1
ECE-542 (Spring 2004)
TERM PROJECT
DUE DATE: Thursday, May 6, 4PM
LATE SUBMISSIONS ARE NOT ACCEPTED
General remarks: You are asked to collaborate with each other in small groups. Each student
must submit his/her own report. The report must include an introduction, analysis, numerical
results, thorough discussion of the results, a conclusion, and references. Lengthy derivations
should be included in an appendix. I expect to see typed reports that are at a professional level of
quality. All figures, graphs, and tables must be adequately labeled and captioned. The report must
be connected and exhibit smooth logical flow. Start early!
Project description: This project involves the performance analysis of an on-off keying (binary)
direct-detection optical communication system. In such a system, a binary stream of data is used
to modulate a light source. The modulated light is transmitted through a series of fiber sections
that may involve the detection and regeneration of light prior to each new section. At the
destination, a synchronous receiver observes the signal for the duration of each bit and decides
whether a high (on-light) or low (off-light) signal was sent by the transmitter in the prescribed bit.
The performance of the system can be measured by the bit-error rate (BER), defined as the
average probability of a bit error. The BER is typically plotted as a function of the transmission
rate R (bits per second), the optical energy per bit, length of the fiber, etc. Another common
measure of performance is the receiver sensitivity S, defined as the minimum mean number of
photons per bit necessary to achieve a BER of 10-9
.
The performance of an optical communication system is governed by the performance of its
components, which consist of the transmitter, the link, and the receiver. Factors that affect the
performance of the transmitter include the responsivity of the light source, its bandwidth (time
response), and the transmitter-to-fiber light coupling efficiency. The performance of the link is
governed by the length of each section, the number of repeaters, attenuation loss, modal, material,
and polarization-mode dispersion, inter-link connector loss, etc. For lack of relevance to this
course, repeater loss and all forms of dispersion will be ignored in this project. The performance
of the optical receiver is governed by the detector gain noise (uncertainty), shot noise, circuit
noise, bandwidth of the receiver (which gives rise to intersymbol interference, ISI), and possibly
background radiation noise. We will assume that the receiver integrates the photocurrent over the
duration of a bit (i.e., it measures the accumulated charge per bit) and compares it to an optimal
threshold. This receiver is called an integrate-and-dump receiver. A common simplifying
assumption in the BER calculations is that the receiver output signal (prior to the decision stage)
is assumed to be Gaussian.
We begin by listing the available components:
a) InGaAsP LED with a 0.1mW maximum fiber-coupled power, and a bandwidth of 3 GHz.
b) Fiber sections, each 500 m long, with 1.5 dB/km attenuation at a wavelength of 1300 nm.
c) An avalanche photodiode (APD) with a mean gain in the range 5-50 and quantum efficiency of
0.4 at 1300 nm (see Ref. 1, Ch. 17 and become familiar with APDs and their gain statistics).
Assume that the random impulse response (response when a photon is absorbed) of the APD is
I(t) = Gq
τ
−1 exp{-t/
τ
}u(t), where
τ
is the response time of approximately 500 ps, q is the
