType A driver amplifier at a high output power.Consequently, excellent distortion characteris-tics can be obtained even when operating witha low quiescent current.

The results of calculating the I/O characteris-tics of the Type B driver amplifier under the con-ditions of a 1.95GHz frequency and a 15.0mAquiescent current show an output 1dB compres-sion-point improvement of 2.4dB over the TypeA driver amplifier.

Furthermore, the Type B driver amplifier pro-totype had an output 1dB compression point of15.0dBm and an operating current of 32.6mA.The application of the self base-bias control cir-cuit demonstrated the ability to improve the dis-tortion characteristics when operating with alow quiescent current.

HIGH-POWER AMPLIFIER. High-power amplifiersused in W-CDMA mobile phones must providehigh efficiency while satisfying distortion re-quirements. In multistage high power amplifi-

bias-feed low-noise amplifier, with a gain of14.8dB, a noise figure of 2dB, an output 1dB com-pression point of 4.5dBm, and the third inter-cept point at 0.2dBm, at 6.1mA DC. Furthermore,for comparison, a conventional resistor bias feed-type low-noise amplifier, such as is commonlyused, was fabricated using the same processtechnology. Comparisons confirmed not onlythat the DC current value and the small signalcharacteristics were about the same, but thatthe saturation characteristics and the distortioncharacteristics had been improved by more than5dB.

DRIVER AMPLIFIER. Driver amplifiers must op-erate with a low quiescent current, must haveless distortion than high-power amplifiers, andmust be small in order to be made into mono-lithic microwave integrated circuits. Fig. 3shows an equivalent circuit of the driver ampli-

Resistor bias feed circuit

RB1

M/S

M/S

IC

Vcc

RB

Reference resistor Diode bias feed circuit

Fig. 2 Schematic diagram of dual-bias feed LNA

constant base voltage circuit

self base bias-control circuit

M/S M/S

M/S M/S

Fig. 3 Schematic diagram of driver amplifier

(a) Driver amplifier with a constant base-voltage circuit (Type A)

(b) Driver amplifier with a self basebias-control circuit (Type B)

TECHNICAL REPORTS

· 7June 2003

ers it is not only the FET in the final stage thatoperates with a large signal; the FET driving thefinal-stage FET operates at nearly the same sig-nal strength. It is well known that the optimalsource and load impedances for FETs when op-erating with small signals are different fromthose with large signals. Consequently, in orderto ensure high efficiency while satisfying dis-tortion requirements, it is necessary to researchthe structure of the inter-stage matching circuitsand optimize impedances between stages.

Fig. 4 shows an equivalent circuit diagram ofthe prototype three-stage high-power amplifiermodule. A GaAs pseudomorphic high electron-mobility transistor process was used to obtainhigh efficiencies and low distortion in the am-plifier elements. A high pass filter/low pass fil-ter combined matching circuit was used betweenstages 2 and 3[3]. This high pass filter/low passfilter combined matching circuit makes it pos-sible to ensure an optimal load impedance forthe second stage FET while ensuring an opti-mal source impedance for the third stage FETgiven the W-CDMA specifications. Using a hy-brid phase-shift keying modulated wave with afrequency of 1.95GHz and a chip rate of3.84Mcps, the I/O characteristics of the three-stage high-power amplifier module developedwere evaluated, indicating that a power-addedefficiency of 43.9% with an output power of27.1dBm was obtained with an adjacent chan-nel leakage power ratio of -38.0dBc and a nextadjacent channel leakage power ratio of -49.1dBc.It is evident that the use of the high-pass filter/low-pass filter combined-type interstage match-ing circuit makes it possible to obtain high effi-ciency characteristics while satisfying distortionrequirements.

Input-matching

circuit

1-2 Interstagematching

circuit

2-3 Interstagematching

circuit

HPFMatching

circuit

LPFMatching

circuit

Outputmatching

circuit

IN OUT

Fig. 4 Schematic diagram of the three-stage HPA module using the high pass filter/low pass filter combinedinterstage matching circuit developed in this development project

All of the circuit structures applied to the W-CDMA transceiver amplifiers in this developmentproject are highly effective in communicationssystems requiring low power consumption anda wide dynamic range. They therefore promiseto find wide utility in the next generation of mo-bile equipment. ❑

References[1]E. Taniguchi, K. Maeda, T. Ikushima, K. Sadahiro, K. Itoh, N.

Suematsu and T. Takagi, “Dual Bias Feed SiGe HBT Low NoiseLinear Amplifier”, 2001 IEEE MTT-S IMS, Dig., pp. 285-288, May,2001

[2]S. Shinjo, K. Mori, H. Joba, N. Suematsu, and T. Takagi, “LowQuiescent Current SiGe HBT Driver Amplifier Having Self Base BiasControl Circuit”, IEICE Trans. Electron., Vol. E85-C, No. 7, pp 1404-1411, July 2002

[3]K. Mori, S. Shinjo, F. Kitabayashi, A. Ohta, Y. Ikeda and O. Ishida,“An L-band High Efficiency and Low Distortion Power AmplifierUsing an HPF/LPF Combined Interstage Matching Circuit”, IEEETrans. Microwave Theory and Tech., Vol. 48, No. 12, pp 2560-2566, December 2000

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE8 ·

*Shinya Takahashi is with Information Technology R&D Center, Keiko Yoshida is with Advance Technology R&D Centerand Hitoshi Maruhashi is with the Mobile Terminal Center.

by Shinya Takahashi, Keiko Yoshida and Hitoshi Maruhashi*

W-CDMA Speech and AcousticProcessing Technology

In order to differentiate them from conventionalmobile phones, W-CDMA mobile phones requireexcellent basic audio quality on a par with normalhard-wired telephones. When the videophone func-tion—a unique feature of W-CDMA—is used in ahands-free phone mode, causing loud echoes andbackground noises to be picked up by the micro-phone, it is important to suppress these noises.Because of this, 3GPP has strict standards relatingto speech and acoustic characteristics.

Speech and acoustic processing in W-CDMAmobile phones is performed by a digital signalprocessing unit, comprising a voice CODEC, anecho-cancellation unit, a noise suppresser, andan analog acoustic processing unit consisting ofthe receiver, the microphone, and the periph-eral device structures. Excellence in both thesetechnologies is imperative in providing userswith the excellent audio quality that fulfills the3GPP standard.

The noise suppresser developed at MitsubishiElectric Corporation has superior capabilities, andwas the first in the world to receive 3GPP en-dorsement as fulfilling all the required performancerequirement standards. Overall superior speech andacoustic characteristics have been obtained bydeveloping a high-quality receiver packageapproach that meets the frequency characteristicstandards for 3GPP receivers, and an echo-cancellation unit with excellent performance.

This article describes the 3GPP speech andacoustic standards, and the processing technolo-gies we developed to meet these standards.

Table 1 provides a summary of the adaptivemulti-rate (AMR) method that is the 3GPP stan-dard for speech CODECs, and a summary of theAMR-wide band (AMR-WB) version of it. Bothmethods can operate at variable bit rates for

compatibility with different transmission capaci-ties. The AMR method for the 12.2kbps bit rateused by most W-CDMA mobile phones in Japanhas extremely high sound quality, approachingthat of hard-wired phones. 3GPP has also stan-dardized performance requirements for the fre-quency characteristics for analog audio systemsand for AMR noise suppression.

When mobile phones are used in the hands-free mode, loud echoes reach the microphonefrom the speaker. Although a subtractive methodis used to suppress these echoes (in which theechoes are inferred through adaptive filtering),because the echoes include non-linear compo-nents generated within the audio section, adap-tive filtering alone is often unable to suppressthe echoes completely. The corporation hastherefore introduced technology that attenuatesthe residual echo period without producing anaudibly odd sound quality by controlling fre-quency characteristics. In addition, a delayeddetection method is used to perform high per-formance double-talk detection guarding againstincorrect echo inferences when there is doubletalk (i.e., when individuals on both the trans-mitter side and receiver side speak simulta-neously). [1]

Noise Suppresser

A PROPRIETARY NOISE-SUPPRESSION METHOD.Noise suppression (NS) requires substantial sup-

Fig.1 Block diagram of the W-CDMA mobile phone

Speechinformation(12.2kbps)

SpeechDecoder

DigitalSpeech

processing

AnalogAcoustic

processing

Noisesignal

Ech

o si

gna

l

Ech

o C

ance

ller

Mod

ulat

or /

Dem

odul

ator

W-CDMA terminal

NoiseSuppresser

SpeechEncoder

Speech Codec

D/A

A/D

Table 1 Specifications for the 3GPP StandardSpeech CODEC

CODEC AMR AMR-WB

Bandwidth 300Hz ~ 3.4kHz 50Hz ~ 7kHz

Sampling frequency 8kHz 16kHz

Bitrate 12.2 ~ 4.75kbps 23.85 ~ 6.6kbps8 grades 9 grades

TECHNICAL REPORTS

· 9June 2003

pression of the background noise only, withoutdegrading the speech signal. Fig. 2 shows a blockdiagram of the NS method developed at Mitsu-bishi Electric. This method is based on spectralsubtraction where the spectrum that is prima-rily noise is subtracted in the low frequency do-main where the signal-to-noise ratio (SNR) ishigh, thereby reducing the amount of noise,while attenuating the spectrum that is largelynoise in the high-frequency domain where theSNR is low. This method, which suppresses noisewhile preserving the speech spectrum, gives ex-cellent noise reduction.

3GPP EVALUATION AND TEST RESULTS. In 3GPP,the performance requirements for noise suppres-sion are specified in TS26.077. Methods that ful-fill all of the performance requirements insubjective evaluations and objective evaluationsby third-party testing laboratories are granted3GPP endorsement. Not only have the noise-suppression technologies developed by the cor-poration fulfilled all of these performancerequirements (a total of 74 different require-

Fig. 2 Block diagram of the noise suppresser

Estimated noise spectrum

Adaptive noise reduction

Spectrum subtraction coefficientNoise likeness

analysisInput signal

Output signal

Input signal Ouput signal(Noise suppressed)

Amplitudesuppressioncoefficient

Amplitude spectrum

Noise spectrumestimation

Noisespectrum

subtraction

AmplitudesuppressionNoise suppressed

spectrum

Noisesuppresser

Noise-likenessinformation

Fast FourierTransform (FFT)

Inverse FFT

Phase spectrum

Fig. 3 Selected objective evaluation results

English

Signal to noise ratio (SNR)

Imp

rove

me

nt

fro

m w

itho

ut

NS

Japanese

6dB0.0

0.2

0.4

0.6

0.8

1.0

1.2

15dB 9dB 18dB9dB18dB

ments) in subjective testing by third-party labo-ratories in two different languages (Japanese andEnglish), but they have also fulfilled the perfor-mance requirements in objective evaluations.[2]

The result is that, in May 2002, this Mitsubishinoise suppression was the first in the world toreceive 3GPP endorsement. Fig. 3 shows someselected results of the subjective evaluation re-sults regarding the amount of improvement pro-vided through the use of this noise suppression.

Acoustic Design

RECEIVING FREQUENCY CHARACTERISTICSTANDARDS IN 3GPP. In 3GPP, the standardsfor the acoustic characteristics of the terminaldevice are covered in TS26.131, which incorpo-rate some Mitsubishi Electric proposals.[3] Thedistinguishing features of the various standardsare that they were established based on thepremise that the acoustic measurements of theterminal equipment would be done underconditions that are extremely near to those ofactual use, using, for example, an artificial earthat takes into account the leakage that ispresent in a real ear. Fig. 4 shows the frequencycharacteristics masking pattern (the boundariesshowing the upper and lower limits on thefrequency characteristics for phone communi-cations) of the receiver during handset conversa-tion in these specifications. In line with ourproposal, the upper limit in the low frequencyband (up to 100Hz) was relaxed in anticipationof the implementation of the wide-band CODEC,AMR-WB.

W-CDMA MOBILE-PHONE AUDIO DESIGN. In thedesign process for the structure of the casing thatcontains the receiver audio system in the FOMATM

D2101V, we were able to perform the design work

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE10 ·

efficiently by adopting proprietary audio-designsimulation techniques that take into account, forexample, the effects of leakage between the user'sear and the ear-piece or the surrounding mechani-cal structures within the receiver unit. MitsubishiElectric was the first to apply to domestic (Japa-nese) mobile phones a low acoustic impedanceimplementation that makes it possible to main-tain consistently stable low-frequency character-istics even under varying coupling conditions,thereby producing high-quality performance meet-ing the TS26.131 standard, see Fig. 4.

Fig. 4 Receiving frequency response for theTS26.131 and the FOMA D2101V

100

-30

-20

-10

10

0

100001000freq.(Hz)

Receiving Frequency response of FOMA D2101V

3GPP TS26.131

rela

tive

(dB

)

Users will only welcome the sophisticated func-tions made possible by W-CDMA mobile phonesif, at the same time, the very highest audio qual-ity is maintained in speech, free from the distrac-tions caused by noisy operating environments and“hands-free” operation. Mitsubishi Electric notonly helped to develop the standards necessary toensure this, but also stands in the forefront of ef-forts to incorporate these standards in small, reli-able and efficient devices. ❑

References[1]B. Matsuoka, et. al.: Research into a double talk detection method

for acoustical echo cancellation with a decision delay, 2000Institute of Electronics Information and Communication EngineersGeneral Conference Lecture Proceedings, A-10-4 (2000).

[2]S. Furuta, et al.: A noise suppresser for the AMR speech CODECand evaluation test results based on 3GPP specification, 2002IEEE Speech Coding Workshop (2002).

[3]B. Matsuoka, et. al.: Standardizing the terminal device receiverfrequency sound pressure characteristics in 3GPP, 2001 Instituteof Electronics Information and Communication Engineers GeneralConference Lecture Proceedings, D-14-1 (2001).

TECHNICAL REPORTS

· 11June 2003

*Fuminobu Ogawa and Shin-ichi Kuroda are with the Information Technology R&D Center.

by Fuminobu Ogawa and Shin-ichi Kuroda*

W-CDMA Mobile-Phone Video Codingand Transmission Technology

This article describes the 3G-324M standard andits implementation in enabling videophone func-tions using W-CDMA mobile phones. Througha hybrid structure of software for digital signalprocessing combined with proprietary hardware,a single IC module has been enabled to handlevideo decoding, multiplexing/demultiplexing formultimedia communication, and control formultimedia communication. At the same time,the technology has enabled a high degree of er-ror resiliency while reducing the wait time be-fore communications can be started.

Protocol Stack for VideophonesVideophone functions for W-CDMA mobilephones follow the 3G-324M multimedia tele-phony standard established by the 3rd Genera-tion Partnership Project (3GPP). Conformanceto the 3G-324M standard ensures cross-connec-tivity with the terminal equipment made byother companies. Fig. 1 shows the 3G-324M pro-

used to ensure accurate transmission of theH.245 messages, even in an error-prone envi-ronment.

Video EncodingIn the 3G-324M standard, the MPEG-4 visualSP@L0 can be used. In MPEG-4 SP@L0, imagesof the maximum QCIF size (176 pixels x 144lines) must be decoded at 15 frames/second.

MPEG-4 has an error-resiliency function thatprevents the image quality from degrading evenwhen there are transmission errors. The use offunctions such as resynchronization markers toenable rapid recovery when an error has oc-curred, data partitioning to prevent the propa-gation of errors into the image data, andreversible variable-length code (RVLC) to reas-semble from the opposite directions words inwhich there were errors, makes it possible toprevent degradation of the playback image. Thisis achieved even in environments with a randomerror rate of 10−4 and a burst error rate of 10−3.Fig. 2 shows typical decoding results for encodedvideo data to which a random error (BER=10−4) hasbeen applied.

Video encoding:H.263

MPEG-4

MultimediaCommunications

Control:H.245

Audio:AMR

NSRP/CCSRL

Multiplexing: H.223 AnnexB

Wireless Network Interface

Fig. 1 Protocol stack for 3G-324M multimediasystem video encoding: H.263 MPEG-4

tocol stack. ITU-T H.263 is mandatory as thevideo codec, and ISO MPEG-4 visual Simple Pro-file Level 0 (SP@L0) can also be used. The 3GPP-AMR is mandatory for the audio codec, and theITU-T H.245 standard is used for the multime-dia communications control that, for example,exchanges capability information between theterminals. ITU-T H.223 is used for multiplexingthe various media, such as video, audio, and data.In addition, numbered simple retransmissionprotocol (NSRP) frames and a control-channelsegmentation and reassembly layer (CCSRL) are

Represented Image of Coded Data

Error Resiliency OFF Error Resiliency ON

Fig. 2 Represented image of coded data

MPEG-4 standardizes a deblocking filter thatreduces encoding noise, which tends to appearat block boundaries. The deblocking filter ishighly calculation-intensive, and accounts formore than half of the decoding process, and it istherefore necessary to find a way of reducingthe calculation overhead without compromisingquality.

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE12 ·

Multiplexing/Demultiplexing for MultimediaCommunicationsIn the 3G-324M standard, multiple media, suchas video and audio, are multiplexed into a singledata stream, and transmitted according to themultimedia multiplexing procedures standard-ized in ITU-T recommendation H.223. This rec-ommendation establishes a multiplexing methodwith a two-layer structure comprising an adap-tation layer (AL) that performs framing depend-ing on the characteristics of each media type,and a multiplexing layer (MUX) that mixes themedia and does the framing depending on thecharacteristics of the transmission path. Fig. 3shows an example of multimedia multiplexingaccording to H.223. H.223 stipulates differenttypes of ALs tailored for each set of media char-acteristics, and different levels of MUXs withdifferent error resilience. Combining appropri-ate H.223 ALs and MUXs enables multimediacommunications for a wide range of applications.In the 3G-324M standard, MUX Level 2 is appli-cable in view of the level of transmission errors,and the type of AL is selected with respect tothe transmission delays.

CR

CH

eade

r

Hea

der

Hea

der

Hea

der

CR

C

CR

C

CR

C

Adaptation Layer (for each type of media)

Multiplexinglayer

(mixed media)

Multimediamultiplexing

frame(MUX frame)

Audio encoding period

Audio AL frame(Framed by the audio encoded data unit)

Video AL frame(Framed by the video encoded data unit)

Fig. 3 Multiplex Format in H.223

Transmission error has a large impact on thequality of communications in mobile networks.The frame length for multiplexing should be shortto enable rapid recovery when an error has oc-curred. On the other hand, because overheadsuch as the various types of headers, the CRC,etc., also has a large impact on the multiplex-ing efficiency, the frame lengths should be aslong as possible. Given these conflicting require-ments, the multiplexing efficiency can be im-proved through innovations such as dynamicallyand adaptively changing the frame length.

Control for multimedia communicationIn multimedia communications combining au-dio, video, and other media, parameters includ-ing media type used and the encoding methodmust be selected appropriately for the terminalcapability. The multimedia communicationscontrol functions are used, e.g., to exchange ca-pability information between terminals. Thesemultimedia communications control functionsinclude capabilities negotiation and logical-channel signalling to determine the media types,encoding and transmission methods to be used,along with commands and indications carryingmultimedia synchronization information. In 3G-324M-standard terminal equipment, the multi-media communications control method used isa general-use protocol based on ITU-T Recom-mendation H.245. H.245 is used in a variety ofapplications, such as B-ISDN, LANs, and mo-bile communications, and has a wide variety ofcommunications control functions. Assuming itis used in an error-free environment, H.245 en-ables reliable control using in-channel request-response messaging.

To ensure the correct transmission of the H.245messages, the underlying layer provides the error-free environment. The NSRP frames, shown inFig 1, add error-detection information to the data.On the receiver side, the transmission is only ac-knowledged when the data has been receivedcorrectly. Error-free states are ensured by se-quential data transmission at the sending side. Inaddition, the data is segmented at the control chan-nel segmentation and reassembly layer (CCSRL)to reduce the packet error rate.

Before the start of communications, the num-ber of H.245 messages exchanged between theterminal equipment amounts to between 10 and20. Each message is sent sequentially after wait-ing for the transmission acknowledgement.When there is a transmission delay of severalhundred milliseconds in the W-CDMA network,there will be a wait time of about ten secondsbefore communication begins. This wait time isreduced by cancatenating multiple H.245 mes-sages together into each transmission frame.Concatenating many messages together canreduce the transmission time, but may involvethe delay caused by linking the messages afterwaiting for many of them to be generated. Fur-thermore, message concatenation lengthens thepacket, increasing the error rate. Concatenat-ing the H.245 messages adaptively so that mes-sage concatenation causes neither longer delaysnor higher error rates makes it possible to re-duce the wait time before the commencementof communications by about 50 ~ 70%.

TECHNICAL REPORTS

· 13June 2003

3G-324M-Compatible IC ModuleA 3G-324M-compatible IC module has been de-veloped to produce MPEG-4 video communica-tions equipment such as videophones, imagecontent players, and similar devices. The pri-mary functions of this module are MPEG-4 videodecoding, H.223 media multiplexing/demulti-plexing, and H.245 control for multimedia com-munication. The MPEG-4 video encoding isperformed at the camera unit. The video encod-ing unit and the audio codec are separate fromthe module; the data flow is switched using awireless interface unit when there is a video-phone rather than speech communicationsalone, and both the power supply and the clockare supplied according to the operation mode un-der the control of the host CPU. This makes itpossible to reduce power consumption duringvoice communications and during playback oflocally stored video.

As is shown in Fig. 4, the module comprises aVDEC unit that performs the MPEG-4 video de-coding, and a MUX unit that performs the multi-media multiplexing/demultiplexing and themultimedia communications control. Both of theseunits use a hybrid structure with a general-pur-pose DSP and proprietary acceleration hardwarefrom Mitsubishi Electric Corporation, with boththe flexibility and upgradability provided by DSPsoftware processes and the high-speed processingprovided by hardware processes.

Table 1 shows the features of this IC module.The size of the circuit is 245K gates (includingthe DSP core) in the logic unit, and 3.2Mbit (in-cluding the on-board DSP memory) in the on-board memory unit. It uses a 0.18micron CMOSwafer process to achieve a power consumptionof 69mW when performing MPEG-4 QCIF 15frame/second video decoding and multimediamultiplexing/demultiplexing processes.

Table 1. Features of the 3G-324M IC Module

Video decoding

Encoding method MPEG-4 simple profile level 0H.263

Image size QCIF (176 x 144 pixels)SQCIF (128 x 96 pixels)

Decoding Bit rate: max. 64Kbpsperformance Frame rate: max. 15fps

Multiplexing method H.223 AnnexB

AV communications control H.245

Size of circuitsLogic circuits 245K gates

Memory 3.2Mbit

Power consumption69mW (at 48MHz,internal voltage 1.8V)

The article has discussed techniques for enablingtwo-way multimedia communications in the W-CDMA environment. A videophone multipointconnection service was launched on October 1,2002, and the videophone services in the mo-bile-phone environment are anticipated to go be-yond merely “chatting” to include a broad varietyof developments such as remote monitoring,education, contents distribution, etc. ❑

CPU

Displaycontrol

CPU

Camera(Video Enc.)

Wirelessinterface

unit

Audiocodec

MUX unit(H223,H245)

VDEC unit(MPEG-4/H.263 Video Dec.)

HOST I/F HOST I/F

Displayinterface

Interruptcontrol unit

Interruptcontrol unit

SRC

ENC I/F

Core I/FDemux I/F

DSP DSP

IQ

IDCT

Fig. 4 Block diagram of the 3G-324M IC module

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE14 ·

der to detect unauthorized data modification. Itis also known as message authentication. Func-tion f9 is used to generate the authenticationcode (see Fig. 3) to be added to the data in orderto check for unauthorized data modification andensure data integrity. The data, the up/downlink, a counter, a random number for each user,and the integrity key are input to function f9 togenerate the message-authentication code. Thereceiver compares the message-authenticationcode sent by the sender to the message-authen-tication code generated by the receiver, makingit possible to confirm, when the codes match,that there has been no illegal data modification.

*Tetsuo Nakakawaji, Mitsuru Matsui and Takeshi Chikazawa are with the Information Technology R&D Center

by Tetsuo Nakakawaji, Mitsuru Matsui and Takeshi Chikazawa*

W-CDMA Mobile-Phone Data-Security Technology

The article describes the security technologiesrequired in the W-CDMA mobile phone system,including wireless communications-level secu-rity functions, data-communications level secu-rity functions, and contents-level securityfunctions.

Wireless Communications-level SecurityAuthentication.This is a technology for authenticating that theuser has the authority to use the wireless com-munications, or, in other words, that the user isthe subscriber for whom the communicationscharges have been paid. When a call is re-quested, the network side authenticates themobile phone based on the subscriber informa-tion in the handset. If the subscriber informa-tion were sent to the network side in theplain-text mode, it would be possible for an eaves-dropper to steal the subscriber information andsuccessfully impersonate the subscriber. Au-thentication is therefore performed by both sidesperforming calculations using the subscriberinformation and comparing the results. In W-CDMA, there are functions f1 to f5 defined asthose by which the calculations are to be per-formed. The algorithms within these functionsare not subject to standardization, but rather aredetermined by the operators.Note: In the process by which the calculations forauthentication are performed, the mobile phonesand networks share the authentication keys and theillegal data-modification prevention keys.

DATA AUTHENTICATION. In W-CDMA, functionf8 is used to generate a series of random num-bers, and exclusive logical OR sums are per-formed for each bit of the user data and signaldata to perform the encoding, see Fig. 2. The bitlength for encoding/decoding, the up/down link,the counter, the logical channel identifier, andthe key for data confidentiality are input to thelogic function f8 to generate the sequences ofrandom numbers.

DATA INTEGRITY. This refers to the technologyin which authentication codes are added to thesignal data in wireless communications in or-

KASUMI

SSL/TLSclient

Contents ContentsDownload

Externalinterface

Portabletelephone ServerGround station

gateway

Internet

Contents-level

security

Memory cardUSB

Infrared

Datacommunications-

level security

Wirelesscommunications-

level security

SSL/TLSgateway

SSL/TLSserver

KASUMI

Fig. 1 W-CDMA mobile mobile-phone datasecurity technology

Fig. 2 Function f8 to insure data confidentiality

f8

Counter Up/downlink

Bitlength

Key stream block Key stream block

Sender side Receiver side

Non-encodedblock

Encodedblock

Encodedblock

Logicchannelidentifier

Confidentialitykey

f8

Counter Up/downlink

Bitlength

Logicchannelidentifier

Confidentialitykey

TECHNICAL REPORTS

· 15June 2003

Fig. 3 Function f9 to insure data integrity

f9

Messageauthenticationcode

Counter

Data Randomnumber

Up/downlink

f9

Messageauthenticationcode

Integrity keyIntegrity key

Counter

Data Randomnumber

Up/downlink

Sender side Receiver side

ENCRYPTION ALGORITHM KASUMI. The en-cryption algorithm that forms the core of thedata-confidentiality function f8 and the data-in-tegrity function f9 is known as KASUMI.

The following conditions to be fulfilled whendeveloping an encryption algorithm were definedby the Third Generation Partnership Project(3GPP) consortium, which is researching tech-nical standards for W-CDMA:* Security must be preserved under open speci-

fications.* Packaging restrictions in mobile phones mean

that the algorithm must be implemented inhardware using no more than 10K gates.

* W-CDMA traffic considerations mean thatprocessing must be at 2Mbps.

Because there was only a six-month lead timeavailable to develop the encryption algorithm,it was decided to work on an existing encryp-tion algorithm rather than to develop a new al-gorithm from scratch. A search for existingencryption algorithms fulfilling the requirementsdescribed above showed that the only one avail-able was the MISTY[1] algorithm at MitsubishiElectric Corporation; KASUMI was developedbased on reworking the MISTY algorithm.

Data-Communications Level SecurityAuthentication and confidential communicationsfor wireless telephones generally use the SSL/TLS(Secure Socket Layer/Transport Layer Security)that has become the global standard for securityprotocols between web servers and browsers onthe Internet. However, not all SSL/TLS functionsare implemented in mobile phones: the specifica-tions are a subset of SSL/TLS.

There must also be a security function to guar-

antee the legitimacy of the various contents andmodules downloaded from the network. Here“legitimacy” refers to the data having been gen-erated by the correct provider and not havingbeen illegally modified after it was generated,and refers to a function on the mobile handsetside to verify and confirm legitimacy. The tech-nology by which legitimacy is ensured uses asystem whereby the server side applies a digitalsignature to the data, and the signature is vali-dated on the mobile phone downloading it, thusconfirming the legitimacy.

Contents-Level SecurityResearch into contents-level security is still atan early stage, and standards have yet to be es-tablished. An external memory card that is in-serted into the mobile phone has been developedat Mitsubishi Electric as a copyright-controltechnology.[2] The technology is enabled throughencapsulation using encryption technology.

When the encapsulated contents are storedexternally, they are encrypted with informationthat is specific to the mobile phone. Similarly,when the contents are played back, the contentsare decrypted using the same data, unique tothe mobile phone. This makes it possible to re-strict the playback of contents to the mobilephone on which they are stored.

The security of communications for the thirdgeneration of mobile phones is critically impor-tant for their widespread adoption. MitsubishiElectric is at the forefront of successful effortsto ensure the very highest degree of security atevery level. ❑

References[1]Matsui, et. al.: “Block encryption algorithm ‘MISTY’,” Mitsubishi

Denki Giho, Vol. 72, No. 5 (1998) (in Japanese)[2]Miyazaki, et. al.: “Digital contents distribution system using a

memory card,” Mitsubishi Denki Giho, Vol. 76, No. 4 (2002) (inJapanese)

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE16 ·

*Tomo-oki Ukiana is with the Mobile Terminal Center and Yoshiaki Katayama is with the Information TechnologyR&D Center.

by Tomo-oki Ukiana and Yoshiaki Katayama*

W-CDMA Mobile-Phone SoftwareTechnology

A mobile-phone software platform known ascellular integrated communicator architecture(CINCIA) has been developed. CINCIA is a hy-brid-structure software platform equipped witha real-time kernel based on the µITRON stan-dard, and a dynamic kernel that controls theapplication. It has a three-layer hierarchicalstructure and modular architecture, providing ahigh degree of scalability and reusability.

CINCIA Mobile-Phone Software PlatformW-CDMA mobile phones are compatible with alarge variety of data and video services, in addi-tion to providing audio and data communicationsservices. Because of this, the application softwareprovided with them must have more sophisticatedfunctions, and is larger, than conventional embed-ded system software. This increases the amountof work involved in development and test activi-ties, making it difficult to develop the requiredfunctions and perform quality assurance withinlimited development lead times. The CINCIAmobile-phone software platform was developed toimprove the productivity of software developmentso as to resolve the above issues.

CINCIA has a modular architecture and a hi-erarchical structure, comprising three layers: anapplication layer, a library layer, and a hardware-abstraction layer (HAL). With this, and the seg-mentation of software into modules, CINCIAprovides excellent scalability and reusability, seeFig. 1. W-CDMA mobile phones for whichCINCIA is applicable comprise over 170 mod-ules, and these are packaged for implementa-tion within 7MB of ROM and 3MB of RAM. Thevarious functions are introduced below.

HALHAL, the lowest layer, conceals any differencesin hardware in the individual equipment typesfrom the software at higher layers, ensuring port-ability of the software as a whole. Because thereal-time performance will vary for each softwaremodule, the kernel is a hybrid OS equipped witha real-time kernel known as a configurable real-time operating system (CROS), based on theµITRON standard for controlling real-time tasks,and also equipped with a dynamic kernel that

Non-realtime system

Realtime system

PIM Browser Video

Device driver(s)

Dynamic kernelRealtime task(s)

Realtime kernel - CROS

Application layer

Library layer

Hardwareabstruction layer

File

DB

Graphics

UI Internet

Shell

Phone Audio

Target device

Fig. 1 Mobile-phone software platform CINCIA

adjusts thread priorities dynamically in order tocontrol applications, enabling the complex sched-uling required by mobile systems. Not only arethe basic OS functions such as memory/timermanagement, semaphore and messaging pro-vided, but also power-saving functions, memory-protection functions and other extended functionsrequired in mobile terminals. The use of theµITRON standard, a Japanese industrial standardfor embedded systems, makes it possible to useexisting software developed both within the cor-poration and elsewhere. It also facilitates devel-opment lead-time reductions by making it easierto train engineers.

Library LayerThe library layer, the intermediate layer, pro-vides a variety of services to the higher-levelapplications. The library, which provides theCINCIA, is device independent, making it us-able across a wide variety of mobile phones. Con-centrating the functions in the library makes iteasy to respond to new applications and addi-tional products, and makes it possible to limitthe amount of software that needs to be devel-oped for each type. Furthermore, the various li-brary interfaces are designed in accordance with

TECHNICAL REPORTS

· 17June 2003

industrial standards wherever possible, simplify-ing the development of applications that use thevarious types of services supplied by the library.The primary library functions are described be-low.

The file system has interfaces by which appli-cations can acquire standardized access to theROM, RAM, flash memory, removable media,and other types of storage devices, and has da-tabase functions and registry functions, amongothers.

The graphics system has low-level graphicinterfaces with an off-screen sprite model anduser interface (UI) components enabling cus-tomization of the look and feel for each type ofequipment.

The telephony system supplies functions com-patible with voice communications and a vari-ety of network services.

For the Internet, high-level communicationsprotocols such as TCP/IP and SSL are provided,along with basic functions such as character-set control, along with functions for handlingmessaging services.

The shell controls the launching and termi-nation of applications, the screen order of theapplications that are running, and event deliv-ery.

Application LayerThe application layer, the highest level, includesa variety of applications such as a phone book, abrowser, etc. The UI module of these applica-tions can provide new functions more easily, andprovides for development and customizing foreach type. In CINCIA, this is handled by provid-ing software development tools that increase theefficiency of development and testing of themobile-phone application UI module.

The article has described the approach taken instructuring the CINCIA mobile-phone softwareplatform. The technologies described in this ar-ticle have already been applied to the develop-ment of W-CDMA mobile phones. In the future,as there are software enhancements for newfunctions, and as a variety of development sup-port systems are provided and improved, the de-velopment of high-functionality/high-qualitysoftware will be carried out with even higherlevels of efficiency. ❑

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE18 ·

*Shiro Takada is with the Advanced Technology R&D Center and Atsushi Musha is with the Mobile Terminals Center.

by Shiro Takada and Atsushi Musha*

Mobile-Phone Structural DesignTechnology

W-CDMA mobile phones must be light andhighly reliable, and from the perspective of struc-tural-design technology development, drop-im-pact and heat-dissipation technologies arecritical. This article provides an overview of tech-nology development focusing on validationthrough analytical techniques able to forecastdrop-impact characteristics and temperatureprofiles at the design stage, and technology de-velopment focusing on model experiments. It alsodescribes the application of these techniques toW-CDMA devices.

Development of Drop-Impact DesignTechnologiesThe ability to withstand the shock of impact isimportant in ensuring reliability in the environ-ments in which mobile phones are used. AtMitsubishi Electric Corporation, drop-impactanalytical technologies are being developed toforecast quantitatively, in the early stages ofdesign, the response of equipment to physicalimpacts. This makes it possible to develop ap-propriate structures. The article describes boththe drop-impact simulation technologies beingdeveloped and their application to W-CDMAmobile phones.

A drop-testing machine with a sloping slidesystem, as shown in Fig. 1, has been developedin order to quantify the shock experienced bydevices after freefall. This test equipment is con-structed so that a test sample is dropped whileheld in a test jig in which the holder releasesthe test sample immediately before it strikes thefloor. As a result, the orientation with whichthe test sample is dropped is controlled until im-mediately before impact, making it possible toreproduce “freefall” consistently. This test equip-ment makes it possible to measure the impactload, acceleration and strain during the contact,and the velocity of the test sample immediatelybefore impact.

Straight-type mobile-phone drop-impact analy-sis is described next. In the modeling, the maincomponents of the chassis are assembled froma front and rear case, and the efficiency of theanalysis is increased by using a fine-elementmesh in the vicinity of the impact, and a coarse-

Fig. 1 Drop-testing machine with sloping slidesystem

element mesh elsewhere. In addition, a printed-circuit board, a shield casing, a liquid-crystal dis-play component, a battery, an I/O connector, andother critical components were modeled as in-ternal parts. With the top of the antenna as thepart making the impact, the drop impact analy-sis was performed for a freefall to a rigid floorfrom a height of 1.5 meters (with an impact ve-locity of 5.4m/s).

Fig. 2 shows the results of the above analysis,with the experimental results of an analysis ofthe time-history waveform of the drop-impactload as measured by the drop-testing machinewith the sloping slide system under the sameconditions. Not only does the shape of the graphof the impact load vs. time show the reproduc-ibility of the test, but it also validates the ana-lytical model because it is in close conformitywith the predictions.

The results of applying this drop-impact simula-tion technology to the early stages of W-CDMAmobile-phone chassis design are as follows. Insuch devices, a magnesium alloy is used for the

TECHNICAL REPORTS

· 19June 2003

front case in order to ensure rigidity, and a poly-mer is used for the rear case. It has been shownclearly and quantitatively that the impact loadwhen dropped is greater than it was in the all-polymer chassis of the past. Additionally, simu-lation technology was used to make structuralimprovements such as the optimization of thechassis fit, the locations of the fitting attach-ments, internal reinforcement ribs, etc., as wellas improvements in the shape of the antennaand in the structure surrounding the cameramount.

The typical optimization of an antenna struc-ture is described below. Fig. 3 shows the ana-lytical model used at the early stages of theW-CDMA device chassis-design process, withthe stress distribution indicated by the analysis.

0

0.5

1.0

1.5

0 0.5 1.0 1.5 2.0

Sample No.1Sample No.2Analysis

Impa

ct lo

ad

kN

Time ms

Fig. 2 Time history waveform of impact load

High

Stress level

Low

Model A Model B

Fig. 3 Stress distribution around the top of the Mg chassis

When dropped from the antenna side, thestresses produced in the magnesium chassis varygreatly depending on the shape of the top of theantenna, so the shape of the antenna wasoptmized. The reinforcing ribs were also opti-mized to reduce the strresses that occur aroundthe place to which the antenna is attached.

Development of Heat-Dissipation DesignTechnologiesIt is difficult to estimate increases in tempera-tures in electronic components and on the sur-faces of the device chassis in electronic devicessuch as mobile phones because the thermal gen-eration profiles are non-uniform. Thus, the useof numerical simulation technologies is more ef-fective. A description of the thermal analysisused in the development of W-CDMA mobilephones follows.

Mobile phones, as shown in Fig. 4, are madefrom a printed-circuit board on which the elec-tronic components are mounted, a shieldingcase that covers the transceiver circuits, a liq-uid-crystal module, a battery, a speaker, a re-ceiver, a microphone, keys, and a variety of othercomponents. These items are housed in a chas-sis. The primary heat sources are the electroniccomponents on the printed-circuit board, wherethe heat generated is conducted to the mountedcomponents and air gaps, ultimately to escapeto the ambient air through the convective airflow against the surface of the chassis andthrough radiation from the chassis. Note thatthe components are tightly packed inside themobile phone, and thus the effect of heat trans-fer through natural convection is minimal, and

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE20 ·

analytical values at locations between the struc-tural parts of (a) and (d). Between (a) and (b), thethermal flux from the surface of the heater tothe back surface of the printed-circuit board waslarge, and difference arising from the state ofcontact at the various locations appear to besubstantial. Furthermore, between (b) and (c),there was an effect from the ribs on the LCDholder in the LCD module, formed as it was froman air gap, a glass panel, a polarizing plate, anLCD holder, and other components. Locationsbetween (c) and (d) are between the surface ofthe glass panel in the liquid-crystal module andthe surface of the front case, where differencesin the thermal diffusion effect appeared second-arily.

Consequently, the temperature differentialsbetween the internal components can be cor-rected through, for example, configuration analy-sis when packaging, and measurements of thematerial properties. The effectiveness of thisanalytical technique is clear.

The analytical model developed was appliedto a W-CDMA mobile phone, and not only wasit possible to select the component layout thatshowed the best distribution of heat generationat an early stage of the design but also heat-dissipation strategies using heat spreaders andhigh-thermal-conductivity rubber could be em-ployed.

As mobile phones become more sophisticated,more aggressive development of mechanicaldesign technologies will become necessary, inwhich the mounted components become largerand the external chassis becomes lighter andthinner, and in which designs with efficient ther-mal dissipation will be suitable for increasinglylarge amounts of generated heat. ❑

the effects of thermal radiation are also smallwhen compared with thermal conductance. Asa result, evaluations were performed taking onlythermal conductance into account.

Finite-element analysis was used as the nu-merical technique required in the design. Anaverage value for the heat-transfer coefficientdetermined experimentally for convection to theair and thermal radiation from the surface of thechassis was used. Furthermore, measurementswere made of thermal conductance for the mul-tiple compound components in the structure,such as the printed circuit board, and the modelwas simplified by using these individual compo-nents assumed to be made of anisotropic mate-rials.

Experimental equipment with simulated heatgeneration was used in which the conditions ofthermal generation could be set explicitly for astructure in which dummy elements, copperplates, and heaters were stacked on a printed-cir-cuit board. This was used to perform validationexperiments. The temperature measurementswere performed for the surface temperature of themounted components from the front case throughto the rear case on the central axis of the dummythermal-generation elements, as shown by thedotted line in Fig. 4.

Fig. 5 shows the experimentally-derived andanalytically-derived thermal distributions. The

Microphone

Key

Front case (d)

LCD module (c)

Shielding caseRear case

Battery

Heat generatingdevice (a)

Principal heat flow

PCB (b)

Fig. 4 Schematic cross-sectional diagram of amobile phone

Dim

ensi

onle

ss te

mpe

ratu

re

Location along the thickness

Rear case Front case

ExperimentAnalysis

(a)

1

0(c) (d)

(b)

Fig. 5 Temperature distribution along thethickness of a mobile phone

solid line shows the experimental values and thedotted line the analytical values. In the figure,(a), (b), (c), and (d) show the measurement points(a), (b), (c), and (d) on the component surfaces inFig. 4. Fig. 5 shows the excellent agreementbetween the analytical and experimental valuesfor the temperatures on the surface of the chas-sis, and that it is possible to estimate accuratelythe temperatures in the actual mounted com-ponents. On the other hand, there were discrep-ancies between the experimental values and the

TECHNICAL REPORTS

· 21June 2003

*Yoshiko Hatano is with the Advanced Technology R&D Center

by Yoshiko Hatano*

W-CDMA Mobile-Phone ImagingTechnology

One of the features of W-CDMA mobile phonesis their videophone functions, able to receivevideo content. In order to provide these func-tions, a camera signal-processing integrated cir-cuit has been developed to transmit the capturedimages, along with a display controller for syn-thesizing the display of video images and graph-ics on both transmitter and receiver sides.

Fig. 1 shows the flow of signals in the imagingsystem. First the image captured by the cameramodule is input to the camera signal-processingIC and after image processing (such as zoom,special effects, etc.) the signal is MPEG-4 en-coded. The encoder’s data-stream output is mul-tiplexed with the speech data, and sent on tothe radio unit.

The raw image, prior to MPEG-4 encoding, isalso output from the camera signal-processingIC and sent to the display controller. The dis-

play controller combines this pre-encoding rawimage and the video-image output from theMPEG-4 decoder, overlaying any still imagegraphics generated by the CPU, and sends theresults to the LCD. The various parts of the sys-tem are as described below.

Camera Signal-Processing ICA block diagram for this unit is shown in Fig. 2.The input image is the QCIF size (176 pixelshorizontal by 144 pixels vertical), with eight bitseach for R, G, and B colors. Electronic zoom andspecial effects are performed on the capturedimage, and still-image graphics are overlaid. Thestill image graphics overlay is more than justthis, generating a “hold” screen when, duringthe videophone session, the transmission of thelocally generated image is paused.

The video encoder offers selection of MPEG-4Simple Profile Level 0 and H.263. When using

BitstreamYCbCr RGB

Transparent/Semitransparent

Overlayframe memory

(176x144)

4:2:0 4:4:415Hz

176x144RGB(8,8,8)

Cameramodule

To display controller

Effects ZoomCameramoduleinterface

Video encoderMPEG4H263

Downsampling MATRIX

Fig. 2 Block diagram of camera signal processor LSI

Fig. 1 Total block diagram for video-system processing

Graphics 2

MUX /DEMUX

MPEG-4Stream

Camera Signal Processing LSI (M64144)

MPEG-4decoder

MUX : MultiplexerDEMUX : Demultiplexer

Cameramodule

Video 1

Video 2

Display controller

LCD

MPEG-4 encoder

Imageprocessing

Mix /overlay

Graphics 1

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE22 ·

MPEG-4 encoding, data partitioning and revers-ible variable-length codes (RVLC) are used toachieve error resilience throughout the video-phone session. Furthermore, to reduce delaysduring the videophone session, the system ex-erts control to ensure that no frame occupiesmore than 512 bytes of the buffer.

The most time-consuming process in the videoencoder is motion estimation. The IC uses amotion-vector detection algorithm that deter-mines a search area adaptively based on thesurrounding motion vectors, thereby reducingthe average calculation load to 27.2% of that inprevious systems. Furthermore, a low powerconsumption of only 50mW has been achievedby optimizing the partitioning of tasks betweenhardware and firmware, and by the use of a gatedclock. The features of this camera signal-pro-cessor IC are summarized in Table 1.

Display ControllerThe display controller has a CPU interface, anMPEG-4 decoder interface, a camera-signal in-terface, and an LCD-module interface, and isable to synchronize graphics and video images,partition areas as desired, generate mixed im-ages, and display the images on the LCD mod-ule. Fig. 2 shows display-controller features,while Fig. 3 shows the system configuration.

Because two video streams are input asynchro-nously, the display controller has two framememories for video, where the timing is adjustedby switching between writing and reading.

The video and graphics synthesis is synchro-nized with the completion of the writing for eachimage, where the timing is adjusted to outputto the LCD only the regions to be written,thereby reducing the power consumption.

The above article summarizes the imaging tech-nology used in W-CDMA mobile phones. The video

communications functions of these mobile phonesseem sure to advance still further in the future.This will make it necessary to improve the perfor-mance of cameras, LCDs, and video-processingICs, and to reduce their power consumption. Fur-thermore, the graphic functions in the mobilephones must be easy to operate, fun to use, andstable whatever the environment in which theyare used. Mitsubishi Electric is therefore commitedto ongoing research in imaging technologies forthese types of application. ❑

Display controller

Camera signalprocessing IC

output176 x 144

176 x 216

Camerafader

Mixframe

memory

3/4resizer

CPU I/F

132 x 162 132 x 162

Graphicsframe

memory2

Graphicsframe

memory1

1/2SEL

1/2SEL

SEL

VRAM

LCDunit

132 x 162LCDI/F

Videoeffect

&inversematrix

MPEG-4Decoder outputQCIF:176 x 144SQCIF:128 x 96

Y,Cb,Cr

CPU BUS

Fig. 3 Clock diagram of display controller

Table 1 Features of the Camera Signal-Processor IC

Input video QCIF (176 x 144 pixels), RGB (8 x 8 x 8 bits)

Electrical zoom 1.5X/2X

Image effects Posterization, monotone, sepia, negative

Still picture overlOverlaying image: 176 x 144 pixels,RGB (5 x 5 x 5bit)Transparent, semitransparent

MPEG-4 encoder

Simple Profile Level 0Bit rate: 64kb/s (max)Frame rate: 15Hz (max)Data partitioningReversible VLC

Power consumption Approx. 50mW

Package 175-pin FBGA

Table 2 Features of the Display Controller

Input Video 1: QCIF (176 x 144 pixels)Video 2: sub-QCIF (128 x 96 pixels)Graphics: 132 x 162 pixels

Output 132 x 162 pixels, 262,144 colors, LCD format

Functions Level control, Mirror image, Rotation for Video 1, 2.Size conversion by 3/4 for Video 1, 2.Still image capture and still image display for videoPicture in picture with Video 1, 2.Overlay graphics 1, 2 on Video 1, 2.Transparent, semitransparent.

TECHNICAL REPORTS

· 23June 2003

*Yoshihito Nakahara and Tomoaki Tsukada are with the Industrial Design Center.

by Yoshihito Nakahara and Tomoaki Tsukada*

W-CDMA Mobile-Phone Design

A W-CDMA (Wideband-Code Division MultipleAccess) service has been developed by NTTDoCoMo. The service provides high speed/highquality communications, prompting changes inthe ways in which mobile terminals are used,and changes in the design of mobile phones. Thisarticle describes an image of the future offeredby this W-CDMA service to its users, identifiesnew concepts, and addresses the issues of mo-bile-phone design and a new GUI for them. Thekey concepts in the new design are as follows:

Proposal: Double ScreenOne proposed form of the mobile phone, using adouble screen, is presented as a multimedia tool,conceived as able to handle more sophisticatedservices and capable of executing multiple ap-plications simultaneously.

Proposal: Separable TerminalsAn evolved form of mobile terminal is presentedas a communications tool, demonstrating thefeasibility of multitasking and the use of the

mobile terminal while engaging in other activi-ties.

As a result of the design work, prototypemockups were exhibited in Germany at both the2001 and 2002 CeBIT, where they were seen notsimply as studies in styling but as definite stepstowards identifying specific advances and point-ing the way to the next generation of mobileterminals.

Envisioning the Types of Use in the FutureVideo contents, video mail, and videophoneservices are at the heart of multimedia servicesin W-CDMA today.1. In the introductory phase, the technologies

will develop more varied and sophisticatedfunctions.

2. During the growth phase, as the degree ofdiversity of integrated services increases, wecan expect to see increasingly large variationsand changes in the terminal itself.

3. In the mature phase, as the terminal becomesubiquitous, there will be advances in the

Fig. 1 The mobile-phone concept mockup displayed at CeBIT 2001

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE24 ·

direction of wearable and customizableterminals.

This article discusses the double-screen concept,envisioned for Stage 1, and the separable terminalconcept envisioned for Stage 2.

Feasibility of the Double-Screen DesignThe double-screen design is a prototype basedon the concept of providing more sophisticatedservices.

BACKGROUND. There are calls to increase thescreen area in the pursuit of more easily legibleliquid-crystal displays that can display a largeramount of text in mobile phones. Note that thetypes of media handled by mobile phones are di-versifying, as shown in Fig. 2.

Fig.2 The wearable concept mockup

One way of fulfilling the dual objectives ofincreasing screen area (in order to cope with thesophisticated functions that will result from thewider range of media handled) while maintain-ing the mobile phone’s compact form factor isto link together two screens.

THE DOUBLE-SCREEN CONCEPT. This can besummarized as a multilayer compilation com-munications terminal. The concept is to makeit possible for users to access sophisticated func-tions and services without having to go throughcomplicated operations. This will contribute highadded value in the form of linked media com-munications that will accompany increasingly

sophisticated communications. These are thebasic desires of mobile-phone users. The design-ers focused on an effective method of using asecond layer screen, where the screens slide andare displayed only when operations require theiruse. Applications envisioned as using these sec-ond-layers screens include:1. Checking or manipulating attached images

when replying to mail.2. Displaying information about stores, restau-

rants, etc., on the map while using naviga-tion functions.

3. Displaying, at the same size, both the image ofthe user and the image of the person to whomthe user is talking when using the videophonefunction.

4. Playing competitive games while watchingthe opponent’s face.

GUIS FOR DOUBLE SCREENS. The interface struc-ture is being studied based on the following threescenarios:1. Mail functions and browser functions are

linked together.2. Videophone functions and other functions are

linked together.3. Navigation functions and local information

search functions are linked together.

In these studies, the basic requirements are:1. That the basic operations can be performed

even on a single screen, and2. That the interface shifts to the double-screen

mode as and when necessary.

Voice

Text

Music

Photos

Video

VoiceText

MusicPhotosVideo

Downloaded applications

Past

Com

mun

icat

ions

leve

lLo

wH

igh

Present Future

Downloaded music

Linked media

Fig. 3 Increasing media on mobile phones

TECHNICAL REPORTS

· 25June 2003

Envisioning these basic requirements makesit possible to satisfy the needs of both basic us-ers and those who take advantage of more so-phisticated communications functions.

Given this, the guiding principle for the GUIis that of a “support display” that can clearlyshow the superiority of the shift to the double-screen mode. In this proposal, the two types ofuser support come down to specific examples ofmaking it possible to multitask clearly distinctapplications, making it easy to find other infor-mation required in the various operations, andmaking it possible to arrive at one’s objectivewith a minimum of operations.

The double-screen concept also has the ad-vantage of increasing the degree of freedom indesign, maintaining portability and power-sav-ing, in additional to its benefits for user inter-faces. The double-screen approach seems sureto play a powerful role in differentiating suchunits from other mobile phones.

These results attracted a great deal of atten-tion when exhibited at the information deviceexhibition in March of 2002 (CeBIT2002). Beforethe exhibition, studies had been performed toevaluate the degree to which the concept wouldbe accepted by the market.

The design of future devices (phones and ter-minals) will depend critically on studies of themarketability of new devices and interface de-sign will be difficult.

Developments in Separable DesignSeparable design builds on a design concept en-visioning multichannel services in the future,and has been validated in the creation of imagevideos depicting use scenarios.

BACKGROUND. When multichannel services areused in the future, it will be possible to operatethe terminal equipment while multitasking--looking at the display while listening to the au-dio or while taking pictures. Because of this, theseparable design approach has been proposed asa structure with even more flexibility than thedouble-screen design.

THE SEPARABLE DESIGN CONCEPT. This conceptprovides flexibility in the way the device is heldby providing independent units, with a handset

Fig. 4 Concept mockup demonstrated at CeBIT2001

Fig. 5 Separate type mockups

for phone conversations and a viewer for han-dling large volumes of data. The scenarios foruse include phone or videophone conversations,while sharing image information, holding theviewer in one hand and the handset in the other.Stackable units and pendent-type units are alsoenvisioned, with portability in mind.

EXAMPLES FEATURING SEPARABLE DESIGN. Firsttype: Stackable units, see Fig. 4. Second type: Aunit that can be split into two from a shape foldedinto two parts, where the handset part insertsinto the viewer, see Fig. 5 and 6.

TECHNICAL REPORTS

Mitsubishi Electric ADVANCE26 ·

Future OutlookAlthough continuing proposals and research fromthe users’ perspective will be needed in the fu-ture, ultimately it will be necessary to considerways of identifying specific hardware and soft-ware proposals. For example, proposals have al-ready been made for how to implement functionsin the double-screen and separable hardwarestructures. Even for past products, wheneverlinks have been made between distinctive oper-ating devices and GUIs, those links have becomethe distinguishing product features.

Knowledge of both hardware and softwaretechnologies will be essential in creating mar-ketable designs for information equipment suchas mobile phones. It will also be important forhardware and GUI design to combine in mak-ing the product appeal to the users in a varietyof ways. This presupposes proposals and researchfirmly rooted in human experience.

The article discussed two approaches to W-CDMAmobile-terminal design developed from future usescenarios. The double-screen design, is particu-lar, is eminently feasible from the perspective ofimprovements to the operating interface, and theproposal has been subjected to thorough research,including that into the associated GUI. However,packaging constraints present a variety of techni-cal issues to be resolved in marketing such a prod-uct. These include the issue of size and of wiringwithin moveable parts. Further development workwill be performed in cooperation with the engi-neering design department. ❑

Fig. 6 Alternative separate type mockup

MITSUBISHI ELECTRIC CORPORATIONHEAD OFFICE: MITSUBISHI DENKI BLDG., MARUNOUCHI, TOKYO 100-8310, FAX 03-3218-3455