Direct Conversion Transceiver Design for Compact
`Low-Cost Portable Mobile Radio Terminals
`
`A.Bateman & D.M.Haines
`Centre for Communications Research, Queens Building
`University of Bristol, University Walk
`Bristol. BS8 lTR, U.K.
`
`Abstract
`
`The full benefit of using linear modulation methods for the
`transmission and reception of data in the mobile environ-
`ment will only be realised when compact and power effi-
`cient linear transceiver equipment becomes a commercial re-
`ality. Moving from superheterodyne to direct conversion
`transceiver architectures is seen as a major step towards
`achieving this goal. Several characteristic problems with di-
`rect conversion technology have however hindered progress
`in this area.
`This paper describes methods for overcoming many of
`the difficulties associated with direct conversion architec-
`tures and presents results from a working prototype devel-
`oped by the Centre for Communications Research at Bristol
`University.
`
`Introduction
`
`The availability of high quality mobile telephone systems
`has proven to be an invaluable resource in many facets of
`business and social life. Yet, despite the economic benefits
`of a mobile phone, widespread acceptance has been ham-
`pered from the user perspective by the short battery life,
`terminal size and cost, and from the operator perspective by
`the shortage of spectrum, hindering service expansion. The
`spectral efficiency of radio communication systems is central
`to the future of mobile communications industry. It is pre-
`dicted that in the UK alone, the present cellular telephone
`population of 1.5 million users will rise to 10 million by the
`year 2000 [I].
`Relatively inefficient use of spectrum by the present FM
`system has manifested itself as crowding on the available
`channels. This is particularly prevalent in cities and on sec-
`tions of motorways at times of peak demand. Overcrowding
`in the present UK TACS system results in long delays for
`connection and (occasionally) call termination. Battery life,
`whilst of less importance in vehicle-based equipment, has
`proved a decisive factor in the serviceability of hand held
`units. Present hand-held designs are a compromise between
`transmission range and battery life, and their performance
`falls short of similar vehicle-based transceivers. Also, even
`the smallest hand held units require a larger than average
`pocket.
`
`Future public land mobile telecommunication systems
`(FPLMTS) will be required in both vehicle-based and hand-
`held form. The challenge of meeting the needs of power
`and spectrum efficiency in a compact package is a hard one.
`However, if any FPLMTS is to be truly universal and widely
`available, the eventual goal of a pocket-sized “personal com-
`municator” must be realised. This has been recognised by
`several working groups [2,3], and a target of 200g weight
`and 200cm3 volume has been suggested [4]. There is also
`some consensus that a FPLMTS should be based on low bit-
`rate (16 kbits/s or less) voice technology. The transceiver
`systems employed in any FPLMTS must be power efficient
`and suitable for use with data modulation. In addition, the
`modulation scheme adopted must be as efficient in its use of
`spectrum as possible.
`Previous work at the Centre for Communications Re-
`search at Bristol [5,6,7] has shown the arguments for a lin-
`ear modulation (LM) scheme being one of the most spec-
`trally efficient. Other work [S,9] has supported these ar-
`guments. Yet existing LM equipment of sufficient quality
`is bulkier at present than comparable FM equipment, and
`more power hungry. To utilise the obvious advantages of LM,
`a departure has to be made from conventional transceiver
`designs. A direct conversion transceiver system eliminates
`many of the space critical components found in superhetero-
`dyne designs, and allows the engineer much scope for in-
`tegration. There are, however, severe problems with this
`approach [10,11]. This paper shows how one type of direct
`conversion transceiver has been developed at Bristol to the
`point where a prototype is under evaluation. Test results
`are provided to show the performance of this transceiver to
`be equivalent to (and in some cases better than) existing
`systems.
`
`The Transmitter
`The requirements for a transmitter in a LM system are that
`it should have adequate output power, low out-of-band emis-
`sions and (if possible) be power efficient. The last two at-
`tributes are incompatible in conventional transmitter power
`amplifier designs. Linear amplifiers generally have large qui-
`escent currents, and efficient Class C amplifiers are highly
`non-linear.
`One method of meeting all the above requirements is the
`
`57
`
`CH2379-1/89/0000/0057 $1.00 1989 IEEE
`
`TCL EXHIBIT 1047
`Page 1 of 6
`
`

`
`RF Amdifw
`
`U1
`
`C l a s s A/AB
`
`I
`
`C
`
`I
`
`I
`
`RF Loco1
`osciltotor
`
`Figure 1: Cartesian Loop Transmitter with Phase Shifter
`
`Cartesian Loop transmitter [12,13]. Figure 1 shows the en-
`tire transmitter in a Weaver method [14] configuration. This
`method was adopted as it allows image suppression which is
`superior to other techniques. It is far from being a new idea,
`but its development has been limited in the past because
`of three problems; loop stability, generation of accurate I
`and Q signals and maintenance of quadrature on the local
`oscillators.
`These problems have been overcome by several develop-
`ments. Firstly, loop stability is governed primarily by the
`phase shift needed between mixer pair AB and mixer pair
`CD. This phase shift corresponds to the delay through the
`power amplifier. Maintenance of the correct phase shift is
`therefore essential for spurious-free operation. A voltage-
`variable r.f. phase shifter B has been incorporated into the
`design. This enables the relative phase of the two sets of
`local oscillators to be varied over a wide range, and also over
`a wide range of frequencies.
`A related network has been developed to provide accurate
`quadrature for each set of local oscillators. Fine adjustment
`is possible by means of a voltage level, and quadrature can be
`maintained over a large bandwidth with minimal amplitude
`fluctuation.
`Finally, a digital signal processing (DSP) system enables
`the production of very accurate I and Q baseband signals.
`This also allows control and calibration of the transmitter,
`during an initial calibration period and subsequently during
`operat ion.
`The results of these developments are dramatic (Fig-
`ure 2). An existing LM amplifier, biased in Class A or AB
`mode is palpably worse than the linearised Class C ampli-
`fier. Output powers of 20W PEP have been obtained at
`VHF with excellent suppression of spurii up to the band-
`width of the differential amplifiers. This may be termed the
`linearising bandwidth. This is different from the transmit-
`ter’s bandwidth of operation, which is dependent upon the
`quadrature local oscillator networks, the phase shift network
`and the amplifiers. This can be very wide (several tens of
`MHz). There are, however, limits on the Cartesian Loop
`Transmitter’s linearising bandwidth, governed primarily by
`the phase delay within the feedback control loop. With too
`wide a loop bandwidth or too high a loop gain, instability
`will occur. Linearised bandwidths of several hundred kHz
`have so far been achieved with this technique.
`The above results have been reproduced at UHF (900
`MHz). The nature of components at these frequencies means
`
`C l a s s C
`
`C l a s s C
`( line a r i s e.d )
`
`Figure 2: Outputs from different VHF amplifier types with
`a 4 kHz notch-filtered noise modulating signal
`
`that the resulting transmitter may be even smaller in size
`than its VHF counterpart.
`The gains in power efficiency are equally remarkable.
`Whereas a conventional LM amplifier (for 20W output) may
`have a quiescent current of about lOOmA at 12V, the cor-
`responding current in a linearised Class C amplifier is zero.
`LM is also generally more power efficient than FM, as Fig-
`ure 3 shows. Here an FM system with a Class C power
`amplifier is compared with an LM system with a linearised
`Class C amplifier. Where the modulating signal has a high
`peak-to-mean ratio (such as speech) the overall gain in power
`efficiency is impressive.
`A final advantage is that the Cartesian Loop Weaver
`method transmitter uses components which may all be re-
`duced in size to chip or hybrid level integration.
`
`.
`e-1
`- 5
`
`b
`
`FM
`
`(v
`0 U
`c ’
`
`.- z
`$?
`Y 0 n
`
`U
`
`U
`
`1
`
`l l l B
`sinqle 2 equal random
`tone
`tones
`noise speech
`Figure 3: Comparison of battery drain (in amps) for an FM
`Class C amplifier and for an LM linearised Class C amplifier
`with different modulation types
`
`TCL EXHIBIT 1047
`Page 2 of 6
`
`

`
`The Receiver
`
`A Weaver architecture has also been employed in the receiver
`section. There are severe difficulties inherent in this type of
`receiver [15], but they have largely been solved by careful
`design and advanced DSP techniques. These difficulties will
`be considered together with the requirements for sensitivity,
`dynamic range and carrier feedthrough.
`
`LM
`
`U
`Preamplifier
`
`oscillator 4
`
`RF local
`
`L
`
`Low-noise Anti-olios
`amplifier
`filter
`
`Dual
`16-bit
`0
`converte
`and
`DSP
`
`Sensitivity
`Determining the sensitivity requirement for a receiver is not
`straightforward. It is desirable to have system sensitivity as
`high as possible, but not to the point where other aspects of
`receiver performance are degraded, particularly strong signal
`handling. A guide to necessary sensitivity may be obtained
`by looking at two comparable radio sets; an Aerotron ACSB
`Pioneer 1000 system (Set l), and a Securicor T530 FM Mo-
`bile (Set 2). Their relative sensitivities (for 12 dB SINAD)
`are given in Table 1. From these values, it is clear that a
`sensitivity of around -120 dBm is necessary [18].
`The front-end of the direct conversion receiver is shown in
`Figure 4. This compares with the front-end of a superhetero-
`dyne (superhet) receiver shown in Figure 5. A noise figure
`summary for a typical superhet receiver is given in Figure 6,
`for a sensitivity of -120 dBm (noise figure of 7.2dB). The
`top portion of the figure shows noise temperature and gain
`data for the individual stages. The bottom portion shows
`how the total noise figure of the receiver is built up from the
`individual stage data. Equations for deriving noise figures
`in cascaded stages are well-documented elsewhere [19] .The
`output of the IF strip is used as a reference - it is assumed
`that after this stage little noise will be added. A correspond-
`ing noise figure analysis for the direct conversion receiver is
`shown in Figure 7. Here the input to the A/D converter is
`used as a reference.
`This analysis shows that the direct conversion receiver
`relies heavily upon the preamplifier for even this modest sen-
`sitivity requirement. The presence of the signal splitter and
`passive mixers at such an early point in the signal path is un-
`desirable. However, their inclusion (instead of active mixers)
`was necessary for improved intermodulation performance.
`The analysis also demonstrates that the audio amplifiers are
`pushed to the limits of their noise capabilities.
`Future
`The required sensitivity can just be achieved.
`work will include the evaluation of higher quality active mix-
`ers and a very low noise audio preamplifier to enhance sen-
`sitivity.
`
`Transceiver
`Aerotron 1000 with preamplifier
`without preamplifier
`
`Securicor T530
`
`Sensitivity (dBm)
`-125
`-119
`-120
`
`Figure 4: Digitally-implemented Weaver Direct Conversion
`receiver
`
`Preomplifier
`
`I----
`
`IF strip and filter
`
`Figure 5:
`Single-conversion Superhet Receiver front-end
`components
`
`Individual Stages
`Noise Temperature (K)
`Cain (ratio)
`
`Noise Figure (dB)
`. .
`Gain (dB)
`
`290
`
`5
`
`3
`
`7
`
`2100
`
`0.1
`
`9.2
`
`-10
`
`Bandpass
`Filter
`
`Mixer
`
`230
`
`30
`
`2.5
`
`14.8
`
`900
`
`1
`
`6.1
`0
`
`IF filter
`
`C u m ula tive
`Noise Temperature (K)
`Goin (ratio)
`
`Noise Figure (dB)
`
`1230
`
`15
`
`7.2
`
`Goin (d8)
`
`11.8
`
`260
`4700
`- c -
`3
`30
`
`900 e-
`1
`
`12.4
`
`4.8
`
`2.8
`
`14.8
`
`6.1
`
`0
`
`Figure 6: Superhet receiver noise figure analysis
`
`Individual Stages
`Noise Temperature (K)
`
`Gain (rotio)
`
`Noise Figure (dB)
`
`Goin (d8)
`
`170
`
`100
`
`2
`
`20
`
`4306
`
`0.063
`
`12
`
`-12
`
`2828
`
`31 6000
`
`10.3
`
`55
`
`1.1 4x1 09
`1
`
`69
`
`0
`
`Cumulative
`Noise Temperature (K)
`6435
`106458
`e - - -
`31 6000
`Goin (ratio) 1 . 9 9 ~ 1 0 ~
`19908
`
`1234
`
`Noise figure (dB)
`
`Cain (dB)
`
`7.2
`
`63
`
`25.6
`
`43
`
`13.6
`
`55
`
`1.1 4x1 09
`
`1
`
`69
`
`0
`
`Figure 7: Direct Conversion receiver noise figure analysis
`
`59
`
`TCL EXHIBIT 1047
`Page 3 of 6
`
`

`
`I
`
`I
`
`,
`
`
`
`I
`
`I
`
`I
`
`I
`
`~~
`
`I
`ComDonent
`I Preamplifier I Mixers I Amplifier/Filter I A/D Converter
`.
`+ 23
`. ~-
`Maximum Signal /dBml 1
`--
`I
`1
`I
`-20
`-20
`4-23
`Minimum Signal (dBm) I
`I
`I
`[ -130
`-60
`-115
`-120
`Table 2: Dynamic ranges for Direct Conversion Fkceiver
`components
`
`I
`
`20
`
`-20
`
`-40
`
`-60
`
`-80
`
`-100
`
`-120
`
`n
`
`Dynamic Range
`There are two aspects to the dynamic range of a receiver.
`Firstly, there is the range of signals in the wanted band that
`the receiver is required to tolerate. This may be as high as
`120 dB. Secondly, there is the largest ratio of adjacent signal
`to wanted signal that can be processed before third-order
`distortion products swamp the wanted signal. This might
`reasonably be 70 to 80 dB [16] and is termed the spurious-
`free dynamic range (SFDR). Into this area also comes the
`question of adjacent channel rejection and selectivity.
`SFDR will be considered first. It is affected by every
`component in the signal path. Referring to the direct con-
`version receiver front-end (Figure 4), each component can be
`assigned a dynamic range. That is, it can be given a range
`of signals which can be amplified without distortion (at the
`high end) or without disappearing into noise (at the low
`end). These are summarised in Table 2. Figure 8 shows how
`an incoming signal range from -120 dBm to 4 0 dBm can
`be translated to -67 dBm to 23 dBm for a 16-bit analogue
`to digital (A/D) converter. SFDR is directly applicable to
`A/D converters as well. The signal-to-noise (S/N) ratio of
`an n-bit A/D converter is given by [17]:
`
`[$] =4.8+6n
`
`dB
`For a 16-bit converter, the S/N ratio is approximately 96 dB.
`This does not take account of thermal noise. It also does not
`correspond to dynamic range. The smallest signal that may
`be accomodated 12 dB above the noise floor is 96 - 12 = 84db
`below the full-scale signal. This assumes a perfectly ideal
`converter. Until recently, 16-bit converters at reasonable
`cost achieved, in practise, only 14 or 15-bit resolution. The
`advent of self- calibrating converters which provide true 16-bit
`resolution with very low differential non-linearity has made
`the construction of this receiver possible. F'uture oversam-
`pling converters which require minimal anti-alias filtering
`should enable the receiver to be simplified further. The point
`has now been reached where the linearity of the receiver is
`determined by other receiver components, in particular the
`audio amplifiers.
`The direct conversion receiver developed uses digital sig-
`nal processing (DSP) to perform several tasks usually un-
`dertaken by analogue circuitry. Adjacent channel filtering,
`automatic gain control (AGC), squelch control and demod-
`ulation are all carried out on one DSP chip. In particu-
`lar, digital adjacent channel filtering is a significant advance
`as it allows linear phase low-pass filters to determine the
`channel characteristic. Existing LM (and FM) equipment
`suffers from the inclusion of a crystal IF filter. The gain
`and group delay characteristics of a typical 10.7 MHz IF fil-
`ter, compared with similar characteristics for a digital filter,
`are shown in Figure 9. It is clear from these graphs that
`the digital filtering provides a much better behaved channel
`characteristic, particularly suited to data transmission.
`In addition, digital channel filtering provides well-defined
`
`- -
`
`(-1
`
`-60
`-so
`
`-100
`
`-120 I i j
`
`Figure 8: Translation of dynamic range from input to output
`in the Direct Conversion Receiver
`
`0.4
`
`0-2
`
`0
`-0.5
`-1.5
`-1
`Frequency off&
`
`0.5
`1.5
`1
`fmm conin (kHz)
`
`2
`
`...
`
`Filter
`-tal
`Digital Filter
`
`-40
`-5a
`._
`-711
`4 . 5 i i i . 5 -1 -0.5 0 0:s i 1:;
`Frequency o H r t fmm c a A r (kiiz)
`Figure 9: Comparison of Gain and Group Delay Character-
`isitcs for Digital and Crystal Channel Filters
`
`i 2 5
`
`adjacent channel performance. Figure 10 shows the selectivity'
`curves for the direct conversion receiver, set 1 and set 2.
`n o m these curves, it is clear that the direct conversion re-
`ceiver has both a tighter and better defined selectivity per-
`formance than either of the other radios.
`The direct conversion receiver has, as required, an SFDR
`of about 80 dB. This SFDR may be located between 0 and
`-120 dBm by varying the gain of the preamplifier. Under
`software control, the gain of the preamplifier may be ad-
`justed from -20 to 20 dB, and thus the overall dynamic range
`of 120 dB is achieved.
`
`'The input signal level is set to give 12dB SINAD. An interfering
`signal is then introduced at a variable frequency offset, and its level is
`adjusted to degrade the wanted signal to 6dB SINAD
`
`60
`
`TCL EXHIBIT 1047
`Page 4 of 6
`
`

`
`- Direct Conversion
`
`......
`
`Set 1
`
`Set 2
`
`Interfering Signal level (dBrn)
`
`-20
`-30
`-40
`
`-50
`
`-60
`-70
`
`-80
`-90
`
`-100
`-1 10
`,
`,
`,
`,
`I
`,
`I ~ l
`,
`I ~ I
`,
`,
`,
`I
`,
`I
`,
`I ~ I
`-120 ~ ,
`0 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3 1 4 1 5
`Int. Signal offset from centre frequency (kHz)
`
`,
`
`I
`
`,
`
`,
`
`,
`
`j
`
`
`
`Figure 10: Adjacent Channel Rejection responses for the
`receivers under test
`
`A related problem is that of gain and phase matching
`in the I and Q signal paths. Accurate quadrature of the
`local oscillators is maintained by a network similar to that
`outlined in the transmitter section. The amplitudes of the
`I and Q signals are compared in software and any required
`adjustments made on a continuous basis. These techniques
`have the effect that image suppression of 40dB or better is
`maintained at all times.
`
`Carrier Leakage
`Carrier leakage and feedthrough onto the I and Q signal
`paths results in a d.c. level being present. Also, the audio
`amplifiers and anti-alias filters may contribute some d.c. off-
`sets. If fed through to the second mixing stage, this d.c. level
`appears as an undesirable tone in the centre of the band.
`Three methods of addressing the carrier leakage/d.c. off-
`set problem have been identified 115). The most promising
`of these is d.c. correction - a software method whereby the
`incoming signal is averaged over a relatively long period, and
`the result subtracted from the signal. This method has been
`implemented, and works to a large extent. It is a substan-
`tial improvement over a.c. coupling for two reasons. If the
`I and Q paths are a.c. coupled at (say) a cut-off of 50 Hz,
`
`then a significant amount of information is lost in the notch
`created [lo]. If the coupling is reduced to 5 Hz, the notch is
`narrower, but the group delay characteristic of the a.c. cou-
`pling filter adversely affects the channel characteristic. The
`DC correction technique differs from the above in that it is
`essentially a discrete system, correcting for d.c. error only
`at specific instants in time. The result is that the group de-
`lay response of the receiver is thus left largely unaffected. A
`notch is still created, but it can be made very narrow without
`introducing significant delay distortion. In a pilot-based sys-
`tem, such as TTIB, where the pilot is nominally in the centre
`of the frequency band (and thus appears at d.c. in a Weaver
`demodulator), the effects of the d.c. nulling notch can be
`overcome by introducing a small frequency offset (10 Hz or
`so) into the downconverted signal.
`The d.c offsets on audio amplifiers and anti-alias filters
`are unfortunately not constant, and even very small varia-
`tions have a significant effect if the incoming signal is also
`very small. Carrier leakage too is not constant, depending on
`environmental and circuit effects. Correction for d.c. offsets
`therefore has to occur at a rate sufficient to counteract the
`change. At present, correction is used in the direct conver-
`sion receiver at approximately 2-3 times per second. Careful
`front-end redesign and the use of low-drift amplifiers should
`allow this rate to be reduced at least ten times, with the re-
`sult that an effective notch width of < 1 Hz is experienced.
`Other techniques, including local oscillator dither to reduce
`the d.c. content of the down converted input are under in-
`vestigation. A combination of d.c. correction and dither
`may well prove to be the most practical solution.
`
`Receiver Fading Performance
`
`The test arrangement for assessing the direct conversion re-
`ceiver’s performance in a fading environment is shown in
`Figure 11. The test signal used was a sine wave, for sim-
`plicity. The SINAD results are shown in Table 3. Set 1
`was employed, with its preamplifier connected. This demon-
`strates the superior sensitivity of Set 1, and the superior
`strong-signal handling of the direct conversion receiver. Pre-
`liminary tests with speech indicate similar conclusions. It is
`expected that future data trials will demonstrate the advan-
`tage of the linear phase channel characteristic.
`
`Receiver
`
`Set 1
`
`Direct
`I1 conversion I1
`
`Signal Strength (dBm)
`-27
`-47
`-77
`-107
`-27
`-47
`-77
`
`I
`I
`
`SINAD measurements
`No fading 1 Moderate fading 1 Severe fading
`I
`I
`18
`19
`9
`12
`23
`2 1
`12
`23
`21
`16
`10
`18
`21
`10
`22
`10
`22
`21
`10
`22
`20
`
`I
`I
`
`I
`
`Table 3: Comparative SINAD measurements for different
`levels of fading
`
`61
`
`TCL EXHIBIT 1047
`Page 5 of 6
`
`

`
`[8] Y. Akaiwa , Y. Nagata A Linear Modulation Scheme for
`Spectrum Efficient Digital Mobile Telephone Systems :
`International Conference on Digital Land Mobile Radio
`Communications, Venice , July 1987.
`[9] A T 6 T Digital Cellular System Proposal : Submission
`to EIA Technical Subcommittee TR 45.3 June 1988.
`[lo] C. J. Collier , C. R. Poole Digital Correction of Chan-
`nel Mismatch for a Digitally Implemented Direct Con-
`version Radio : 4th IERE conf ‘Land Mobile Radio’
`Warwick 15 Dec 1987 p19
`[ l l ] R. Zavrel State-of-the-art IC’s simplify SSB Receiver
`Design : ‘Electronic Components and Applications’ Vol
`7 NO 4, pp223-228
`[12] V. Petrovic Reduction of Spurious Emission from Ra-
`dio l’kansmitters by Means of Modulation Feedback :
`IEE Conf on Radio Spectrum Conservation Techniques,
`September 1983, pp44-49.
`[13] V. Petrovic Application of Cartesian Feedback to HF
`SSB h n s m i t t e r s : IEE Conf on HF Comkunications
`Systems and Techniques, 1985, pp81-85.
`[14] D. K. Weaver A Third Method of Generution and De-
`tection of Single Sideband Signals : Proc IRE December
`
`1956, ~ ~ 1 7 0 3 - 1705.
`[15] A. Bateman, D. M. Haines , R. J. Wilkinson Linear
`Ransceiver Architectures : IEEE Conf VT-88, Philadel-
`phia June 1988.
`[16] I. White Modem VHF/UHF fiont-End Design : ‘Radio
`Communication’ April 1985, pp264-268.
`[17] F. G. Stremmler Introduction to Communication Sys-
`t e m s : Addison-Wesley, USA 1982 p511.
`J. N. Gannaway The Effects of Preamplifiers on Re-
`ceiver Performance and a Review of some Currently
`Available 144 MHz Preamplifiers : ‘Radio Communi-
`cation’ November 1981, pp1026-1031.
`H. Taub , D. L. Schilling Principles of Communications
`Systems : McGraw-Hill, USA 1987 p624.
`
`-
`
`Sins waw
`inwt
`
`TrIB
`-
`Modulator
`
`U(
`Tmnsrnitter
`
`I
`
`Figure 11: Test arrangement for fading comparison tests
`
`Conclusions
`
`The results presented in this paper show that a direct con-
`version transceiver system can be made to work at both VHF
`and UHF. The improved channel characteristics facilitated
`by digital channel filtering make this configuration partic-
`ularly suited to data transmission. In addition, the inte-
`gration potential of all the system components is the best
`available route to the concept of the ‘personal communica-
`tor’.
`Future work in this area will concentrate on techniques
`for automatic calibration and testing of the transmitter and
`receiver, the research and development of improved linear
`RF component technology, together with further evaluation
`of the prototype operation and reliability.
`
`References
`[l] P. Carpenter From Mobile to Personal Communications
`: Copenhagen Oct 1988 p1.7
`R. MacNamee , S. Vadgama , R. W. Gibson Universal
`Mobile Telecommunications - A Concept : 4th IERE
`conf ‘Land Mobile Radio’ Warwick 15 Dec 1987 p19
`R. Gibson, G. MacNamee , S. Vadgama Universal Mo-
`bile Telecommunications System A Concept : ‘Telecom-
`munications’ USA vol 21 no 11 p23-6 Nov 87
`
`Draft Report M/8 Future Public Land Mobile Telecom-
`munications Systems
`A. Bateman , J. P. McGeehan Phase-Locked Banspar-
`ent Tone-in-Band (TTIB): a New Spectrum Conjigu-
`ration Particularly Suited to the h n s m i s s i o n of Data
`over SSB Mobile Radio Networks : IEEE Trans. COM-
`32, 1984 ~ ~ 8 1 - 8 7 .
`J. P. McGeehan , A. Bateman Data Ransmission over
`UHF Fading Mobile Radio Channels : IEE Proceedings
`Pt F Vol 131, 1984 pp 364-374.
`H. Hammuda , J. P. McGeehan Spectral Efficiency
`of Cellular Land Mobile Radio Systems : IEEE Conf
`VT-88 Philadelphia June 1988 pp616-622:
`
`62
`
`TCL EXHIBIT 1047
`Page 6 of 6