вход по аккаунту



код для вставкиСкачать
(vi) Each branch starts at state S,(a,, a 2 , . ... a,) at depth p
and is connected with state S , , ,(al. a 2 , .. ., a,) at depth p + 1,
which can be defined as follows:
S , + , ( a , ,a 2 , ..., a,) = Sp(ai,a z r ..., a,)
+ C;
There are
p= I
distinct paths through this trellis diagram and each path corresponds to a unique GAC codeword.
Example: Let the aim be to design the trellis diagram of the
binary (16, 11,4) GAC, described above.
Conclusion: A new encoding technique, called the generalised
array code (GAC) is proposed. The technique allows the
design of array codes with higher information rate, while the
size of the array and the minimum Hamming distance remain
the same. A trellis decoding technique for such codes is proposed with lower decoding complexity compared with conventional techniques. A technique has been presented to
design a trellis structure of (16, 11,4) GAC.
Acknowledgments: This research has been supported by the
UK Science and Engineering Research Council (SERC). We
are grateful also to V. Zyablov from the Russian Academy of
Sciences for helpful discussions and advices towards this
15th February 1993
B. Honary (Hull-Lancaster Communication Research Group, Engineering Department, Lancaster University, Lancaster LA1 4 Y R , United
G. S . Markarian (Armenian Academy of Sciences, at present at the
Engineering Department, Lancaster University, Lancaster LA1 4 Y R ,
United Kingdom)
P. G. Farrell (Electrical Engineering Department, The University of
Manchester, Manchester M I 3 9PL, United Kingdom)
(i) The code has the following generator matrices:
G , = [l
(ii) Following the procedure outlined above, we determine the
trellis depth N , = n2 + 1 = 5 and number of states N , = z4 =
(iii) We identify each state of depth p ( p = 1, 2, . . , , n,
4-tuple binary vector S,(a,, a 2 , a3,a4).
I ) by
(iv) At depth p = 0 and p = n2, the trellis has only one state,
namely So(oooO) and S,(OOOO), respectively.
(v) The trellis branches are labelled with X p / C b and are
obtained from the generator matrices of GAC (8). At depth
p = 1, the branches are as follows:
ELIAS, P.:‘Error free coding’, IEEE Trans., 1954, IT-4, pp. 29-37
FARRELL, P G.: ‘Array codes’, in LONGO, G. (Ed.): ‘Algebraic coding
theory and applications’(Spnnger-Verlag.1979),pp. 231-242
s., and NAMEKAWA, T.:
‘New classes of binary codes constructed on the basis of concatenated codes and product codes’, IEEE Trans., July 1976, IT-22,
(4),pp. 462-468
BLOCH, A. L., and ZYABLOV, v. v.: ‘Coding of generalized concatenated codes’, Problems of Information Transmission, 1974, 10, (3),
pp. 45-50
J. K.,and SLOANE. N. 1.: ‘The theory of error correcting codes’(North Holland, New York, 1977)
G . s., and DARNELL, M.: ‘Trellis decoding
technique for array codes’. Eurocode ‘92, Italy, Udine, 1992
WOLF, J. K.: ‘Elfcient maximum likelihood decoding of linear
block codes using a trellis’, IEEE Trans., 1982, IT-28, (2)
G. o.,JUN.: ‘Coset codes-Part
2: Binary lattices and
related codes’, IEEE Trans., 1988, IT-34, (3,pp. 1152-1187
= (oo00).
G l = (oo00)
(OOO1).G l
= (OOO1)
ci=15= (1110). GI =
C;=l6 =(1111). GI = (1111)
Following the above procedure, the trellis diagram for (16,
11, 4) GAC is presented in Fig. 1. It is clear from Fig. 1, that
the designed GAC has d,, = 4. The trellis structure of this
code has N o = 211 paths and is similar to that given by
Forney [SI for the Reed-Muller codes, however the encoding
procedure is easier.
S. S. Ou, J. J. Yang, C. Hess a n d M. Jansen
Indexing terms: Semiconductor lasers, Integrated optoelectronics
short-cavitv. strained-laver InGaAs/
” _
GaAlAs laser diodes that are suitable for two-dimensional
electronic-photonic integrated circuit (EPIC) applications
have been demonstrated. Continuous-wave threshold currents as low as 4mA, and singlemode output powers in
excess of IOOmW (maximum output power: 12OmW) were
achieved for as-cleaved, reactive-ion-etched, 3.5 urn-wide,
200 um-lone. InGaAdGaAIAs ridee-waveguide laser diodes
in the junclon-side up configuration.
Fig. 1 Trellis diagram of (16, 11, 4 ) G A C
Laser diodes with low power dissipation, good beam quality
and high power levels are desirable for two-dimensional
monolithic electronic-photonic integrated circuit (EPIC)
applications. Conceptually, these requirements are not significantly different from those of discrete devices; however, practically, they are somewhat difficult to meet. This is because,
first, there exists a tradeoff between a low threshold current
and a high output power. Secondly, because laser diodes
78th March 7993
Vol. 29
occupy more space than most other electronic components in
EPICs, in order to obtain a high fill factor the dimensions of
the laser diodes should be minimised. Thirdly, for wafer integration, the fabrication yield should be high. Fourth, to simplify process integration and packaging, laser diodes must
often be mounted junction-side up, which results in thermal
buildup. This in turn impacts device performance and reliability.
Generally, to achieve stable high-power fundamental-mode
operation, both the ridge-waveguide geometry and layer compositional are critical and require two dimensional modal
analysis [I]. In practice, the ridge width and the wing thickness (the distance between the edge of the GRINSCH region
and the bottom of the ridge) have to be precisely controlled.
The tolerance in wing thickness can be as low as 50nm. For
low-threshold current applications, the ridge is generally
narrow and the etching can be stopped at the interface
between the doped p-type cladding layer and the undoped
waveguide layer [2, 31. For high-power operation, large spot
sizes are required in order to lower the optical power density
on the facet. O n the other hand, as the modal overlap with the
laser active region decreases, the threshold current increases.
Therefore, it is difficult to achieve both high-power operation
and an ultralow threshold current from a specified ridge structure for 2-D EPIC applications.
Dry etching has been the favourite technique in the fabrication of narrow stripe ridge waveguide laser diodes for EPICs.
It offers several distinct advantages over conventional wet
chemical etching such as high fidelity pattern transfer, precise
process control, high fabrication yield, flexible layout/circuit
designs, and simple process integration. GaAs/GaAIAs and
InCaAs/GaAs ridge waveguide laser diodes fabricated by
using reactive-ion-beam etching and in-situ laser-monitored
RIE with Cl, gas have shown threshold currents of 5 mA and
3.6 mA from as-cleaved devices, respectively 13, 41. However,
high output powers have not been reported from these
devices. Recently, we have demonstrated singlemode output
powers in excess of 200mW (maximum power of 320mW) for
3.5pm-wide, loo0 pm-long InGaAs/GaAlAs dry etched ridge
waveguide lasers in the junction-side down configuration [ 5 ] .
These devices are suitable for high-power discrete device
In this Letter we report CW threshold currents of 4mA, and
singlemode output powers in excess of l00mW for 3.5pmwide, 200pm-long unmounted, as-cleaved, InGaAs/GaAlAs
devices tested junction-side up. The CW threshold current was
reduced to 2.3mA by coating the laser facets with highreflectivity (HR) coatings. For devices with CW threshold currents of -4mA and singlemode output powers of 100mW,
the fabrication yield is 90%. This can be attributed to the
uniformity of the dry etching process. These results pave the
way for future high performance EPICs.
The device structure is a graded index separate confinement
heterostructure with AI,.,Ga, ,As cladding layers surrounding an undoped active region consisting of a 9nm thick
In,.,Ga,.,As quantum well sandwiched between 20nm thick
GaAs layers. The ridge of the device was defined by reactive
ion etching two 3pm-wide groves. The wing thickness was
0.15pm. Reactive ion etching was carried out by using
99.99% pure SiCI, plasma at a flow rate of 7 . 5 sccm, IOmtorr,
50W at 13.56MHz and 90V selfgenerating bias. The background pressure was 1 2 x IO-' torr. A single photoresist
layer was used as the dry etching mask. The etching rate is
l w A / m i n . Both atomic force microscopy and 3-D optical
interference microscopy measurements of RIE etched sidewalls indicated a surface roughness of less than 8 n m over a
10pm measurement region. The pure SiCI, assisted RIE fabrication method has been published previously 161.
A plasma enhanced chemical vapour deposited Si,N,
dielectric layer was used to confine the current laterally. After
ohmic contact deposition, the wafer was cleaved into 200pm
short cavity bars. For low-threshold device applications, the
back and front facets of the devices were coated with SiO,/Si
alternating layers resulting in 95 and 70% reflectivities. Fig. 1
shows the cross-section of the device.
Typical room temperature CW power against injection
current (P-I) characteristics of as-cleaved and HR-HR coated
devices with 200pm-long cavity lengths are shown in Fig. 20
and h, respectively The characterisation was carried out by
probing individual devices on the bar in the junction-side up
configuration and without mounting or active cooling The
P -electrode
SaOCAl A i
vndoped Stro rec
Fig. 1 Schematic diagram ofdeuice structure
c u r r e n t , rnA
Fig. 2 Output power against current characteristics of as-cleaved, and
95-70% coated 2Wpm-long devices
(i) As-cleaved. I,, = 4mA, q = 0.8 W/A
(ii) 95-75% coated: I , , = 2-3mA, '1 = 0.45 W/A
CW threshold current is 4mA for as-cleaved devices, and is
reduced to 3.5 mA for HR-cleaved and to 2.3 mA for HR-HR
coated devices. The total differential quantum eficiency is
0.8 W/A for as-cleaved devices and is reduced to 0.45 W/A for
HR-HR coated devices. Typical CW singlemode output
powers of 100mW, which correspond to 30 x threshold, and
maximum output powers of 120mW, which were thermally
limited, have been observed from as-cleaved devices. The
power was reduced to 32mW before C O D occurred for
HR-HR devices. This is attributed to overheating at the
facets. The CW wallplug eficiency is 43% at l00mW of
output power. Fig. 3 shows typical lateral far-field (parallel to
the junction) patterns under CW operation at output powers
of 20, 50, and 100mW. The full width at half maximum
(FWHM) intensity is 18" at 100mW. The FWHM in the
direction perpendicular to the junction plane is 37.5", which
yields an aspect ratio of 2.1. The slight expansion of FWHM
with increasing output power is most probably due to self-
ELECTRONICS LETTERS 78th March 1993 Vol. 29 No.6
angle, deg
Fig. 3 Typical lateral far-field (parallel to junction) patterns under CW
operation at output powers of 20, 50, and 1WmW
focusing as a result of gain spatial hole burning. Beyond
100mW, further broadening of the far-field pattern with some
steering at the beam centre occur indicating multispatial-mode
operation. This is most probably caused by thermal waveguiding, because the heating of the device is considerable in the
junction-up configuration. Fig. 4 shows the emission spectrum
for the device at powers of 5, 20, 50, and 100mW. Below
D. J. Roscoe, A. Ittipiboon, L. Shafai and M. Cuhaci
Indexing terms ’ Noise, Active antennas
A noise model for a developed active antenna is presented.
The active antenna is a series of distribution of active devices
and microstrip transmission lines. This poses an interesting
problem in terms of noise analysis, because the active devices
and antenna can no longer be separated. The developed
noise model is verified with experimental measurements.
Introduction: An active integrated antenna is defined here as a
Fig. 4 Emission spectrum for device at powers of 5, 20, 50, and 100mW
5 mW, the devices show multilongitudinal mode operation.
The laser diodes operate in a single longitudinal mode at
wavelengths in the vicinity of 970nm. At 100mW, the lasers
exhibit single longitudinal emission at 971 nm with 17 dB sidemode suppression. This can be partly attributed to the short
resonant cavity. The short cavity stabilises longitudinal modes
at high injection currents [7]. The characteristic temperature
and the thermally-induced wavelength tuning coeficient are
measured to be 150K and 2 . 2 A / ” C , respectively. To the best
of our knowledge, 120mW (100mW) is the highest output
power (singlemode output power) from short-cavity ‘bare’
devices in the junction-side up configuration. The threshold
currents reported here are also among the lowest obtained for
InGaAs/GaAs ridge waveguide lasers and are consistent with
previous work [3].
In conclusion, we believe that our result is the first demonstration of a ridge waveguide structure which achieves both
low threshold current and high output power by opthising
the fabrication conditions required by EPlC applications. We
have obtained the highest output powers from short cavity
‘bare’ InGaAs/GaAs ridge waveguide lasers in the junctionside up configuration.
A c k n o w l e d g m e n t : The authors express their thanks to D.
Botez for useful discussions.
1st February 1993
S. S. Ou, J. J. Yang, C. Hess and M. Jansen ( T R W , Research Center,
Space and Technology Group, One Space Park, Redondo Beach, C A
90278, U S A )
OU, S. S.,
c., JUN., and PANISH,
U. 8.: ‘Heterostructurelasers, Part
A: Fundamental principles’ (Academic, New York, 1978)
2 WADA, O., SANADA, T., KUNO, M., and Rill, T . : ‘Very low threshold
current ridge-waveguide AIGaAs/GaAs single-quantum-well
lasers’, Electron. Lett., 1985, 21, pp. 1025-1026
3 CHAO, C. P., HU, S. Y., FWYD, P., LAW, K . - K , CORZINE, S . W., MERZ, J.
A. c., and COLDREN,
L. A.: ’Fabrication of lowthreshold InGaAs/GaAs ridge waveguide lasers by using in-situ
monitored reactive ion etching’, IEEE Photonics Technol. Lett..
1991,3, pp. 585-587
WEBB,0.:‘Ridge formation for GaAlAs GRINSCH lasers by CI,
reactive ion etching’, IEEE Photonics Technol. Letr., 1990, 2 , pp.
YANG, 1. I., JANSEN, M., HESS, C., SERGANT, M., TlJ, C.,
ALVAREZ, F., and LEMBO, L. 1.: ‘High-power InGaAs/GaAs singlemode laser diodes with reactive-ion-etched ridges’, Electron. Lett.,
1992,28, pp. 2345-2346
6 ow, s. s., YANG, J. I., and JANSEN,
M.: ‘5W GaAsJGaAIAs laser
diodes with a reactive ion etched facet’, Appl. Phys. Lett., 1990, 57,
pp. 1861-1863
A. G., and
MARCUSE, D.:’Short-cavity InGaAsP injection lasers-dependenceof
mode spectra and single-longitudinal-mode power on cavity
length’, IEEE J . Quantum Electron., 1982, QE-18, pp. 1101-1105
radiating element on which the current distribution along the
structure may be altered in a controlled manner. This involves
the concept of integrating active components into the antenna
itself. An active antenna provides a control mechanism to vary
charateristics such as peak gain, pattern shape, axial ratio and
its physical size. This type of active antenna has been developed and presented in References 1 and 2 (see Fig. 1). The
active devlces nr,n+pdm
Fig. 1 Active integrated antenna
a General configuration
h Fabricated four-arm active antenna
developed antenna is a travelling wave structure into which
amplifiers have been integrated. In general, the resultant structure is a distribution of microstrip transmission lines and
amplifiers. To use this antenna in a receive application, its
noise temperature must be determined. This Letter presents a
theoretical and experimental analysis of the noise temperature
of such an active antenna.
T h e o r e t i c a l analysis: Within a typical communication system,
the amplification stage follows the passive radiator. The noise
contributions from each stage can be clearly segregated and
the noise calculation of the system can be determined based
on a cascade network of noise sources 131. The developed
active antenna introduces an interesting problem when calculating the system temperature due to the integrated devices. In
this particular situation, amplifiers are located within the
antenna and in series with radiating sections of the antenna
and hence function as an integral part of the antenna. Thus
there are several noise sources, series distributed, within one
antenna element. These noise sources are affected by the gain/
loss of each active device and combine in some manner t o
produce an output noise temperature.
An equivalent noise model of the general structure, Fig. la,
is presented in Fig. 2 0 . The noise model is a cascade network
representation of the antenna, where the antenna is replaced
with a combination of resistors, attenuators, and active
devices. The noise model is applied to the fabricated prototype which consists of four radiating arms and three amplifiers. The equivalent noise model of this antenna is presented
in Fig. 2 h . The matched termination is represented by the
resistor R,, and the radiation resistance of the ith segment is
represented by R i . Each resistor is referenced to its environment temperature, with R , referenced to the sky temperature
as seen by the ith segment. The ohmic attenuation due to the
microstrip transmission line is represented by L,, referenced to
its environment temperature TL,.The gain of the ith amplification stage is represented by G , , where the noise figure of the
amplifier is expressed in terms of temperature by TA,.
18th March 1993 Vol. 29 No. 6
Без категории
Размер файла
316 Кб
Пожаловаться на содержимое документа