B. FSO Link
In the optical link experiment, the IC laser was modulated with a pseudorandom bit sequence (PRBS) signal from a bit-error-rate transmitter (BERT) and a bias tee was used to combine the high-speed modulation and laser dc bias current. The laser operated at
$T=77\ \hbox{K}$ under applied dc bias of about
$I=100\ {\hbox {mA}}$ with output power of
$\sim$10 mW at 3.0
$\mu{\hbox {m}}$ [5] and the laser emission was collimated by a ZnSe lens. After a free space transmission of about 1 m, the optical signal was collected by a second lens and focused on a high-speed mercury–cadmium–telluride (MCT) detector manufactured by Vigo. Compared to QWIP [10] used in the initial high-speed laser characterization experiments, Vigo MCT detectors have much better sensitivity and higher operational temperature reachable with thermoelectric coolers that make them more practical for use in FSO experiments, yet their useful bandwidth was limited to about 70 MHz. After the detection, the high-frequency photocurrent signal was fed into a high-speed amplifier followed by a second BER receiver or high-speed oscilloscope after the bias tee.
Fig. 1 shows the results of an indoor FSO transmission experiment for data transmission rates from 10 to 70 Mb/s. The open eye diagrams acquired for different data transmission rates are clearly seen in this graph. At data rates of 30 and 50 Mb/s, the observed large size of eye opening, small distortion, and time variation of zero crossing demonstrate the high performance of the FSO link based on the developed IC lasers. The degradation of the open eye diagrams at data rates below 10 Mb/ was attributed to the laser heating/cooling during long on/off times and the upper limit of data transmission rate of about 70 Mb/s was set by the Vigo MCT detector bandwidth. These limitations to achievable data transmission rates can be overcome by using IC lasers with better thermal properties such as narrow waveguide ridge devices and by utilizing detectors with higher bandwidth.
The performance of FSO links for different attenuations of the optical signal was also studied and BERs were measured using the BER receiver. Fig. 2 shows the open eye diagrams acquired for data transmission at a rate of 30 Mb/s for different attenuations of FSO link and the measured BER versus optical attenuation, showing exponential dependence at low BER. This graph demonstrates that an FSO link utilizing developed IC lasers can operate with BER less than
$10^{-8}$ that was the sampling time limit in this experiment, with attenuation up to 14 dB. The link BER, in fact, was limited by the receiver sensitivity where, for example, 18-dB attenuation corresponded to a received power of about 50
$\mu\hbox{W}$. With the availability of more sensitive and high-bandwidth detectors [11], [12], our estimate suggests that the transmitters could support much higher photon efficiency optical links with attenuation exceeding 60 dB.
C. FSO Link at 1.5
$\mu{\hbox {m}}$ versus 3.0
$\mu{\hbox {m}}$
The indoor experimental test-bed was used to study the effects of fog on the performance of the demonstrated mid-IR FSO link and for comparison between mid-IR and 1.5-
$\mu{\hbox {m}}$ FSO links. The mid-IR FSO link utilized the developed IC laser, and a commercial telecom laser and detector were used for the 1.5-
$\mu{\hbox {m}}$ link. In this test, the collimated optical beams from both lasers were combined by a beam splitter, and then both signals passed through the fog so the mid-IR and near-IR FSO link shared the same optical path through the fog. After completing the assembly, both mid-IR and near-IR FSO links were tested and BERs versus attenuation for both links were measured (not shown).
Next, properties of the FSO link were evaluated in clear air and in the presence of “dry” and “wet” fogs, where the former was created by the flow of liquid nitrogen
$(\hbox{LN}_{2})$ in the air and the latter by a commercial fog generator. For the three different “wet” fogs generated (Fogs 1, 2, and 3), the fog density and the water particle size was minimal in “Fog 1,” and both density and particle size were increased in the “Fog 2” and “Fog 3” conditions. Based on the fog generator specifications, the average size of the water particles was estimated to increase from 7 to 15
$\mu{\hbox {m}}$ between Fogs 1 and 3 but that was not verified experimentally.
The influences of the different fog conditions on the FSO link were investigated by measuring the temporal fluctuations of the received optical power for both mid-IR and near-IR channels at the different fog conditions, as shown in Fig. 3. Initially, the power fluctuations were measured in the absence of any air flow (denoted as “Clear” in the figures), then in the presence of the fog created by
$\hbox{LN}_{2}$ (dry fog) and in the wet fog (Fogs 1, 2, and 3). Fig. 4 shows the attenuation of the average optical power (relative to the optical power measured in clear air) over the 1-m link for both channels at different fog conditions. This graph shows that optical power of near-IR link decreased faster than the power of the mid-IR signal for all fog conditions, and at the “Fog 3” conditions for example, the attenuation of 1.5-
$\mu{\hbox {m}}$ channel is 2 times stronger than that of the 3-
$\mu{\hbox {m}}$ channel for 1-m-long FSO link.
Fig. 5 shows the probability distribution function of the optical signals with and without fog for 1.5- and 3-
$\mu{\hbox {m}}$ channels. The optical power of the near-IR channel showed a normal distribution with no flow, but in the dry fog it exhibited a strongly log normal distribution typical of a turbulent atmosphere. The scintillation index,
$\sigma_{I}^{2}$ calculated from
$\sigma_{I}^{2}=\langle{I}^{2}\rangle/\langle{I}\rangle^{2}-1$, where
$I(t)$ is the measured signal strength, is large
$(\sigma_{I}^{2}=0.35)$ for the near-IR channel in the dry fog showing that near-IR channel is strongly affected by scintillation. In contrast, the mid-IR 3-
$\mu{\hbox {m}}$ data showed only normally distributed power in both scenarios (without and with fog from the
$\hbox{LN}_{2}$ flow), demonstrating the advantage of using 3-
$\mu{\hbox {m}}$ wavelength link for transmission under strongly scintillating effects.
In the wet fog, the scintillation indexes were found to be small indicating very weak turbulence. Also, the measured ratios of the scintillation index did not follow that predicted by theory [13] when comparing the two wavelengths. These results indicated that in wet fog conditions the scintillation was not dominated by the wavelength dependence but by the detection and/or transmission power variations. This was also confirmed by looking at the probability distribution functions that show a normal behavior for both no flow and under fog conditions for both wavelengths (not shown).