Performance of optical intersatellite links

INTERNATIONAL JOURNAL OF SATELLITE COMMUNICATIONS, VOL. 6, 153-162 (1988) PERFORMANCE OF OPTICAL INTERSATELLITE LINKS M. WIlTIG AND G . OPPENHAUSER European Space Agency, ESTEC, Keplerlaan 1, N L 2200 AG Noordwijk, The Netherlands SUMMARY The performance of optical intersatellite links is described by a model. A significant difference between optical and microwave intersatellite links is the occurrence of so-called burst errors. The statistics of bit error rate for an optical link are derived from simulations taking into account beam mispointing resulting from satellite motions and tracking channel noise. The effect of coding on the bit error rate is also shown. The mutual tracking system between two optical terminals located on different satellites can become unstable if the noise of the tracking loop exceeds a critical value. The results of tracking simulations are presented. KEY WORDS Intersatellite links ODtical communication Laser Bit error rate Burst error rate Tracking loop OOK PPM Simul’ation 1. INTRODUCTION 1.06 ~ r n The . ~ COz laser communications system was investigated in the U.S.A.’ and Europe.2 The Intersatellite links (ISLs) can be established between system studied in Europe has reached an advanced two geostationary spacecrafts (GEOs), between a state of development. Optical phase-locked loops4 low earth orbiting (LEO) spacecraft and a GEO, and optical Costas loops5 operating at 10 pm have or between a deep space probe and a GEO or been realized. The COz laser communications system LEO. The main traffic to be carried by a GEO-GEO is very similar to the well-known phase modulated link would be telecommunications (e.g. telephony, microwave systems, using phase modulation of the business, user data or TV). These intersatellite links optical carrier. The receiver consists of an optical would allow network interconnectivity or provide PLL for carrier recovery and demodulation. A new services in a more favourable manner. For a laboratory breadboard of the C 0 2 laser communi- LEO-GEO link, transmission of data collected at cations system is being completed. the LEO seems to be the most attractive application. The rapid pxogress made with terrestrial optical This data relay service would improve the collection fibre communications systems, and especially the of earth observation data and be of great importance improvements achieved in the development of for the future space station. reliable and powerful semiconductor laser diodes, These links can be established either at r.f. offers the possibiIity of using solid state technology frequencies or at optical frequencies. The main for optical intersatellite links. Frequencies corre- advantage of using optical frequencies is the high sponding to wavelengths of around 800 and 13001 antenna gain of optical ISL. Other advantages of 1500 nm are available. Optical communications with optical systems are the large available bandwidth semiconductor laser diodes was identified as a and the ease of suppression of external interference promising candidate6 for the European data relay due to the extremely narrow bandwidth. Addition- satellite (DRS). But system feasibility has still to be ally, optical frequencies are suitable for applications demonstrated, so significant development effort is outside the earth’s atmosphere which would have a necessary before the flight hardware is produced. strong absorption for nearly all optical frequencies. Within its Payload and Spacecraft Development and For data transmission between a deep space probe Experimentation program (PSDE) the European and an earth orbiting spacecraft, the data rate Space Agency (ESA), has started to develop an achievable would be higher, and the required deep optical payload based on today’s technology for a space (optical) antenna diameter lower, than for planned in-orbit demonstration starting in 1993.’ In today’s microwave systems if communication systems the U.S.A. the development of an optical terminal operating at optical frequencies were to be used. In for space-to-ground and space-to-space links is under this paper only near-earth ISLs are considered. way. This terminal will be flow on ACTS.8 The Optical communications systems for space appli- Japanese also are working on an optical terminal cations have been considered since the early 1970s. for space-to-ground and space-to-space links, to be At that time the two potential systems were based flown on ETCS VI.9 on COz lasers, operating at 10-6 pm,l. and In this paper the performance characteristics of flashlamp pumped Nd : YAG lasers operating at optical ISLs links are described, with particular 0737-2884/88/020153-10$05.00 Received January 1988 01988 by John Wiley & Sons, Ltd. 154 M. WITTIG AND G. OPPENHAUSER emphasis on how they differ from the well-known For the antenna mispointing angle, a Gaussian microwave links. mispointing error is usually assumed for each of the two orthogonal axes of the satellite. For this model one arrives at the equationlo 2. OPTICAL INTERSATELLITE LINK CHARACATERISTICS (3) 2.1. Link budget which relates the so-called burst error probability (PBE) with the antenna mispointing angle E. A For optical ISLs the same formalism for link burst error occurs if the instantaneous bit error rate budgets can be used as for microwave systems. (BER) drops below a defined value. Let us now However, the antenna gain of the optical telescope define this value as occurring when the antenna has a great influence on the overall system design. gain is 3 dB below the on-axis gain value. This This is caused by the fact that the optical (full) corresponds to du = 115 8Iu. With this relation we beamwidth 6 is expressed by obtain the PBE as a function of the ratio between 6 = AAID, 217~< A < 3 (1) optical beam divergence and the r.m.s. value of the statistical mispointing angle a; Figure 2 shows this and is several orders of magnitude lower than the relation. If we now specify a PBE and we have attitude stability of the host satellite. Here A is the an optical communications system operating at wavelength and D the telescope diameter, and the wavelength A, we obtain the relation between the factor A characterizes the dependence on the shape allowable antenna diameter and the r.m.s. beam of the radiating aperture. For a semiconductor laser mispointing as diode system with a wavelength of around 830 nm and a telescope diameter of 30 cm, 6 = 6.75 p a d . (4) The host satellite's attitude stability is assumed to be 0.1" = 1.75 mrad. The optical beam is thus about 260x260 narrower than the attitude stability. This leads to drastic reductions of the antenna gain unless special care is taken to reduce the uncertainty between the two communicating spacecrafts. This is the most serious technological problem of optical ISLS. A simple quantitative analysis gives more insight into the problem. Let us start with the antenna gain as a function of mispointing angle: where Jl(x) is the Bessel function of first kind. Figure 1 shows the influence of mispointing angle on antenna gain. It can be seen that for 61e = 115 Figure 2. Burst error probability PBE as a function of beam the antenna gain reduction corresponds to 3 dB. divergence 0 normalized to r.m.s. beam mispointing u This function is shown in Figure 3 for two different wavelengths. It is obvious that an optical communi- cations system with a long wavelength is less prone to antenna mispointing than a system with a shorter wavelength. Nevertheless, optical communications systems using solid state lasers at wavelengths near 1OOOnm (800, 1060, 1300 and 1500nm) are technologically superior to gas laser systems operating at longer wavelengths in their reliability, lifetime, mass and power consumption. If we now substitute equation (4)for the maximum allowable antenna diameter into the general equ- ation for the antenna gain Figure 1. Antenna gain as a function of mispointing angle E normalized to beam divergence 8  PERFORMANCE OF OPTICAL INSTERSATELLITE LINKS 155 need to achieve precise pointing of the beam remains a technical challenge. Next we consider the gain of the receiver antenna. Provided the receiver is located in the far field of the transmitter, a plane wavefront will arrive at the receiver telescope. The reduction of receiver antenna gain due to mispointing of the receiver telescope is then given simply by the projection of the entrance aperture onto the arriving wavefront: For mispointing angles of interest the cosine term is always close to unity, and equation (7) reduces to equation (5). With the transmitter antenna gain described by equation (2) and the receiver antenna gain by equation (7), we obtain the received power asll where T is the optical transmission. The important results of this consideration are that higher receiver antenna gains can be achieved with either larger antennas or shorter wavelengths, and that the transmitter antenna gain is constrained by the required burst error probability and the r.m.s. value of the antenna mispointing angle, and is independent of wavelength and antenna size. Figure 3. Allowable transmitter antenna diameter as a function For a unidirectional link a small transmitter of r.m.s. beam mispointing for a burst error probability of low6 and for two wavelengths: (a) 10.6 pm and (b) 0433 pm antenna and a large receiver antenna are the optimal choice, whereas for a bidirectional link a trade-off between pointing accuracy and link performance is we end up with the result required. 1*53 G T = (uV(-2lnPBE) P That is, the transmitter antenna gain as a function 2.2. Modulation-detection scheme The received power required for a given BER is a function of the modulationkoding scheme. With of beam mispointing and PBE is independent of the optical communications systems, either direct detec- wavelength (Figure 4). This is one reason why tion or heterodyne detection may be used. The semiconductor laser communications systems are of choice depends mainly on the laser technology. great interest for ISL applications. However, the For the longer wavelength region around 10 pm, heterodyne detection using an optical phase-locked loop (OPLL) was successfully demonstrated several years ago. With semiconductor lasers operating at 800 and 1300/1500nm, both direct and heterodyne detection were demonstrated. But the complexity of a semiconductor laser heterodyne detection system restricts its application to the laboratory at present. Direct detection systems are simpler than heterodyne systems, and are used in terrestrial fibre optical links. For ISL applications in the near future, it is assumed that direct detection will be used. The optical receiver front end of the direct .* C detector consists of an optical system which concen- 0 I3 80 I I trates the received power on a photodetector. Since the photocurrent is directly proportional to the incident optical power, optical direct detection systems require the transmitted information to modulate the intensity (power) of the transmitter 156 M. WITTIG AND G.OPPENHAUSER laser. The one-to-one coding of a serial non- cess.13 With very high degree of confidence, a return-to-zero (NRZ) data-stream into an intensity Gaussian distribution for the photocurrent generated modulated light sequence is called on-off keying by an APD optical front end can be assumed.14,l5 (OOK) modulation. Another very important modu- The signal and noise currents required for the lation scheme for optical systems is to decode a calculation of the BER as a function of received serial arriving bit stream of length n after buffering signal and background power are given by into its decimal equivalent. This gives one high state out of M=2” possible states, and only this one pulse is transmitted for n bits. This modulation scheme, r14 i,=U-PPR known as M-ary pulse position modulation ( M - hf PPM), has the advantage that the transmitter is operated over 1/M of the bit duration, with M times the C.W. power. This gives the receiver a better chance to distinguish a received pulse from the unavoidable background radiation. This is important in a LEO-GEO link for the high-data-rate receiver at the GEO, which has the earth as a permanent background noise source in its field of view. For OOK, with the usual setting of the threshold level midway between the signal level for a received The BER as a function of received optical signal high and a received low state, we obtain the BER power for the different modulation/detection as a function of the photodetector current is schemes and the parameters given in Table I are BER = Q (k) (9) plotted in Figure 5. The BER after forward error correction coding with a block code of total length 31 bits and a two-error correction capability is also where Q(x) is the well-known integral over the shown. 9 bits of redundance are required. For a Gaussian error function, and us is the noise current BER in front of the decoder of less than no for the laser switched on. improvement as a result of error correction can be However, the noise level for a received high and expected. a received low state after the photodetection is not the same. This is in contrast to r.f. and microwave systems, where the noise is independent of the signal 2.3. Performance of tracking loop level and stems from the thermal noise introduced As indicated in Section 2 . 1 , precise pointing by circuit elements. The detection of light by between both communicating satellites is essential avalanche photodetectors (APDs) is a non-deter- for an optical intersatellite communications link. ministic process; there exists only a statistical relation The tracking loop consists of a beam displacement between the incident light power and the generated sensor which corrects the detected displacement electrons. For an ideal photodetector and a constant with its pointing system. Usually the pointing system incident optical power, this relation is described by consists of coarse and fine pointing elements. The the Poisson distribution. It is well known that the variance of a Poisson distributed variable is given Table I. Basic parameters of the simulated link by the square root of the mean value, and this explains dependence of the noise level on the signal Transmitter power level ADPs. For OOK detection with an Wavelength 830 nm optimal setting of the threshold,12 we obtain the Transmitter power 30 mW following relation for the BER as a function of is: Transmitter antenna diameter 35 cm Pointing error mean value 500 nrad BER = r.m.s. value 300 nrad uns+us Channel where uns is the noise current if the laser is Distance 45 000 km Receiver switched off. Finally, the BER for M-PPM can be Received antenna diameter 20 cm approximated by Background power -70 dBm Data detector APD M-1 is-&, APD gain 80 BER 5 ~ Q (-l/($lo@V)) (11) 2 uns Quantum efficiency 0.8 Bitrate 60 Mbps The statistical distribution of photocurrent is deter- Modulation 4-PPM mined first by interaction between radiation and Tracking detector CCD matrix matter (in the ideal case it is given by the Poisson Quantum efficiency 0.8 distribution), and second by a complicated statistical Bandwidth 2 kHz Receiver field of view (total) 20 prad law which models the carrier multiplication pro-  PERFORMANCE OF OPTICAL INSTERSATELLITE LINKS 157 - - - nocoding (31,22,2) coding BER,PBE,R,B I I NEAs I 8s u NEA= f(G,,,.PBEI BL = f l N E A s , 6,. N E A ) I B, =f(BER,PBE,R,B) I Figure 6. Relation between required link performance, satellite environment. communications svstem narameters and trnrkino antenna gain by 3 dB from its on-axis value. An ~ 65 -60 - 55 important aspect of the optimal design of an optical Pa ld0n Figure 5. Bit error rate (BER) as a function of received optical communications system is indeed the influence of power for link parameters of Table I and different modulation/ beam mispointing, coming from uncorrected satellite detection schemes motions and/or tracking channel noise, on the link performance. coarse pointing elements have a large angular range A better answer to this problem can be obtained and low bandwidth, whereas the fine pointing only as a result of system simulation. It is possible elements have a small angular range, typically of to construct a ground-based optical communications the order of the satellite attitude uncertainty, and system which is representative of systems to be a large bandwidth. The total bandwidth of the flown on satellites. The influence of the tracking control loop is an important parameter of the optical channel noise on link performance is the same on communications system. This can be visualized with the ground and in orbit. The influence of satellite the help of Figure 6. If the designed system requires motion on link performance can be verified on the a received power P R to achieve a BER over the ground by using in-orbit measured satellite motion link distance R, we can choose optimal values for data to excite the optical communications system. the transmit and receive antenna diameters. If However, there is one problem which cannot be we have a r.m.s. mispointing angle (NEAs) and a resolved with a ground-based set-up. A transmitter noise equivalent angle (NEA) coming from the antenna far field is assumed for the operational noise of the tracking channel (the most important system. This far field is reached at a distance R noise source is indeed the tracking detector), we greater than end up with very stringent requirements for the bandwidth BL of the beam steering control loop. LINK SIMULATION To visualize this relation, consider that at this From the previous discussion we may conclude distance the diameter of the transmitted beam is, that the performance of an optical intersatellite as a result of beam divergence, just twice the communications system depends on various par- diameter DT of the. radiating aperture. Therefore ameters. The analytical model describing the relation the wavefront at this distance is not planar. The between beam mispointing and burst error rate PBE difference between the rim of the beam and the was derived under the assumption that the burst centre is just half the wavelength. If the receiver error is defined as the decrease of the transmitter aperture is a plane perpendicular to the beam 158 M. WI'lTG AND G. OPPENHAUSER Signal Generator Transmitter Antenna Channel Antenna Receiver IER Figure 7. Block diagram of communications systems direction and is not located on the axis, destructive resulting from the tracking channel noise is Gaus- interference will occur at the detector surface. sian,16 and satellite induced mispointing can be For a wavelength of 830nm and an antenna described in principle by a transfer function which diameter of 30 cm, the far-field condition is fulfilled is exited by Gaussian noise. at distances greater than 220 km. This clearly As an example a link from LEO to GEO is excludes a representative ground-based verification simulated. The link parameters are as given in Table of link performance. This is the main reason why a I. To generate the beam mispointing angles a simulation of an optical intersatellite communi- statistical generator was used to produce Gaussian cations system is necessary. values for a specified mean value and variance. A Figure 7 shows the block diagram of one communi- simulation run with 20 000 samples was performed. cation path. The signal source is a pseudo-noise- The instantaneous BER for the first 200 samples is generator which generates different patterns. These shown in Figure 8. It can be seen that the design data patterns are then transformed into an optical goal of a BER < lop6, i.e. a PBE < lod6, is signal. The transmitter is characterized by its transfer achieved. Figure 9 shows a histogram of the function, its transmitted bit energy, the transmitted instantaneous BER for the whole simulation. The wavelength and the antenna gain given by equation peak of probability is around a BER of 1O-l1, and (2), which takes beam mispointing from the nominal is vanishingly small for BER > lop6.The occurrence pointing direction into account. The space loss is a of a maximum of BER probability distribution is a function of wavelength and distance. The receiver consequence of the nonzero mean value of the beam antenna gain shows the mispointing dependence as mispointing error. For the same variance of beam given by equation (7). The receiver front end pointing errors, but with a zero mean value, we consists of a photodetector, a low-noise preamplifier obtain the BER probability distribution of Figure and a filter. The parameters describing the receiver 10, which is a monotonic function. front end are wavelength, quantum efficiency, dark current, internal gain, excess noise factor, load resistance, noise temperature and transfer function -I- of the following low-noise preamplifier. -2 - To evaluate BER performance, the eye-pattern -3 - obtained after the optical receiver front end (ORFE) -4 - is calculated. From the eye-opening, the signal-to- -5- noise ratio can be calculated, and from the signal- -6- to-noise ratio the BER for the different modulation/ -1 - detection formats can be determined. The antenna -8 mispointing angles for both the transmit and receive antennas are derived from two orthogonal mispoint- -p -g -10 ing angles. Either these mispointing angles can be v generated by dedicated statistical generators, or - -11 -12 measured values can be applied. For the following results, a Gaussian distribution of mispointing angles is assumed. This is a good approach because Figure 8. BER without coding as a function of time for the link measurements have shown that the mispointing parameters of Table I  PERFORMANCE OF OPTICAL INSTERSATELLITE LINKS 159 Figure 9. Probability histogram of BER for a mean pointing error of 500 nrad and a r.m.s. mispointing of 300 nrad b) coded BER 0 , 20 19- IS- 17- 16- 15- 14- 13- 12- 11- -7 Figure 10. Probability histogram of BER for a mean pointing error of zero and a r.m.s. mispointing of 300 nrad w 8 -9 B -10 -12 0 timi 4L I 1.5 6 .5 9 Figure 11. BER for a mean pointing error of zero and a r.m.s. value of 800 nrad; (a) without FEC and (b) with FEC using a (31, 22, 2) block code In many other simulation runs the variance of reality. To visualize the problem, consider Figure beam mispointing angle was observed. With such a 12. An optical transmitter located at satellite A data set the influence of forward error correction transmits a beam towards the optical terminal (FEC) coding was investigated. A simulation result located at satellite B. Let us assume that perfect with a r.m.s. mispointing of 800 nrad is shown in alignment is achieved. The ORFE of terminal B Figure ll(a). It was found that a short block code injects some noise into the tracking control loop of with high redundancy is the most efficient code for terminal B. This means that the signal received from this application. A reduction of instantaneous BER terminal A generates a beam mispointing at terminal can be achieved (Figure ll(b)). However, if the B. The transmitter antenna gain of terminal B instantaneous BER is below lo-* no improvement influences the signal detected by the ORFE of is possible. This means that the FEC alone is not terminal A. The tracking loop of terminal A again sufficient to compensate for higher beam mispoint- injects noise into the tracking control loop, which ing. Unfortunately the other extreme occurs if the affects the signal received at terminal B. It can be BER is below a code dependent threshold: the FEC expected that the noise occurring in the ORFE can then increases the effect of burst errors. lead to instabilities of the mutual tracking of both In all the above cases it was assumed that the terminals. To investigate this effect in more detail, receiver was always locked. In reality this is not so a further system model was established and some and the effect of coding is further reduced by loss simulation runs were performed. For the above- of clock and frame synchronization during the mentioned LEO-GEO link, the beam mispointing bursts. angle of the x-axis of terminal A is shown for a All the results presented so far have been derived tracking loop signal-to-noise ratio of 19 dB in Figure under the assumption that the signal-to-noise ratio 13. The noise equivalent angle of the tracking loop in the tracking channel is always high enough, and is below 300 nrad. But if we decrease the tracking consequently that large beam mispointing has no loop S / N to 16 dB, the system becomes unstable effect on the tracking loop. But that is not so in after some time. For this simulation an ideal beam 160 M. WI'ITIG AND G. OPPENHAUSER Terminal A 0 'c) . 3 - 0 D .- .2- $ .A ' b 3 6 '9 i2 is ie ii A4 h time/s Figure 13. (a) Noise equivalent angle (NEA) and (b) beam Figure 14. (a) Noise equivalent angle (NEA) and (b) beam mispointing (e) of terminal A for a tracking loop signal-to-noise mispointing (6) of terminal A for a tracking loop signal-to-noise ratio of 19 dB ratio of 16 dB mispointing control loop was assumed, and only the quality of the link (the BER) depends on precise influence of noise on the system performance was beam pointing. This beam pointing is influenced by evaluated. In reality some degradations must be the motion of the satellite and by the noise of the expected due to the nonlinear tracking detector tracking system itself. Unfortunately real in-orbit characteristic and the remaining tracking loop point- measured satellite motion data with microradian ing error. But the results obtained give an indication accuracies exist only up to frequencies of some 10 of the required tracking loop S/N. Hz . A full-scale ground-based test and performance CONCLUSIONS evaluation of an optical intersatellite communi- The difference between optical ISLs and microwave cations system is not possible. The main reason is systems in terms of link performance has been that the far-field conditions of the optical antenna shown. However, the great advantage of very narrow require there to be a distance of over 200 km beams (high antenna gain, reduction of interference) between the terminals. As a consequence the brings with it a considerable disadvantage: the performance of the optical intersatellite communi-  PERFORMANCE OF OPTICAL INSTERSATELLITE LINKS 161 cations system can be evaluated only outside the on-ff keying earth’s atmosphere. optical receiver front end In order to improve the level of confidence for a optical transmitter successful in-orbit operation, a Simulation of the probability distribution of x system performance is required. The models used background radiation power for this have to be as realistic as possible. Special received power experiments using stratospheric ballons as carriers transmitter power in order to reduce the influence of the earth’s probability of burst error atmosphere1’ are one possibility. electron charge The simulation of the communications subsystem, integral over Gaussian error function taking into account beam mispointing for a perfect link distance tracking system, resulted in the BER varying as a far-field distance function of time. The histogram of the BER verifies detector load resistance the design goal, that the link budget yields a BER signal function not greater than lop6. signal-to-noise ratio A second simulation was carried out for the signal bit duration tracking performance. A critical tracking loop signal- LNA noise temperature to-noise ratio can be determined. Below this value receiver loss mutual tracking between both terminals is imposs- transmitter loss ible, and the tracking loops become unstable. The internal gain of APD critical tracking loop S/N determines the optimal mispointing angle partition of received optical power into the tracking quantum efficiency and data channels. beam divergence Optical intersatellite communications systems wavelength have some advantages over microwave systems, but r.m.s. value of mispointing angle some new problems arise. The design of an optimal noise current for laser on system is more complex than for a microwave noise current for laser off system. To achieve the optimal solution, extensive 70 transmitter pulse width simulations are a good and helpful tool while refining 72 received pulse width the model step by step. ABBREVIATIONS AND SYMBOLS REFERENCES aperture shape factor 1. J. H. McElroy ef al., T O 2 laser communication systems for near-earth space applications’, Proc. IEEE, 65, 221-251 transmission bandwidth ( 1977). control loop bandwidth 2. W. Englisch, ‘Study on intersatellite laser communication bandwidth of satellite vibrations links’, ESA Contract No. 3555/78, Final Report, 1979. 3. M. Ross ef al., ‘Space optical communications with the bit error rate Nd : YAG laser’, Proc. IEEE, 66, 319-344 (1978). control loop equipment 4. E. Bonek et al., ‘Optical PLLs sees the light’, Microwaves antenna diameter & RF, 22, 65-70 (1999). 5. H. K. Philipp ef a/.. ‘Costas loop experiments for a 10.6 p n laser extinction ratio communications receiver’, IEEE Trans. Commun., COM-31, transmitter pulse energy 1000-1002 (1999). optical carrier frequency 6. ‘Assessment of optical communication systems for data relay satellite’, ESA Contract, Final Report, 1987. excess noise factor of APD 7. J . L. Vanhove and C. Noeldicke, ‘In-orbit demonstrations antenna gain of optical IOLiISL - the Silex project’, this issue. Planck’s constant 8. R. R. Lovell and C. L. Cuccia, ‘Switches, lasers and electronically hopped beam antennas in space - NASA’s transmitter filter transfer function Advanced Communications Technology Satellite’, Paper receiver filter transfer function presented at ‘Space Tech 8 6 Int. Conf., Geneva, May 1986. signal current for laser on 9. Y. Furuhama et al., ‘Present status of optical ISL studies in Japan’, SPIE Proc., 810, 147-149 (1987). signal current for laser off 10. P. W. Scott and P. W. Young, ‘Impact of temporal detector dark current fluctuations of signal-to-noise ratio (burst error) on free- Bessel function of the first kind space laser communications system design’, SPIE Proc., 616, 174-181 (1986). Boltzmann’s constant 11. J. L. Vanhove, B. Laurent and J . L. Perbos, ‘System analysis channel loss of optical interorbit communictions’, SPIE Proc., 810, 164-171 (1987). number of time-slots 12. J. B. Abshire. ‘Performance of OOK and low-order PPM M-ary pulse position modulation modulations in optical communications when using APD- additive noise based receivers’, IEEE Trans. Commun., COM-32, 114G1143 (1984). noise equivalent angle 13. R. McIntyre, ‘The distribution of gains in uniformly multiply- noise equivalent angle of satellite ing avalanche photodiodes: Experimental’, IEEE Trans. vibrations Electron Devices, ED-19, 713-718 (1972). 162 M. WITI’IG AND G . OPPENHAUSER 14. R. M. Gagliardi, ‘On Gaussian error probabilities in optical 16. T. M. Duncan and T. H. Ebben, ‘Measurement of pointing receivers’, ZEEE Trans. Commun., COM-28, 1742-1747 error distributions in tracking loops of optical intersatellite (1980). links’, SPIE Proc., 756, 54-61 (1987). 15. N. Sorensen and R. Gagliardi, ‘Performance of optical 17. U. Johann and W. Reiland, ‘Stratospheric balloons as a receivers with avalanche photodetection’, ZEEE Trans. testbed for optical space communications technology’, Znt. Commun., COM-27, 1315-1321 (1979). J . Satell. Commun., 6, 225-230 (1988).